Objective Function Design For Video Stream call management in ...
Post on 31-Oct-2014
737 Views
Preview:
DESCRIPTION
Transcript
Objective Function Design For
Video Stream call management
in 3G Mobile Wireless Networks
Project report submitted to the department of Electrical and Computer Engineering
as partial requirement for M. Eng (Master of Engineering) degree.
Submitted By:
Raza Ahmad Bhatti On April 23, 2003
Supervisors Signature Page
UNIVERSITY OF ALBERTA
DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING
THE UNDERSIGNED CERTIFY THAT THEY HAVE READ, AND ACCEPT THE
DOCUMENT ENTITLED "Objective Function Design for Video Stream Call
Management in 3G Mobile Wireless Networks ",
SUBMITTED BY: RAZA AHMAD BHATTI
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING.
____________________________ (Dr. Mrinal Mandal)
Department of Electrical and Computer Engineering, University of Alberta.
_____________________________ (Dr. Ehab Elmallah)
Department of Computer Science, University of Alberta.
_____________________________ ( ) Department of Electrical and Computer Engineering, University of Alberta. Date:
Dedicated to My father Azmat Ullah Bhatti, mother Shamim, wife Zarqa, children Asad, Daniyal and my Brothers & Sisters
for their continuous love, prays and support to encourage me to fulfill my dreams to learn and be creative
Also Dedicated to Personalities who have worked and those who are
working towards making our globe a better place
to live
Abstract
The commercialization of mobile wireless networks has increased significantly
in recent times. It is primarily due to the ease of access to critical users services in
voice and data, and this includes Internet browsing, email, online banking, video
streaming and e-fax. Due to the increasing demand of mobile phones, it is becoming
critical to keep the users satisfied with the QoS of a mobile network at an affordable
cost. Hence an efficient and profitable Scheduling and Call Admission Control
(CAC) scheme is crucial for a mobile wireless network.
In this research we have considered developing an Objective Function that will
output the effectiveness of a Cost Function in comparison to a given QoS for mobile
users. Hence at any given time instance it can easily be deduced that if certain QoS,
Scheduling and Call Admission Control (CAC) schemes are profitable for the service
provider. This information will be of great benefit for the service provider in order not
to keep the users happy up to a level where it starts costing service provider. At that
point the service provider has to take decision on which user(s) to pick for either
degrading the QoS or disconnecting the service based on service contract. The users
willing to pay more for guaranteed QoS will remain at the same QoS. On the other
hand, the CAC will induce new users wanting guaranteed QoS at the cost of
degrading or disconnecting service of users according to the service contract. Which
basically indicates that such users can tolerate degraded service or disconnections by
having the advantage of paying less in comparison to the users having guaranteed
QoS.
Considering the design of such an objective function will be an enormous task,
if considering real world 3G wireless cells. That is multiple services for multiple
cells. Thus we have divided the problem into several subtasks, and we only
considered providing video streaming service in a single 3G W-CDMA wireless cell.
i
Table of Contents
Abstract ……………………………..…………………………………… i
Table of Contents ………………..……………………………….…… ii
List of Abbreviations ……..………………………..…………..…… v
List of Figures ………………..………………………………..……….… vii
List of Tables ………………..…………………………..………….…… viii
List of Symbols ………………..…………………………..……….……… ix
Acknowledgements ………...…………………………….……………… xi
1. Introduction ………………………………………………..….… 1
2. Introduction to Mobile Wireless Networks ……….… 4
2.1. Wireless Telephones, Computers and Networks ………………... 4
2.2. Wireless Networks ……………………………………...………... 6
2.3. 3G Mobile Wireless Networks …………………………………... 7
2.4. Overview on UTRA ….……………………………………….…. 9
2.5. Importance of Power Control in UTRA ……………………… 12
2.6. Summary ……...………………………………………..……... 14
3. Video Transmission Over Wireless Networks ….…….… 15
3.1. Video Service in Mobile Wireless Networks ………...……… 15
3.2. Digital Video Coding Standards ………………………………….... 17
3.3. Video Streaming Service in Mobile Wireless Networks …..………. 20
3.4. Packet Scheduling Methods for Mobile Wireless Networks ….… 21
3.5. Call Admission Control Methods in Mobile Wireless Networks ….. 28
3.5.1. CAC Based on Resource Availability …….……...…… 29
3.5.2. CAC Based on Number of Users ……………………… 30
3.5.3. CAC Based on SIR ……………………...……….….…. 30
3.5.4. Transmit and Received Power Based CAC ……...….…. 31
ii
3.5.5. Reservation Scheme……………………………….….…. 31
3.5.6. Linear Weighting Scheme………………………....….… 31
3.5.7. Distributed Admission Control Scheme ……………...… 32
3.5.8. Shadow Cluster ……………………………………….... 32
3.5.9. Call Bounding …………………………………..…….… 33
3.5.10. Cutoff Priority ………………………………………...… 33
3.5.11. Call Thinning …………………………………...………. 34
3.5.12. Miscellaneous CAC Schemes …………………………... 34
3.6. Summary ………….…………….……………………………..……35
4. Video Streaming Service in 3G W-CDMA Network …..….… 36
4.1. Problem Definition ……………………………………………...… 36
4.2. System Model ……………………………………….…..…… 37
4.3. Objective Function …….….………………………………....…… 39
4.4. Implementation …….…………………………………….....… 42
4.4.1. Algorithm Architecture …………………………….....… 42
4.4.2. Calculation of Path Loss and SIR …………...………..… 45
4.4.3. Packet Scheduling …………………………………….… 46
4.5. Summary …………………………….……………………….…..… 47
5. Conclusions and Future Directions ……….…………...… 48
References …………………………………..……………….……………… 50
Appendix A: Wireless Technology: History of Development …………….… 54
A.1 Magnetism ………………………………….……... 54
A.2 Electricity ……………………………….………... 56
A.3 Electromagnetism …………………………………... 57
Appendix B: Overview of W-CDMA ………………………….…….………… 59
B.1 Transport Channels …………………………………... 66
B.2 Physical Channels …………………………………... 66
iii
B.2.1 Dedicated Physical Channels …………. 67
Appendix C: Project Code …………………………………….………….…… 70
iv
List of Abbreviations
3GPP Third Generation Partnership Project
AMPS Advanced Mobile Phone Service
BER Bit Error Rate
BS Base Station
CAC Call Admission Control
CDMA Code Division Multiple Access
DL Downlink
Eb/No Bit Energy to Noise Density
FCC Federal Communication Commission
FDD Frequency Division Duplex
FDMA Frequency Division Multiple Access
GGSN Gateway GPRS Support Node
GMSC Gateway MSC
GPRS Global Packet Radio Service
GSM Global System/Service for Mobile Communication
HLR Home Location Register
IMT-2000 International Mobile Telecommunication 2000
ME Mobile Equipment
MPEG Multimedia Photographic Expert Group
MSC Mobile Switching Center
MT Mobile Terminal
PCS Personnel Communication System
PG Performance Gain
PSTN Public Switched Telephone Network
v
PL Path Loss
RNC Radio Network Controller
RNS Radio Network Subsystem
RRM Radio Resource Management
SGSN Serving GPRS Support Node
SIR Signal to Interference Ratio
SNR Signal to Noise Ratio
TDD Time Division Duplex
TDMA Time Division Multiple Access
UE User Equipment
UMTS Universal Mobile Telecommunication System
UPS Universal Parcel Service
USIM Universal Subscriber Identification Module
UTRAN Universal Terrestrial Radio Access Network
UL Up Link
W-CDMA Wideband CDMA
vi
List of Figures
Figure 2.1: Field created by a charged body.
Figure 2.2: Changing electric field due to moving negative charged particles
Figure 2.3: Formation cause of electromagnetic waves.
Figure 2.4: UMTS Network Architecture
Figure 2.5: UTRA proposed spectrum allocation
Figure 2.5: Multiple Access Schemes
Figure 2.6: Multiple Access Schemes
Figure 2.7: Spread Spectrum Technology
Figure 2.8: Jamming Operation in Spread Spectrum Technology
Figure 2.9: Spread Spectrum Technology
Figure 2.10: Performance Gain using Spread Spectrum
Figure 2.11: UTRA Physical Channels
Figure 2.12: Structure of UTRA Physical Channels
Figure 4.1: System Model
Figure 4.2: Packets Scheduling Model
vii
List of Tables
Table 2.1: Organizations Submitting Proposals for IMT-2000 System.
viii
List of Symbols
ηAD Threshold Value of Inter-Packet Delay.
ρ System Workload
C System Capacity
Cf Cost Function
CapacityResidual Residual Capacity of a System.
Da Average Inter-packet Delay
Dt1 Inter-packet Delay Threshold Value
Eb Energy per bit.
FA Cost Effectiveness Generated by Service_Level_A Users.
FB Cost Effectiveness Generated by Service_Level_B Users.
FC Cost Effectiveness Forecast from Waiting Service_Level_A Users.
FD Cost Effectiveness Forecast from Waiting Service_Level_B Users.
I Total Received Interference.
Pr Power Received by MT.
Pt Power Transmitted from Base Station.
RA Average Requested Data Rate.
RF Reference Rate.
Tv Total Volume
Td Total Volume Delivered
Tf Packets with Earliest Finish Time.
Tdif Traffic Volume Delivered for Interrupted Flows.
Tsim Simulation Time
Tc Current Time
ix
Vi Traffic Volume of Video Clip i.
W System Bandwidth
Z Effective Throughput
PAi , PBp Price for a certain date rate of class-A and class-B users respectively,
where PAi > PBp
NA, NB Total number of service_level_A and service_level_B data rates in
service respectively.
NWA, NWB Total number of service_level-A and service level-B data rates
respectively, waiting to get served.
Nusers_i Equivalent No. of Users in Class i.
SRAi , SRBp Number of service level-A and B users respectively, served at a
common data rate.
Q(k) UMTS QoS classes assigned weighting factors: Q(1)=4, Q(2)=3, Q(3)=2,
Q(4)=1
TRAj , TRBl Total serving time left for service level-A and B users respectively
served at a common data rate.
TRWAo , TRWBs Service time requested by waiting users of service level-A and B
respectively.
WRAi, WRBp Total number of waiting users asking for service level-A and B
respectively.
x
Acknowledgements
I take this opportunity to express thanks to my supervisors Dr. Marinal Mandal
and Dr. Ehab Elmallah for their continuous guidance and support throughout the
project. Apart from that, I will never be able to forget the support I have received
from my department, family members, seniors and friends during one of the worst
patches of my life (the car accident); without their love, help and support I would
never be able to finish this important task. I would like to individually name some of
my friends, seniors and family members as well to show the importance of their
contributions towards my rehabilitation to get enough physical and mental strength to
finish my project work.
Ishtiaq Bhatti and Family, Elder Mamu Jee’s Family, Uncle Sultan Bhatti’s Family,
Saeed Bhatti and Family, Elder Khalu Jee and Family, Zia Bhatti and Family,
Waheed Bhatti and Family, Majid Bhai and Family, Uncle Majeed and Family, All
Bhatti Families, my and Family friends, neighbours (who visited me to show their
support), Iffat Tamoor and Family, Tahir Siddique and Family, Imran and Family,
Baba Jee and Family, Liaquat Ali, Shamim, and Other Hiking Club Friends, Uncle
Abbas and Family, Uncle Barkat Ullah Bhatti and Family, Abbas and Zafar, Bhatti,
Dr. Farheen Imran, Dr. Zafar, Dr. Afridi, Qari Idrees, Chacha Saeed, Wasi uz Zaman,
Dr. Sohail Zubairi, Dr. Ishtiaq Ahmad, Dr. Tahir Rasul, Uncle Nazeer, Mike, Adil
Akbar and Family, Mumtaz Sb. and Family, Tariq Sb. And Family, Dr. Nadeem
Khattak and Family, Munawwar and Family, Iqbal Soomro and Family, Amir,
Khurrum Shehzad, Shahid, Dr. Tajjamul, Ghulam Farooq and Rizwan.
xi
Chapter 1 Introduction
This chapter presents an overview of our project work. It starts with mentioning
the importance of mobile communication in our daily lives and what are the new
challenges the industry is facing. It then mentions our primarily work on designing an
objective function for 3G W-CDMA mobile wireless network. The objective function
calculates the effectiveness of cost functions by considering factors like QoS and
users satisfaction. The service provider can use this figure of merit to determine if at a
given time instance the service is generating profit, if not system has to be improved.
The use of mobile phones has increased rapidly in recent times. Since consumer
dependence on these products has phenomenally increased, the service providers are
improving the system further so that consumers will be offered better services on
mobile phones for a profit and growth in the industry.
This is what has led to the development of 3G UMTS mobile wireless
infrastructure. The planning, design and recommendations for 4G and onward
systems is well underway. The 3G UMTS infrastructure is designed for providing
multimedia services in mobile wireless environment. It primarily divides the offered
services into four QoS classes: Conversational, Streaming, Interactive and
Background.
In a wireless cell, the users mobility is a challenge for the service provider due
to the Near Far Effect. The base stations generally have has fixed power budget. In
other words, the combined transmitted power from a base station to users mobile
equipment cannot exceed a certain maximum (for example 50 W). The users moving
away from the base station require more power whereas the users coming closer
require less power from the base station.
1
In this project, we have designed an objective function that will demonstrate the
effectiveness of cost functions. At any given time instance, it can be determined from
the objective function if the offered services are generating profit for the service
provider. In addition, this objective function will provide a trade off between the
profit margin and user satisfaction. This is very important because if we keep on
serving users paying less in comparison to those who are waiting and willing to pay
more for a particular service, service inefficiency will result. In worst case, this may
lead to events such as corporate downfall, downsizing, cut backs and layoffs.
Designing an efficient objective function is a difficult task especially if
considered for an ideal 3G UMTS wireless environment, where users are requesting
services belonging to one, multiple or all of the UMTS QoS classes in a multiple
cellular environment. Therefore we have narrowed down the problem for this project.
We have decided to evaluate performance results for Video Streaming service in a
single 3G UMTS/W-CDMA mobile wireless cell. In order to achieve our objectives
we considered the following core issues:
• Packets Scheduling.
• Call Admission Control.
• Power & Rate Control.
• Rate Control.
• Predict effectiveness of cost functions.
Packet scheduling arranges queues for incoming users streams. Hence the main
task is to minimize queuing delays and deliver packets in choice of the objective
function. Call admission control deals with the problem to see if/which new service
call(s) can be admitted in the system from waiting users in choice of the objective
function. This will in turn look for the power and data-rate requirements for that
(those) particular user(s) for fixed time duration. Our formula will then take care of
predicting the output of cost functions, which if exceeds a certain threshold indicates
that the service provider will meet its profit margin target by admitting new user(s)
2
along with the currently adopted packet scheduling, power and rate control schemes.
Chapters 4 cover these aspects in precise details, giving out our assumptions and
proposed service contract.
Significant work has been done by different researchers on the first four issues
mentioned above. However, not much work has been done in the fifth issue. Hence,
we are addressing this problem in this project. So, it is a new challenging problem we
are addressing in our project work.
Organization of the report:
The report is organized as follows:
• Chapter-2: It first provides a historical perspective on mobile wireless
industry and then gradually presents the development 3G mobile wireless
networks.
• Chapter 3: This chapter covers details on the requirements and trade offs
of providing video services over mobile wireless infrastructure.
• Chapter 4: This chapter provides details on our work on the development
of an objective function for video streaming service in a single 3G
UMTS/W-CDMA wireless cell.
• Chapter 5: It presents concluding remarks and future directions.
• Appendix A: It presents an overview of the historical development of
wireless technology.
• Appendix B: It provides an overview of W-CDMA access scheme for 3G
mobile wireless networks.
• Appendix C: It presents the MATLAB code for evaluating the performance
of the proposed objective function.
3
Chapter 2
Introduction to Mobile Wireless Networks
The importance of mobile phones has increased phenomenally in recent times.
The Internet growth use of critical applications like email, e-news and web browsing,
have made the use of mobile phones more than simple voice communication. First
and second generation (1G and 2G) mobile wireless phones were primarily aimed at
providing efficient services relating to voice communication. The growth in
computers, networks and specially Internet has opened doors to many critical users
applications. Peoples dependence on such applications were the cause of developing
future mobile wireless networks. The chapter presents a brief overview on the critical
development stages leading to mobile wireless networks.
2.1. Wireless Telephones, Computers and Networks
Successful wireless transmission was demonstrated by, Morse in 1832. But it
was not possible until 1921 to achieve commercial wireless voice communication. In
1921 the mobile radios started operating in the 2 MHz range, and the system was used
by the New York law enforcement departments. The system was half duplex and used
amplitude modulation. It had some limitations, for example bad weather conditions
affecting transmissions, size of radio equipment and operating cost. In 1924, Bell
Labs had invented the bi-directional wireless voice band, which was one step closer
to make it possible to offer the system for public use. The discovery of frequency
modulation by Armstrong in 1935 paved the way for improved voice communication
and the reduced equipment size, because with FM modulation can be done at high
frequencies. There was rapid development (e.g. development of circuit boards) of
wireless technology during Word War-II. The first commercial mobile phone service
4
(AMPS, 800 MHz) was offered in 1947 in St. Louis, Missouri using a single
transmitter for a large area. However due to F.C.C. limited frequency use policy, only
23 simultaneous connections were possible in a certain area. The mobile phones were
offered only 6-channels with 60 KHz spacing, which resulted in frequent cross talk or
poor voice quality. AT&T later developed the concept of using several small
transmitters for smaller areas. However this did not solve the voice channel problem.
The demand of mobile phone users was huge. In 1976 there were approximately 600
mobile users in NY and 3500 were on the waiting list; 45000 mobile users in entire
USA and 20000 on the waiting list. Thus further research and development was
essential in order to sustain the market and keep customers happy. PCS mobile phone
system with services like messaging, paging and voicemail was offered as an
improved system focusing users requirements in the 1850 MHz range using FM. The
invention and improvements in digital technology lead the development of system we
now see in our daily life, such as GSM, TDMA, and now leading to CDMA and W-
CDMA.
The development of wireless technology in the current era is heavily dependent
on computers and networks. Hence it is very important to have some necessary
background knowledge on the development of computer networking. This review will
give us an understanding of how technology automatically find its ways for further
growth.
In 1821, Charles Babbage invented a machine called Difference Engine. The
system was developed in an effort to solve numerical problems easily by using
method of Finite Difference. While solving Polynomials this system avoids
multiplications and divisions, and actually uses simple additions to solve the problem.
It was a mechanical system used to calculate series of numerical values and printing
results automatically. The inspiration must have come from the earlier Calculator
development works of Pascal, Leibnez and Schickard. This was the basis of computer
development, computers in our era uses simple addition to solve multiplication and
division problems. So, basically its all binary addition inside a computer to calculate
arithmetic. Technology development efforts in World War-II resulted in the
5
development of first decoder called Colossus, which was developed to break German
codes. It was a slow machine taking about 3-5 seconds for each calculations. John
Presper Eckert developed a decoder (ENIAC), thousand times faster than Colossus
taking 160KW power using vacuum tubes. When it was run for the first time. The
invention of Transistor lead by Shockley, Brattain and Bardeen in 1948 at Bell Labs
and is one of the most important inventions in 20th century technology development.
This invention made it possible to develop smaller electronic devices (low power
consumption). Eventually, the IBM in 1981 introduced a product named PC
(Personnel Computer) for public use, the IBM sold 2 millions PCs in 1981, and sold
65 millions more over the next ten year period. Currently as we all know having a
computer in house has become essential to keep ourselves well informed, efficient
and more productive.
As the number of computers increased, first in offices and later at homes, it
created a strong need for networking them. This resulted in the development of
Ethernet in 1970 for LAN (Local Area Network), WAN (Wide Area Network) and
MAN (Metropolitan Area Network).
2.2. Wireless Networks
Since wires have a high cost for setting up a large network. There is a need to
have wireless LANs and we do see wireless LANs in homes, offices, and now it is
reaching to mobile phones in the 3G framework. Wireless phones have started from
simple voice communication (1G) service to new services such as video conferencing
and video streaming (3G).
The wireless technology offers convenience for people in hotels, airports and
other travel spots, who need access to services such as email, voice and web
browsing. The solution to this problem is Wireless Internet Service Provider (WISP).
The demand of wireless networks is ever increasing in the horizontal (public safety,
monitoring applications, delivery services, finance and retail) and vertical markets
(e.g. WAP, SMS, GPS and web surfing). For example, package delivery services,
6
such as UPS, use single channel system model called ESMR (Enhanced Specialized
Mobile Radio).
2.3. 3G Mobile Wireless Networks
Work on 3G mobile wireless networks has been started by ITU’s radio
communication sector (ITU-R) task group 8/1 in late 80’s. The group defined the
requirements for 3G mobile wireless networks. Initially the recommendation were
known as Future Public Land Mobile Telecommunication System (FPLMTS). The
frequency spectrum for such a network was decided on a worldwide basis in 1992,
and these were (1885-2025 MHz and 2110-2200 MHz bands). In year 2000, the
FPLMTS got a new name as IMT-2000 (International Mobile Telecommunication
System-2000). It provides a framework to support services ranging from few Kbps to
2 Mbps data rate requirements, and have transparent worldwide radio coverage for
global roaming. This provides a base infrastructure to connect any two mobile
wireless terminals worldwide. The design of IMT-2000 also takes into consideration
the different propagation requirements, and has the ability to handle circuit and packet
data mode services of variable data rates. All these improvements provide an
acceptable quality of service (QoS), which is competitive to wired networks at a
reasonable cost to the consumer.
Several international telecommunication organizations put their efforts in
developing and standardizing the IMT-2000 proposal. The ETSI (European
Telecommunication Standards Institute), TIA (Telecommunications Industry
Association) of USA, and ARIB (Association of Radio Industries and Businesses) of
Japan are prominent members of that group. In June 1998, fifteen IMT-2000
proposals were submitted to ITU-R in which five proposal were for satellite based
communication solutions. Table 2.1 lists the organizations that submitted proposals. It
can easily be observed in the table that most of the organizations suggested using W-
CDMA technology for 3G wireless networks. Features such as improved user
capacity, coverage in most propagation environments, ability to solve multi-path
fading problem through RAKE receiver, ease in frequency planning due to use of a
7
single band for each user, instead of having a unique band for each user as in the past,
made W-CDMA an attractive technology.
Table 2.1: Organizations Submitting Proposals for IMT-2000
Proposal Description Multiple Access Source
DECT Digital Enhanced Cordless Telecommunications
Multi carrier TDMA (TDD)
ETSI Project (EP) DECT
UWC-136 Universal Wireless Communications
TDMA (FDD and TDD)
USA TIA TR45.3
TD-CDMA Time Division Synchronous CDMA
Hybrid with TDMA/CDMA/SDMA (TDD)
Chinese Academy of Telecommunication Technology (CATT)
W-CDMA Wideband CDMA Wideband DS-CDMA (FDD and TDD)
Japan ARIB
CDMA II Asynchronous DS-CDMA DS-CDMA (FDD) South Korean TTA
UTRA UMTS Terrestrial Radio Access
Wideband DS-CDMA (FDD and TDD)
ETSI SMG2
NA: W-CDMA
North America Wideband CDMA
Wideband DS-CDMA (FDD and TDD)
USA T1P1-ATIS
Cdma2000 Wideband CDMA (IS-95) DS-CDMA (FDD and TDD)
USA TIA TR45.5
CDMA I Multi band synchronous DS-CDMA
Multi band DS-CDMA South Korean TTA
8
The joint efforts of several standard regional organizations towards
standardizing IMT-2000 resulted in two partnership projects: 3GPP1 and 3GPP2. The
3GPP1 project aim at developing technical specifications for IMT-2000 based on
GSM (Global System for Mobile Communication) and UTRA (UMTS Terrestrial
Radio Access). The ETSI (European), ARIB (Japan), CWTS (China), T1 (USA),
TTA (Korea), TTC (Japan) are the organizations involved in 3GPP1. The first
specification for UTRA was released, in December 1999.
The objective of 3GPP2 is to develop technical specifications for IMT-2000
based on CDMA2000 RTT and ANSI-41 core networks. The project is headed by
TIA and other members include TTC, ARIB, TTA and CWTS.
2.4. Overview on UTRA
RACE (Research in Advanced Communication Equipment) and ACTS
(Advanced Communication Technologies and Services) were the research initiatives
taken by EU (European Union) in 1988 and 1995, respectively, for advancement in
the development of UTRA architecture. RACE program had the objectives to
research and develop pilot projects for the possible UMTS air interface technologies.
ACTS project finally chose two technologies for accessing UTRA air interface, and
those were TDMA (Time Division Multiple Access) and W-CDMA (Wideband Code
Division Multiple Access). It was ARIB who first made decision on adopting
and focusing research on W-CDMA in January 1997. The decision was later followed
by ETSI in January 1998.
The initial spectrum allocation for UTRA is shown in Fig. 2.5. This frequency
spectrum proposal came with the philosophy of voice and low date rates services
dominating the IMT-2000 market. The philosophy became invalid soon with
increasing demand of high date rate services such as video streaming, video
conferencing and digital imaging. It has been forecasted that the proposed frequency
bands allocation can sustain until year 2005. High data rate services over UTRA
9
demands an additional 187 MHz of frequency spectrum. It has been forecasted that
such addition will be enough to fulfil the services data rate requirements by the year
2010.
Uu Iu
CellCell
CellCell
CellCell
CellCell
USIM
ME
Cu
RNS
RNS
BS-1 / Node B
BS-N / Node B
BS-1 / Node B
BS-N / Node B
UE UTRAN CN Ext. Net.
RNC
RNC
MSC / VLR
SGSN GGSN
HLR
Circuit Switching Gateways
Packet Switching Gateways
PLMN:PSTN, ISDN
Internet
Iub Iur
Circuit Switching Domain
Packet Switching Domain
GMSC
CellCell
CellCell
CellCell
CellCell
USIM
ME
Cu
RNS
RNS
BS-1 / Node B
BS-N / Node B
BS-1 / Node B
BS-N / Node B
UE UTRAN CN Ext. Net.
RNC
RNC
MSC / VLR
SGSN GGSN
HLR
Circuit Switching Gateways
Packet Switching Gateways
PLMN:PSTN, ISDN
Internet
Iub Iur
Circuit Switching Domain
Packet Switching Domain
GMSC
Figure 2.4. UMTS Network Architecture
DECTW-CDMA
(TDD) MSW-CDMA
Uplink (TDD)W-CDMA
(TDD)W-CDMA
Downlink (FDD) MS
1885 1900 1920 1980 2010 2110 2170 2200
Figure 2.5. Proposed spectrum allocation in UTRA
Frequency (in MHz) →← Frequency (in MHz) →←
10
UTRA development is heavily influenced by the GSM and GPRS networks, and
the architecture mainly composed of two parts: UTRAN and CN. UTRAN provides
air interface for the user mobile terminals where as CN is responsible for routing and
switching voice and data connections to external networks. The architecture has open
interfaces enabling operation or communication among equipments from different
manufacturers. UTRAN contains several Radio Network Subsystem (RNS); each
RNS is composed of several Radio Network Controller (RNC), and each RNC
controls several Base Stations (BS’s) and User Equipment (UE). RNC is responsible
for controlling radio resources and plays a major role in HC (handover control), LC
(load control), PC (power control), AC (admission control) and packets scheduling
(partially done by RNC). “Iur” interface is used for soft handover among RNC’s,
“Iub” to control one BS/Node-B, and “Iu” to interface with CN.
In Fig. 2.4, Node B represents base stations (BS). In UTRA architecture the
BS’s are similar to BS’s used in GSM networks. Base station is a physical unit for
sending and receiving radio signals in cell(s). It’s responsibilities include soft
handover, inner closed loop power control, signal spreading, interleaving, coding
channels and rate adaptation. Base station accesses user equipment (UE) via the W-
CDMA Uu interface. A UE comprises of ME and USIM; and these modules use Cu
interface for communication. The ME is user’s mobile equipment whereas USIM is a
smart card containing unique subscriber identification number and personnel
information. In the core network block of the UMTS network architecture, the home
location register (HLR) is responsible for keeping track of subscribers location and
charging and routing calls to the respective (where mobile phone is registered at that
time) MSC or SGSN. The MSC provides an interface point between the mobile radio
network and the fixed network of the outside world, and thus handles all circuit
switching requests. The visitor location register (VLR) working along with MSC has
the responsibility to keep track the location of roamed mobile terminal. GMSC and
GGSN, MSC and SGSN are similar in functionality, with a difference that MSC and
SGSN serve as the gateway and switching centre for the packet switched network,
respectively.
11
2.5. Importance of Power Control in UTRA
The base stations in a cell have fixed power budget, The combined transmitted
power to all the mobile terminals (MT) cannot exceed a certain maximum (for
example, 50-watts). Thus an efficient power control is necessary. It is also necessary
to solve the near-far problem [3] in a CDMA network. Power control also determines
the capacity and coverage of the system. In FDD mode of operation, the power
control mechanism employed is called Closed Loop Power Control (CLPC). In TDD
mode, it is called Open Loop Power Control (OLPC). We will discuss the two types
of control briefly in the respective order. Before discussing power control, it is
important to understand the following SIR equation for each MT. The ultimate
objective is to get a target Signal to Interference Ratio for each Mobile Terminal.
rx
intra_cell inter_cell o
PWSIR= (in dB)R (γ I + I + η W)
(2.1)
where,
W = Chip Rate in Hz.
R = Transmission Data Rate in bits per second.
Iinter_cell = Interference received from neighboring cells
Iintra_cell = Interference received within cell
Prx = Power received from BS
ηo = Noise Power Spectral Density
γ = Orthogonal Factor
In Eq. 2.1, Prx represents the total power received from BS’s and the
denominator term (excluding R) represents the total interference received at a MT.
12
The CLPC is used both for Uplink (UL) and Downlink (DL) using TPC
commands that are transmitted in the format shown in Fig. A.12(a, b). Hence,
discussing UL or DL power control is equivalent. However, when it comes to
designing an effective CAC scheme for handling UL and DL, give rise to remarkably
different problems.
In UL power control, BS measures received power from an MT on its
respective DPDCH and DPCCH after RAKE. The BS also estimates the total received
interference (information other than from the target MT) in order to estimate the total
received SIR, which is compared to the target SIR for that MT. This operation is
performed every few milliseconds. Based on these information, BS sends TPC
commands to respective MT in order to either increase or decrease transmission
power of DPDCH and DPCCH by a step size of ∆TPC dB (typically 1 or 2 dB) to
achieve target SIR. The transmissions at unnecessary high power, not only reduces
battery life but also results in performance degradation on service quality for other
users.
In OLPC, an MT has a priori knowledge about the interference level at the BS
and DL path loss, sent by BS. Since information is already known, the MT can easily
calculate the required transmit power in order to achieve the target SIR.
13
2.7. Summary
This chapter provides a brief introduction to the wireless networking
technology. It starts by presenting an overview on the technological developments of
wireless networks. A brief overview on the development of computers, wire-line
networks and wireless telephones is given in section 2.1, since these are the major
contributors in the development of wireless networks. We then gave brief overview
on the evolution of 3G mobile wireless networks in section 2.3. A brief review on
UTRA has been presented. The importance of power control in mobile wireless
environment was described in section 2.5.
14
Chapter 3
Video Transmission Over Wireless
Networks
People’s dependence on Internet and growth in mobile wireless industry have
generated new requirements for the industry in order to provide better services to
users and profits for investors. Examples include mobile access to Internet (for
example email, browsing, stocks and banking), video conferencing and video
streaming etc. Mobile access to Internet has already been offered even in 2G mobile
wireless networks. Such a service can tolerate delays while browsing or downloading
emails, but for video service the case is not the same, since most users do not want to
receive still or distorted video clips. Such a condition may be caused either by a
fading channel or network congestion resulting in data rate decrease for the mobile
user. Hence, delivering uninterrupted acceptable quality video service is a real
challenge to address to have satisfied customers.
The traditional concept of treating video as data to achieve certain QoS
objectives fails in wireless networks, since the overall network bandwidth cannot be
guaranteed at a certain time instance. We note that digital video signal is a collection
of frames at a rate of 10-30 frames/sec (FPS). In other words, as long as these 10-30
frames can be delivered every second to the target MT, the QoS objective will be
achieved successfully.
3.1 Video Service in Mobile Wireless Networks
Digital video is a collection of frames, where a frame refers to a still
picture/image. A full depth (8 bits each for red, green and blue, total 24 bits for each
pixel) color image of size 358x288 requires 304128 bytes of storage space, whereas
for a reduced depth colour image (4 bits each for red, green and blue color
components, total 12 bits per pixel) requires 152064 bytes of storage space. The
15
following shows the bandwidth requirements, if we want to send a full and reduced
colour depth video over wireless link.
Data Rate Requirements for Uncompressed Video:
Full Colour Depth Video:
30[fps] x (358x288)[resolution] x 24 [bits per pixel] = 36.5 Mbps
Reduced Colour Depth Video:
30[fps] x (358x288)[resolution] x 12 [bits per pixel] = 18.2 Mbps
Television Quality Video:
30[fps] x (704x576)[resolution] x 12 [bits per pixel] = 146 Mbps
The above bit rate shows that efficient and cost effective transmission of
uncompressed video signal is not possible due to very high data rate requirements.
Compression is the only solution to the problem. There are several very efficient
standards for image (e.g. JPEG) and video compression (e.g. MPEG).
The key aspect in guaranteeing video QoS over wireless networks is the use of
compression. The level of compression depends on channel quality. It is low for good
channels and high for bad channels. Delivering video in 3G wireless networks has
become possible due to the development in compression technology. Now let us take
a brief look on the existing video coding standards.
16
3.2 Digital Video Coding Standards
A video is a collection of frames and each frame represents an image. Video
compression is primarily achieved by transmitting only changes occurring from one
frame to another, and by removing information undetected by viewer’s eyes. The
number of bits used for coding video signal is proportional to the change between two
consecutive frames. There are five widely adopted methods of video compression that
are as follows,
1. Discrete Cosine Transform (DCT): It is a lossy compression algorithm,
analyzing presence of frequency components in samples taken at regular
intervals, and discarding unwanted information as its perceived by viewers
eyes. DCT is utilized in several standards such as, JPEG, MPEG, H.261, and
H.263.
2. Vector Quantization (VQ): A lossy compression algorithm looking at an array
of data and removing redundancy.
3. Fractal Compression (FQ): A similar compression scheme like VQ, where
compression is performed by locating redundant sections of an image and are
regenerated by using a fractal algorithm.
4. Discrete Wavelet Transform (DWT): This compression scheme
mathematically transforms an entire image into frequency components. Its
different from other schemes, which works on sections of an image.
5. Motion Compensation: This scheme keeps track of objects motion in
successive video frames to accomplish compression. It is an effective scheme to
reduce temporal redundancy among consecutive frames.
MPEG (Moving Pictures Expert Group) is a known standard in video
compression technology. The MPEG group was established in 1988 by ISO/IEC for
research and development in digital audio and video compression technology. The
following standards have been developed by MPEG.
17
• MPEG-1: The main focus is to store full motion video in CD-ROMs. It can
support data rates requirements up to maximum of 1.5 Mbps, in which 1.15
Mbps is for video and 192 Kbps for audio. MPEG-1 supports up to 352 x 240
resolution and frame rate of 24. It is a popular standard for distributing videos
over Internet (i.e. .mpg files) and making VideoCD. This is the most popular
video distribution format in Asia.
• MPEG-2: Applications requiring data rates between 1.5-15 Mbps. It was
initially developed for digital television broadcasting. But it is now also used
for DVD compression and high definition digital television (HDTV). For
HDTV the transmission requirement is 18 Mbps. The significant improvement
in MPEG-2 is the ability to compress interlaced video. It supports 720 x 480
resolution and frame rate of 30 or 60 interleaved fields per second.
• MPEG-4: The main focus of this standard is to bring television experience on
desktops. Its was primarily designed for data rates which dialup modems can
support, like 28.8 Kbps, 33.6 Kbps and 56 Kbps. Channels with such capacity
can support frame rate of 12 or 15, with a maximum resolution of 172 x 288.
Technology advancement in data rates for Internet connectivity made it possible
to support higher resolutions and frame rates. MPEG-4 is an object based
technology, compressing and keeping track of different objects in an image.
• MPEG-7: A standard still under development, the main focus is to provide
information about the content, which will include security and integrity,
personalization, filtering and content manipulation.
In addition to MPEG standards, there are other video compression standards
developed by ITU, such as H.261, H.263 and H.264. There are briefly discussed
below,
18
• H.261: The standard was developed for video conferencing over ISDN lines,
supporting data rates, which are multiples of 64 Kbps. Based on in DCT, the
algorithm uses inter-frame and intra-frame compression. It supports QCIF
(Quarter Common Intermediate Format, 176 x 144) and CIF (352 x 288)
resolutions and can be employed in hardware or software.
• H.263: Several enhancements were introduced in this standard over H.261, to
improve video quality over modem connections. It supports CIF, 4CIF (704 x
562), 16CIF (1408 x 1152), QCIF and SQCIF (128 x 96) resolutions. The CIF
standard is also known as FCIF (Full CIF).
• H.264: This standard is a combination of MPEG and H.26L. The ITU-T VCEG
(Video Coding Expert Group) initiated work on H.264 in 1997. Later by the end
of year 2001, MPEG and VCEG joined together to form JVT (Joint Video
Team) and H.264 project was taken over. Here are some of the key advantages
offered by H.264.
o Bit rates savings up to 50%.
o High quality video
o Improved error resilience
o Adaptable to heterogeneous networks
H.264 is a strong candidate for video compression in mobile wireless
networks.
19
3.3. Video Streaming in Mobile Wireless Networks
Our work in this report is on facing the challenge and designing an objective
function for an efficient and profitable video streaming service on 3G mobile wireless
networks. The demand of Internet browsing and its related technologies such as
email, video streaming, video conferencing, online banking and IRC, are in high
demand. As we all know “Need is the mother of invention”, the development of
mobile wireless networks is driven by consumer needs. First generation mobile
wireless networks were analog systems used only for voice communications. Later
came PCS providing efficient services like voice mail and paging.
The rapid development and growth in Internet technologies forced the mobile
wireless network investors to seriously think on providing these services to
consumers of their industry for ease of access, which will increase mobile phones sale
and hence a growth in profit margin. These reasons and previously compiled
problems of mobile wireless networks lead to the development of 3G mobile wireless
networks, which has its roots in technologies like UMTS and W-CDMA. These
technologies provide enough bandwidth requirements to cope services mentioned
above.
Internet growth has proven the efficiency, flexibility and cost effectiveness
offered by IP protocol. Thus implanting IP in wireless networks would be ideal in
order to merge two different worlds (wired and wireless). In other words, services
such as voice, video steaming and conferencing, web browsing, IRC, email can be
exchanged in packet form among mobile wireless and wired network/Internet. We
will focus our discussion in particular to video streaming service. There are two main
bottlenecks in delivering video streaming service over mobile wireless channel.
1. Delivering information within the maximum allowable delay to meet quality of
service objectives and to avoid jitter.
2. The condition of a wireless channel depends significantly on the amount of
network load and the weather conditions.
20
The design of IP networks was primarily aimed at providing delay insensitive
services like web browsing and emails, whereas the design of mobile wireless
networks aimed at providing efficient voice service, which is a delay sensitive
service. Thus creating a challenge in providing delay sensitive services over IP
networks. This also includes video streaming and video conferencing services, which
also requires guaranteed throughput and are delay bound.
Providing guaranteed QoS for video streaming over mobile wireless networks is
difficult since it is difficult to guarantee the followings:
• Packet loss, end to end delay, and delay jitter due to bursty errors caused by
several factors including weather conditions.
• Service bandwidth due to host mobility, handoffs etc.
The base station in a cell has a fixed power budget, varying weather conditions
and user mobility etc. Thus an efficient RRM (radio resource management) scheme is
required in order to achieve services QoS objectives. There are different approaches
to address the problem of efficient multimedia services over wireless channels, and
these include studies on MAC protocol, admission control and packets scheduling.
Our study focuses on packets scheduling and admission control, since the two are
inter-dependent for the design of an efficient and cost effective objective function.
3.4. Packet Scheduling Methods in Mobile Wireless
Networks
Second generation wireless systems enabled voice traffic for commercial use.
Since then the increase in customer’s needs and demands paved the way for the
development of third generation wireless systems, aimed at multimedia traffic
requiring higher data rates. This also opened door for a wide range of applications to
fit in the aimed data rates. That is why it been realized to categories traffic into
different classes of varying QoS requirements. In W-CDMA/UMTS the QoS has been
divided into following four classes.
21
1. Conversational Class
a. For highly delay sensitive applications: video, VoIP (Voice Over IP).
b. Example: Packet or circuit switched core network in UMTS architecture.
2. Streaming Class
a. For applications like streaming video and music.
b. Example: Packet or circuit switched core network in UMTS architecture.
3. Interactive Class
a. For non-real time services like web browsing and IRC.
b. Packet switched core network.
4. Background Class
a. For delay insensitive application such as email and advertisements.
b. In this class, the bit rate is not guaranteed.
Efficient Radio Resource Management (RRM) techniques are required to
achieve the set QoS requirements, while maximizing systems capacity at the same
time. Future evolution of 3G systems are aiming to provide wireless services on an
end-to-end IP network. This is a clear indication that technologies dealing with packet
switched services will gain momentum. The optimization of systems capacity, a
desired target, will depend on the performance of QoS and RRM. Such algorithms
should also exploit the bursty nature of packetized traffic for enhancing the overall
system capacity. Following are some key characteristics of a 3G wireless channel.
• Dynamically varying capacity of wireless channel.
• Channel Contention among mobile hosts.
• Channel errors are bursty and location dependent.
• Scheduling scheme must efficiently tackle uplink and downlink flows.
• Mobile hosts have processing power and battery life time as bottlenecks.
• Mobile hosts do not compete to occupy channel(s) in order to transmit data.
22
The focus of our work is Streaming Class and aimed in particular to video
streaming service in 3G mobile wireless networks. Packet scheduling algorithms
aimed at achieving all or a set of the following objectives.
• Determination and allocation of available radio resources (power and time).
• Monitor radio resource allocation for different services.
• Sharing of radio interface resources among services.
• Monitoring and controlling system load.
Packet scheduler allocates resources for both UL and DL due to asymmetric
nature of user traffic. While allocating radio resource for one direction only the
algorithm has to reserve resources (low date rate channel) for other direction in order
to carry high layer (TCP), data link layer acknowledgements and for power control.
There are two main categories of packets scheduling methods. An algorithm is
typically a combination of the two categories.
1. Time Division Scheduling
o Normally used with shared channels.
o Capacity allocation to few mobile users at a given time instance. The
allocated bit rates can be very high.
o System load determines the scheduling time.
o Downlink shared channel is typically a time division scheduling.
o Advantages: high bit rates/low delay, low SIR.
o Disadvantages: Short transmission time, high setup overhead, high
variations in interference levels.
2. Code Division Scheduling
o Normally used with dedicated channels.
23
o Capacity allocation to large number of mobile users. Low bit rates are
allocated for each service.
o System load determines bit rates allocation.
o Advantages: long transmission time, low setup overhead, no variations in
interference levels.
o Disadvantages: low bit rates / higher delay, high SIR
Following are some popular packet scheduling methods, and their special
features.
First Come First Served (FCFS)
o In this method, the first packet coming in is the first going out to the
destination.
o It works well if no congestion. However the method is inefficient in
handling poor quality links.
Priority Queuing
o Classified packets sent to prioritised queues (0 to N-1), thus packets in
higher priority queue get served first. Queue i gets service only when
queue (i-1) got completely served.
o Advantages: High priority queues achieving low delays, high throughput
and bandwidth.
o Disadvantages: Lower priority queues performance is heavily dependent
on the performance of higher priority queue.
24
Round Robin
o Classified packets sent to respective queues (0 to N-1), and get served in
the same order. This approach is suitable for fixed size packets such as in
ATM networks, since algorithm assumes fixed time slot for each packet.
o Disadvantages: This method cannot guarantee bandwidth or delay. In
addition, it is Insensitive to packet size.
Window Priority Queuing
o Classified packets sent to prioritised queues (0 to N-1). Only a limited
number of packets say N per queue gets served per servicing round. Thus
lower priority queues get attention in servicing time.
o In this scheme, we can easily observe that a large N corresponds to
Priority Queuing, and small N corresponds to Round Robin.
Virtual Clock
o The main aim of this scheduling scheme is to guarantee bandwidth per
flow.
o Classified packets are sent to queues having associated bandwidth
specifications.
o Packets are ordered according to TimeStamp, which is calculated using,
Packet SizeTimeStamp[n] = TimeStamp[n-1] + Queue Rate
(3.1)
o This is still not a fair approach, because classes using idle bandwidth will
be penalized later.
25
Weight Fair Queuing (WFQ, WF2Q, WF2Q+)
o Delivery packet gets selected from a set of eligible packets using
following parameters [queue #m, current time (Tc), packet start time (Ts),
packet finish time(Tf), Packet on top of queue #m (Pm) ]
o The earliest finish time (Tf ) for packet Pm using the following equation,
mf m
Length(P )P ) = + TRate(m) cT ( (3.2)
o This scheme ensures fairness based on queue parameters instead of
packets properties.
Channel State Dependent Packet Scheduling Algorithm (CSDPS)
CSDPS algorithm tries to ensure error free scheduling service by
employing weighted round robin scheme. An error free channel flow is
allocated a slot, and whenever channel errors occur for this particular flow the
algorithm skips the flow and allocates the slot to another error free channel.
Basically it performs weighted round robin among flows to filter out clean
channels. Thus CSDPS does not measure flows lag or lead times, which also
sets no compensation as the bottleneck. Its implementation complexity is low.
Server Based Fairness Algorithm (SBFA)
This scheme reserves some channel bandwidth for compensation to
backlogged flows. If a backlogged flow, which has been allocated a slot, cannot
transmit due to channel errors then such flow queues a slot request in the
compensation flow. The scheme serves the compensation flows along with
other flows. SBFA scheme does not have the concept of leading flow. When
compensation flow has no slots available, the requested bandwidth is shared by
all flows carrying clean channels at that particular time instance.
26
Wireless Fair Service Algorithm (WFS)
It is an enhanced version of WFQ, in which each flow is allocated two
parameters, the rate weight (ri) and delay weight (Φi.). The start tag is computed
same as in WFQ but computation of finish tag is based on Φi. rather than ri. In
WFS the lag for a particular flow (caused by erroneous channel condition)
increases when some other flow (perceiving good channel) can take its place.
The lead and lag times are bounded by per flow parameters.
o Wireless Packet Scheduling Algorithm (WPS), Channel-condition
Independent Fair Queuing (CIF-Q), Idealized Wireless Fair Queuing
(IWFQ) are some other variations of wireless fair queuing algorithms,
whereas General Processor Sharing, Self Clock Fair Queuing (SCFQ),
Stop and Go and Rate Control Service Discipline (RCSD) are some other
methods used for packets scheduling.
Following are the approaches we have considered in our project work for packets
scheduling:
• First Come First Served (FCFS): Here the packet(s) are selected such that
intra-cell or inter-cell interference requirements are within the specified
limits.
• Best Channel First (BCF): Here the channel condition for each user is
different and are uncorrelated. Hence the available resources are utilized
for users having better channel conditions and requiring less radio
resources (e.g. transmit power). The major problem with this approach is
the long term fading for users having poor channel condition for extended
time.
27
3.5. Call Admission Control Methods in Mobile Wireless
Networks
Call Admission Control (CAC) is defined as a system for managing arriving
traffic, at the call, connection or session level based on some predefined criteria. A
CAC scheme either admits, rejects or delays the incoming calls based on a criteria to
achieve some QoS objectives for new and existing users.
In a mobile wireless cell, the base station has limited power resources. In other
words the combined transmitted power to all mobile terminals cannot exceed a certain
maximum level. Each newly admitted call makes this resource further scarce. Hence
it is very important to adopt policy that will determine which calls to efficiently
admit, block or reject based on available resources.
The following outlines some of the basic needs for which CAC is employed.
• Maximize revenue.
• Fair resource sharing.
• Guarantee transmission rate.
• Guarantee QoS at packet level.
• Some UMTS classes given priority over others.
• Guarantee call-dropping probability.
• Guarantee signal quality.
Let us consider a few cases that will demonstrate the importance of the CAC.
1. The base station has limited power resource, and the combined power
transmitted to users cannot exceed the allocated power to the base station. Thus
if admitting new users increases the outage probability of existing users, a bad
CAC decision results.
2. Mobile equipment is moving from one cell to another in which all of the
resources have been fully occupied. A handoff operation is not possible for that
28
particular user, thus indicating a bad CAC scheme. Since a call drop will occur
for such user.
Here is the list of some the most prominent RRM (Radio Resource
Management) schemes used for CAC.
• Power Control.
• Base station assignment.
• Channel allocation.
• Bandwidth reservation.
• Scheduling or buffer management.
Instead of going into the precise details of the CAC schemes mentioned in this
section, we are focusing on the main philosophy adopted in the respective CAC
scheme and evaluate if it works well in choice of our objective function.
A few selected CAC policies are presented in the following,
3.5.1 CAC based on resource availability:
Evans and Everitt [16] proposed the following model to measure the available
resources.
jNj
jij=1 i=1
P X > C ≤ ∝ ∑∑ (3.3)
where C is the total available resources, Nj represents total number of users in
class j and Xji respresents the resources required by user i. of class j. This
particular scheme then defines the admission criteria as,
jj j
j=1K N C≤∑ (3.4)
29
Here Kj represents effective bandwidth of users of class j.
Zhao, Shen, and Mark, [17] used packet delay upper bounds for variable bit rate
calls, and jitter for all constant bit rate calls as a policy for CAC.
3.5.2 CAC based on number of users:
Zu and Hu [18] proposed the following expression to calculate the equivalent
number of users in class i. with respect to UMTS class 1 (the voice service),
jusers_i j i 1 1
r SIRN = N ρ
r SIRj (3.5)
where N represents number of users, ρ as activity factor, r as transmission data
rate and SIR as signal to interference ratio for class j and 1 respectively. In this
particular scheme a Guard band is reserved for handoff calls. Reference [19]
also proposes CAC criteria based on the number of users.
3.5.3 CAC based on SIR:
Ishikawa [19] makes CAC decision based on measured interference level,
whereas Liu [20] uses residual capacity for CAC decision, and it is defined as:
residual1 1 = - Capacity SIR SIRTH
(3.6)
30
3.5.4 Transmitted and Received Power based CAC:
Knutsson [21] proposed a scheme in which a new call is admitted as long as the
total transmitted power from base station do not exceed a maximum threshold
value. On the other hand Kuri [22] based the CAC decision on the 95% of total
power received at the base station.
3.5.5 Reservation Schemes (RS)
This scheme makes CAC decision by looking only on the cell from where the
call has been originated. It is one of the very simple schemes. Let N and Nh are the
total number of channels and number of channels reserved for handoff operations in a
cell, respectively. New calls are admitted if following condition is satisfied,
N - Nh > No (3.7)
where No represent the threshold value.
3.5.6 Linear Weighting Scheme (LWS)
This scheme considers average number of calls within D hops from the
originating cell, before making a CAC decision. Let S represents the set of all cells
within the region of awareness (D-hops), consider N - Nh as the threshold value. The
new calls are admitted to the originating cell only if:
ii S
1 N < (N - N )|S| ε
∑ h (3.8)
For D=0, this scheme reduces to Reservation Scheme.
31
3.5.7 Distributed Admission Control Scheme (DACS)
This scheme considers the number of active calls in the originating and
neighboring cells before making CAC decision. This scheme is considered both for
one and two dimensional models.
Interactive CAC [27] has the philosophy of admitting new users requiring less
power resources after evaluating expressions to confirm if the required SIR level can
be achieved for each user in the cell. This scheme has a few serious drawbacks. It
works well for always active connections, but cannot handle discontinuous
transmissions, a very important aspect while dealing with UMTS network. Secondly
they demonstrate convergence speed problems.
A different approach is to increase power levels to users when interference
increases in order to keep SIR at a target value. Thus this scheme admits new call(s)
only if estimated power increase keep the SIR within the target value.
3.5.8 Shadow Cluster
A mobile terminal exerts influence on the base station in the current cell
depending on its current location, data rate requirements and direction of travel. Such
influence also moves when mobile terminal switches cells. Thus neighboring base
stations that are influenced due to such movements form an area what is called as the
Shadow Cluster. This shadow is strongest at the current location of mobile terminal
and it fades away based on factors like, such as call duration, data rate requirements,
mobile terminal velocity and trajectory.
Due to above mentioned factors, the center of this shadow cluster is not the
center of a geometric area, but it is the current location of the mobile terminal. In
shadow cluster CAC scheme a base station share probabilistic information among
neighboring base stations with a future probability of its active mobile terminal(s)
moving to the neighboring cells. This allow neighboring base stations to reserve
resources for that particular mobile terminal(s) in order to meet a certain call dropping
probability threshold in choice of QoS and objective function in our case.
32
Accepting a new call in a particular cell, also involves neighboring base stations
in the decision making process. After information exchange about current and future
requirements for that particular user in current and neighboring cells, a collective
decision takes place whether to accept or reject the admission request. This scheme
has a distributed framework and works very well in mini or micro-cellular
environment. It does have some processing overheads on the base station and the
wire-line network.
3.5.9 Call Bounding
The main philosophy behind this CAC scheme is to accept a few users rather
than increasing the dropping probability of current users by admitting new users up to
a possible maximum. The concept works well because users are more sensitive to call
dropping than call blocking.
In this scheme, each cell has been assigned a threshold value on the maximum
number of active calls/users/mobile terminals. A new call is blocked only if it
exceeding the set threshold value. On the other hand the handoff calls are blocked
only if all channels are occupied.
3.5.10 Cutoff Priority
Instead of putting a limit on the total number of possible calls in a cell, the CAC
(in this scheme) decision is based on the total number of active calls at that particular
instance in the cell. A new call is accepted if it does not exceed a given threshold
value, otherwise it will be blocked. The handoff calls are always accepted as long as
there is an available channel. This scheme works under the assumption that average
new call channel holding time and average handoff call channel holding time are
equal, allowing the use of one-dimensional Markov chain theory. If there assumptions
are not true this scheme will not work.
Hybrid cutoff priority [28] is an extension of Cutoff Priority scheme from a
single traffic stream to multi-class traffic streams. This scheme supports any number
of traffic classes each with unique QoS Requirements in terms of channels
33
requirements, connection time and a unique Cutoff Threshold value. This scheme
employs finite buffering for handoff and new calls. In addition, the scheme assumes
new calls departing due to unwillingness to wait in the queue, and dropping the
queued handoff calls due to channels unavailability.
3.5.11 Call Thinning
The main idea behind this CAC scheme is to accept new calls with a certain
probability and at the time of congestion the newly admitted calls take up the share to
reduce currently served data rates or to delay in order to reduce load.
Many CAC schemes have been proposed in literature, which are more or less
variables of above mentioned schemes. Our system model, initially only work for a
single cell, so adopting a download power based CAC scheme in choice of the
objective function will be of benefit to us. The performance results of adopted CAC
scheme for a single and neighboring cell environment will set the direction of
selecting variables of different adequate CAC scheme in choice of objective function.
At the very low level its all comes down to power utilization per mobile terminal,
thus any steps taken to make an efficient use of this resource will improve the overall
system performance.
3.5.12 Miscellaneous CAC schemes:
Baldo [23] calculates overload probability (Po) to make CAC decision whereas
Baldo [24] proposed the handoff probability (Phf) for CAC decision. Peha [24]
and Guo [25] proposed to make CAC decision on bandwidth utility function and
degraded service area respectively.
34
3.6. Summary
This chapter focuses on the problem of transmitting video in a mobile wireless
environment. We started the chapter by giving the bottlenecks of transmitting a digital
video without compression. It showed that no compression of video signal requires
very high data rates and hence power allocations for a mobile user. Thus showing the
importance of finding ways to reduce data by compression. We reviewed several
video encoding schemes and then advantages. We also presented an overview of (in
this chapter) packet scheduling and call admission control schemes. We have also
reviewed requirements of providing video streaming service in mobile wireless
environment.
35
Chapter 4
Video Streaming Service in 3G
W-CDMA Network
In previous chapters we have given brief overview on relevant technologies,
algorithms and architectures, which are of importance in our project work. We have
presented (in chapter 2) introduction on the development of personnel computers,
wire-line networks and mobile wireless networks. Then we have presented an
overview on UTRA (3G) mobile wireless networks architecture, and the importance
of power control in 3G networks. We have presented (in chapter 3) an overview on
video transmission requirements and the importance of video compression in
reducing data rate requirements. Then we have presented an overview on several
known video compression technologies.
Our study focuses on the design of an Objective Function that provides cost
functions as an output for the service provider. These functions will be helpful for the
service provider to make critical decisions on CAC, scheduling and power/rate
control based on users service contract, in order to keep the system running in a
profitable and cost effective manner. The objective function will cater Video
Streaming service only in a single cell and later multiple services for UMTS QoS
classes for mobile users in 3G W-CDMA multiple cellular environment.
4.1. Problem Definition
Ideally we would like our objective function working for a real world 3G
mobile wireless networks. Such a task is very wide and can be divided into four
possible cases, as mentioned below.
36
Case #1: Mobile users with varying mobility are requesting only video
streaming service of fixed duration (1 to 5 minutes) in a single 3G
wireless cell.
Case #2: Mobile users with varying mobility are requesting services belonging
to all or a set of UMTS QoS classes in a single 3G wireless cell
environment.
Case #3: Case #1 implementation for a multi-cell environment.
Case #4: Case #2 implementation for a multi-cell environment.
The above case objectives will be taken care of in choice of an Objective
Function. The function will try not only to assure a good QoS but also to achieve a
good profit margin for the service provider. The above two objectives can only be
successfully achieved by achieving efficiency and trade off among the following
issues:
• Packets Scheduling
• Downlink (DL) Power Control.
• Data Rate (R) Control.
• Call Admission Control (CAC).
In this report our main objective is to implement and achieve performance
results for Case #1. The design of our project work will be such that it will pave the
way for future implementations of other cases.
4.2. System Model The overall system model is shown in Fig. 4.1, where Operations A, B and C
are explained as follows:
Operation A: Users requesting a streaming video (for example, video streaming
request for 3-min soccer world-cup highlights).
37
Operation B: Video stream data files reaching the gateway in packetized form. The
stream file is of known volume and data rate requirements. The packets are of varying
size and inter-arrival time.
Operation C: Gateway performing following tasks: Packet Scheduling, Power
Control, Rate Control and CAC.
Figure 4.1: System model considered in the design of our objective function. The users who are requesting video streaming service in a “3G cell”.only
Internet Server Mobile Users
3G Cell
Operation A
Operation B Operation C
Gateway
Figure 4.1: System model considered in the design of our objective function. The users who are requesting video streaming service in a “3G cell”.only
Internet Server Mobile Users
3G Cell
Operation A
Operation B Operation C
Gateway
The following assumptions are made:
• The video streaming service is limited within a single cell.
• The cell has a diameter of d-meters.
• Data rate requirement varies between b1 to b2 bps.
• Users mobility in the range [1,5] meters per seconds.
• Eight allowable moving directions (N, S, E, W, NE, NW, SE, SW).
• IP packets of varying size (s1 to s2) Kbps.
38
The gateway receives the incoming data and after performing operation C the
packets get transmitted to their respective destinations.
We have identified the mobile users as belonging to one of the following two
categories,
1. Service_Level-A: Users requiring guaranteed service, no disconnects or degraded
service, and have willingness to pay high for such a service.
2. Service_Level-B: Users having willingness to pay less at the cost of degraded or
disconnected service.
So, we have to give priority and special consideration to Service_Level_A users
while performing Scheduling or Call Admission Control operations. In order to
simplify our task further, we can treat all users as starting at same service level, and
users can specify one (or more) video resolution(s) that user is willing to pay for. The
higher the resolution indicates higher volume. The CAC can make decisions on which
resolution to serve based on cost function and available power resources.
4.3. Objective Function
The two important factors in the design of objective function is the cost
effectiveness at a given time instance from the system and the effective systems
throughput. The two cases will provide a trade off among service providers profit
margin and user satisfaction. We will be mainly focusing on the later issue in our
project work.
We define the effective systems throughput as,
T (in bits)dZ = T (in seconds)sim
(4.1)
where Td is the total volume delivered and Tsim is the simulation or elapsed time.
39
If there are no packet delay violations then we can calculate the Total Delivered
Traffic after each completion as,
Td = Td + Vi (4.2)
Where Vi represents the traffic volume of a video clip.
If there are delay violations above a Threshold Value, then we proceed as follows,
If Da > ηAD
Then abort or degrade service for Service_Level_B users with certain
probability. If all users have Service_Level A, then make a decision on
degrading or aborting service for those which overall improves the effective
system performance, and keep most of the users satisfied. Also penalize the
systems Total_ Delivered_Traffic to Reflect Interrupted Flows.
where, Da is the average inter-packet delay and ηAD is the threshold value on this
delay.
For the above mentioned cases, we are also assuming a maximum inter-packet
acceptable delay, and service degradation means one of the following cases,
i) Switching from a high data rate to a low data rate if possible.
ii) If case-1 is not possible then delay the packets.
iii) If packets delay exceeds a certain threshold then drop the service for
user(s).
We propose a cost function (Cf) formula to determine the cost effectiveness to
the service provider:
Cf = [FA+FB]-[FC + FD] (4.3)
40
where,
A A
A
N N
Ai RAi RAj ii=1 j=1
F = Cost generated by active Service_Level_A users
= P S Q(k) (T )
∑ ∑ (4.4)
B B
BN N
Bp RBp RB1 pp=1 l=1
F = Cost generated by active Service_Level_B users
= P S Q(k) (T )
∑ ∑ (4.5)
WA WA
CN N
Am RAm RWAo mm=1 o=1
F = Estimated cost generation from waiting Service_Level_A users
= P W Q(k) (T )
∑ ∑ (4.6)
WB WB
DN N
Bn RBn RWBs nn=1 s=1
F = Estimated cost generation from waiting Service_Level_B users
= P W Q(k) (T )
∑ ∑ (4.7)
Here FA is the cost the service providing is getting from Service_Level_A users,
which basically are the one’s requiring guaranteed service and are paying higher for
such a service. Similarly FC is the expected earnings from the waiting
Service_Level_A users, whereas FB and FD are the respective figures from
Service_Level_B users. Our aim is to maximize this function with respect to
increasing workload and QoS requirements. The given Cost_Function formula is a
coarse approach requiring improvements and fine tuning after been put to test for
several different scenarios.
41
4.4 Implementation Let us expand the insight of the key aspects of our system model, and how we
have implemented those.
4.4.1 Algorithm Architecture
We have implemented a flexible architecture, which let us change the
dimension of cell, number of users, packets size and arrival time.
Here is the algorithmic approach we have considered for our project.
i) All channels in good shape and users are requesting only high priority
traffic services (video streaming in our case).
ii) Some channels are in bad shape and users are requesting only high
priority traffic services (video streaming in our case).
iii) All channels are in good shape and users are requesting high and low
priority class services.
iv) Some channels are in bad shape and users are requesting high and low
priority class services.
The first two cases are assumed in this work. We will approach the algorithm
development as the following steps.
Following are the steps taken to accomplish our task.
42
Step #1: The first step is to simulate users profile in a 3G mobile wireless cell. This
will include users mobility pattern and video stream download requests.
Step #2: Packets with varying data rates and inter-arrival time delivered to the
respective serving queues.
Step #3: Simulation Run:
SimulationStartTime=0;
While (Some flow is not completely delivered) {
i) Update users location every t1-seconds:
ii) Call scheduler to determine which users are going to receive data for the next t2 ms:
iii) Wait for t2 ms:
iv) Update the queues:
(1) Update remaining volume to be delivered for each user and average delay per queue
(2) Check if any packets-inter-arrival time delay violations happened: (If true) Give user(s) option to terminate service [with a certain probability]. Also penalize Z to reflect such scenarios.
(3) Check complete volume delivery:
v) Tsim= Tsim + t2 ms:
vi) Compute objective function to see if services are meeting the required profit threshold. If profit margin is below a threshold then call CAC algorithm to decide which user(s) will be dropped or receive degraded service and which new user(s) can be admitted to the system to meet or exceed the profit threshold value.
vii) If system has available resources then perform CAC for new requests.
}
43
Step #4: Calculate and Plot Statistics for Z vs. workload (ρ) and Cf, Z and ρ.
Following are some the remarks taken care of while plotting Effective
Throughput vs Workload graph,
1. Every point is an average of “many” simulation runs.
2. Effective Throughput (Z) gets computed only when the workload (ρ)
is at a required level.
3. Workload is taken with respect to an estimate of overall systems
capacity.
We have used the following expression to calculate workload.
A
F
Rρ = R
(4.7)
where,
RA= Average Requested Data Rate.
RF = Average Reference Rate.
Now in the following bullets we are going to give further details on the insight
of our algorithm writing. First we are going to give information on the formulas we
have used to calculate path loss, required transmission power and SIR. It is then
followed by our approach on packets scheduling. A final decision on a particular
scheduling scheme in choice of our objective function can only be reached after
getting performance results for different schemes in a multi-cellular environment.
44
4.4.2 Calculation of Path Loss and SIR.
We use the log-normal shadow fading model to calculate Path Loss
(equation 4.8). The received power by a mobile terminal at distance d from the
base station is calculated using equation 4.9.
[dB] [dB] oo
d PL(d) = PL (d ) + 10n log ( ) + Xd
s (4.8 )
[dBm] [dBm] [dB]P (d) = P + PL(d)r t ( 4.9)
2
2( )P 4πt = P λr
(4.10)
Where we have considered λ=900 MHz in our project work. Once the
received power is known for each mobile terminal in the cell the next equation
must be satisfied for each mobile terminal in order to make a successful
transmission. The denominator of the equation given below only takes care of
intra_cell interference, since we are only considering a single wireless cell.
E PWb ri= 7 dBI R Prj
i j
≥
≠∑
(4.11)
In the above equation the denominator calculates the sum of power
transmitted to all the mobiles except the one for which SIR is been computed,
and the above equation must be satisfied for each mobile terminal in order to
achieve successful transmissions from base station to the mobile terminal.
45
46
4.4.3 Packets Scheduling
Fig. 4.2 shows our scheduling policy, each user requesting a high or low
priority service creates a service queue for that particular user for system to
manage. FIFO scheduling policy is applied here, where performance
enhancement operations like earliest deadline first or others can also be
applied for increasing effective systems throughput. If system permits then
each queue gets its 10ms frame duration for data delivery, if not then
Service_Level_A users gets priority over others, since they are paying more
for the service. Within Service_Level_A users, those with good channel
conditions get priority over those with bad channel conditions. This particular
issue has also been addressed in objective function formula to calculate the
cost effectiveness for the system, in order to keep satisfied users. Others may
get degraded service if system permits else service disconnects.
Users Packets Arrival (Poisson)
Users With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel Condition
High Priority Packets Queue for User i
Low Priority Packets Queue for User i
High Priority Packets Queue for User i
Low Priority Packets Queue for User i
Figure 4.2: Packets Scheduling Model
Users Packets Arrival (Poisson)
Users With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel Condition
High Priority Packets Queue for User i
Low Priority Packets Queue for User i
High Priority Packets Queue for User i
Low Priority Packets Queue for User i
Users Packets Arrival (Poisson)
Users With Good Channel ConditionUsers With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel ConditionUsers With Bad Channel Condition
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
Figure 4.2: Packets Scheduling Model
Users Packets Arrival (Poisson)
Users With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel Condition
High Priority Packets Queue for User i
Low Priority Packets Queue for User i
High Priority Packets Queue for User i
Low Priority Packets Queue for User i
Users Packets Arrival (Poisson)
Users With Good Channel ConditionUsers With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel ConditionUsers With Bad Good Channel Condition
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
Figure 4.2: Packets Scheduling Model
Users Packets Arrival (Poisson)
Users With Good Channel ConditionUsers With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel ConditionUsers With Bad Good Channel Condition
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
Users Packets Arrival (Poisson)
Users With Good Channel ConditionUsers With Good Channel Condition
Performance Enhancing Operations
Users With Bad Good Channel ConditionUsers With Bad Channel Condition
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
High Priority Packets Queue for User iHigh Priority Packets Queue for User i
Low Priority Packets Queue for User iLow Priority Packets Queue for User i
Figure 4.2: Packets Scheduling Model
4.6 Summary This chapter presents the accomplished project work. It starts by first describing
our project goals. Then we define the problem followed by presenting our system
model. Our approach towards designing an efficient objective function has been
described in section 4.1.1. Section 4.1.2 covers in detail our implementation plan. The
implementation plan has been described in steps.
47
Chapter 5
Conclusions And Future Directions
Our project work of designing an objective function for 3G UMTS/W-CDMA
mobile wireless networks is a very important contribution towards achieving a
balance and tradeoff in QoS/satisfied users and service provider profit margin. The
work needs comprehensive testing in several real world situations in order to fine-
tune the work for better results.
Further studies are required in order to see the performance of several video
codecs in choice of our objective function. An efficient and self-healing video-codec
can reduce data rate and power requirements for the respective user(s), and hence
increase the probability of keeping satisfied users and admitting new ones.
In order to induce efficiency in our objective function it has to get tested for
several appropriate CAC and scheduling schemes on the downlink for services
belonging to all UMTS QoS classes. So, it is a very competitive task and requires lots
of simulations and research. Also in achieving our objective its is imperative to fine
tune the Cost_Function formula so that it can be adopted efficiently and effectively in
a wide range of UMTS wireless network environments.
Our approach is at a very basic level, considering a single UMTS cell for a
single wireless service (Video Streaming), the system has to be tested and further
developed very carefully in steps. So that it will be easy to adopt it for general
purpose use, for all the UMTS QoS classes in a wide range of UMTS wireless
networks.
48
Further research is required to test our approach for several scheduling and call
admission control schemes. So that it can be deduced that which set of scheduling and
call admission control methods can be adopted on permanent basis in choice of our
designed objective function. Also further research is required in studying codecs for
different services, starting with the video streaming service codecs in our case, this
will help reduce data rate requirements for those particular services.
49
References
1. E. S. Elmallah and H. S. Hassanein, “A Power-aware Admission Control
Scheme for Supporting the Assured Forwarding Model in CDMA Cellular
Networks,” Proceedings of IEEE International Conference on Local Computer
Networks (LCN), Tampa, Florida, Nov. 2003
2. L. Wu and E. S. Elmallah, “Supporting Relative Delay Differentiation in
CDMA Cellular Environments,” Proceedings of the International Symposium
on Wireless Systems and Networks (ISWSN’03), Dhahran, Saudi Arabia, March
2003.
3. T. S. Rappaport, Wireless Communications, Prentice Hall Inc., 1996.
4. N. Dimitriou, R. Tafazolli, and G. Sfikas, “Quality of Service for Multimedia
CDMA,” IEEE Communication Magazine, pp 88-94, July 2000.
5. J. S. Blogh and L. Hanzo, Third Generation Systems and Intelligent Wireless
Networking: Smart Antennas and Adaptive Modulation, John Wiley and Sons,
2002.
6. J. Pöllönen, “Quality of Service Based Admission Control for W-CDMA
Mobile Systems,” Master’s Thesis, Nov. 2001, Department of Engineering
Physics and Mathematics, Helsinki University of Technology, Finland.
7. P. Frene, D. Rasseneur, and P. Tournassoud, “Mobile Evolution Towards Full
IP-Multimedia,” Alcatel Telecommunication Review, pp 33-39, 1st Quarter
2001.
8. I. E. G. Richardson, “Introduction to Image and Video Coding”,
http://www.vcodex.fsnet.co.uk/videocoding2b.pdf, (last seen April 2003).
9. 3GPP2 S.R0021, “Multimedia Streaming Services – Stage1,”(version-2),
www.3gpp2.org, (last seen April, 2003).
50
10. K. S. Lee and M. El Zarki,“Comparison of Different Scheduling Algorithms for
Packetized Real-Time Traffic Flows,” International Symposium on Wireless
Personal Multimedia Communications, Denmark, 2001.
11. H., Holma , A., Toskala, W-CDMA for UMTS, Radio Access For Third
Generation Mobile Communications, John Wiley&Sons, 2000, 322 pp.
12. Gilhousen and Jacobs,“On the Capacity of a Cellular CDMA System,” IEEE
Vehicular Technology, Vol. 40 No. 2, May 1991.
13. Sampath, Kumar and Holtzman, “Power Control and Resource Management for
a Multimedia CDMA Wireless System,” IEEE Proc. PIMRC’95, vol. 1, pp 21-
25.
14. Shen, “Resource Management in Wireless CDMA Networks,” Rensselaer
Polytechnic Institute, October 2001.
15. Simon, Omura, Schultz, Levitt, “Spread Spectrum Communications
Handbook,” McGraw Hill Publications, 1994.
16. J. S., Evans and D. Everitt , “Effective Bandwidth Based Admission Control for
Multi-Service CDMA Cellular Networks,” IEEE Transactions on Vehicular
Technology, vol 48, no. 1, pp. 36-46.
17. D. Zhao, X. Shen, and J.W. Mark, “Efficient Call Admission Control for
Heterogeneous Services in Wireless Mobile ATM Networks,” IEEE
Communication Magazine, Volume: 38, Issue. 10, pp. 72-78, Oct. 2000.
18. A. Zhu, J. Ding, and J. Hu, “Adaptive Call Admission Control for Multi-Class
CDMA Cellular Systems,” Fifth Asia Pacific Conference on Communications
and Fourth Opto-Electronics and Communications Conference. APCC/OECC,
pp. 533-6 vol. 1, 1999.
19. Y. Ishikawa, and N. Umeda, “System capacity design based on communication
quality for cellular CDMA systems,” Proc. of Mobile Multimedia
Communications, pp. 43-50, October 1997.
51
20. Z. Liu and M. El. Zarki, “SIR-based call admission control for DS-CDMA
cellular systems,” IEEE Journal on Selected Areas in Communications, vol.12,
no.4, pp.638-44, May 1994.
21. J. Knutsson, P. Butovitsch, M. Persson, R.D. Yates, “Downlink Admission
Control Strategies for CDMA systems in a Manhattan environment,” IEEE 48th
Vehicular Technology Conference (VTC’98), p.1453-7 vol.2.
22. J. Kuri, and P. Mermelstein, “Call admission on the uplink of a CDMA system
based on total received power,” IEEE International conference on
Communications (ICC), pp. 1431-1436 vol.3, 1999.
23. O. Baldo, L. K. Thong, and A. H., Aghvami, “Performance of distributed call
admission control for multimedia high speed wireless/mobile ATM networks,”
IEEE International Conference on Communications (ICC), pp. 1982-1986
vol.3, 1999
24. J. Peha and A. Sutivong, “Admission Control Algorithms for Cellular Systems,”
ACM/Baltzar Wireless Network, Vol.7., No.2, pp.117-125, March 2001.
25. Y. Guo, and B. Aazhong, “Call admission control in multi-class traffic CDMA
cellular system using multi-user antenna array receiver,” IEEE 51st Vehicular
Technology Conference Proceedings, VTC2000-spring, pp.365-369, vol.1,
2000.
26. Y. Xiao, C. L. P. Chen, and Y. Wang, “Quality of service and call admission
control for adaptive multimedia services in wireless/mobile networks,”
Proceedings of the IEEE National Aerospace and Electronics Conference,
2000, NAECON2000, pp.214-220
27. A. Capone, and S. Redana, “Call Admission Control Techniques for UMTS,”
Vehicular Technology Conference, 2001. VTC 2001 Fall. IEEE VTS 54th,
Volume 2, page(s) 925-929 vol.2, 2001.
52
28. B. Li, C. Lin and S. T. Chanson, “Analysis of a hybrid cutoff priority scheme
for multiple classes of traffic in multimedia wireless networks,” Wireless
Networks, Vol. 4, pages 279-290, 1998.
29. S. Wu, K. Y. M. Wong, and B. Li, “A Dynamic Call Admission Policy With
Precision QoS Guarantee Using Stochastic Control for Mobile Wireless
Networks,” IEEE Transactions on Networking, Vol. 10, No.2, April 2002.
53
Appendix A
Wireless Technology: History of Development
In this section, we have presented a comprehensive review of the wireless
technology. The 5W’s (What, Where, When, Who, Why) and 1H (How) of the basics
of wireless technology are explained at a low level.
We note that the word Wireless simply means “no wires”. Because (in order to
understand the underline physics of the technology), wireless technology uses waves
called electromagnetic (electric & magnetic) waves for transmitting information from
one point to another. It is very important to understand the discovery of magnetism
and then electricity. Since magnetic materials were discovered before the discovery of
electricity we will discuss it first.
A.1 Magnetism The keyword “magnetic” came from “Magnus”, the name of a Sheppard boy in
ancient Greece, who placed the tip of his staff on a rock while treating his herd on
mount Ida. However, he later could not free his staff from the rock. So name came
from rocks found in Asia Minor called “Magnesia”. The earliest references to
magnetic material was found in the Chinese literature; it was referenced as the power
of loadstone (magnetitie) between 3rd century B.C. and 6th century A.D. Common
names used for magnetic stone were ‘tzhu shih’ or ‘loving stone’, mentioning a stone
which attracts objects. The power of magnetic material got explored when it first
mentioned by a Greek philosopher Thales in 6th century B.C. in his literature. The
first major device made using the power of magnetism was compass, a tool of great
importance in the development of our civilization. The first compass was build in 1st
century A.D. in China which later reached Europe in 12th century A.D.
54
The presence of compass in Europe was first referenced by an English monk,
Alexander Neckham. But there is no historical information on how he got his hands
on it. In 13th century philosophers thought that compass needle points to the North
star. However this arouse the problem of declination (Horizontal angle between
magnetic north and geo-north), which was first reported by a buddhist astronomer I-
Hsing in 720 A.D. and later also discovered by Christopher Columbus in European
region during his voyage to the west indies in 1492. Eventually research lead to the
conclusion that a magnet hanging in air always points towards North. In 376 B.C.
Haung Ti, a Chinese general ordered to use compass on ships in order to find ways.
Arabs and European sailors adopted it later.
It was later found that the pointer of a compass always points to geo-north
because of Earth’s magnetic field. Since 30km down the earth due to curie
temperature the rocks will be non-magnetic and answer to this questions was given by
Einstien that the magnetism is produced when current flows around a coiled wire,
which implies that 30km down the earth the fluid plays the role of a rotating current
creating magnetism. For further information, look for details on fluid dynamos or
navier-stokes equation.
In 13th century B.C. Thales, a Greek philosopher and mathematician discovered
the fact of forming a magnet after rubbing the stone amber, but he was not able to
conclude why such a magnet is not able to attract heavy metal objects. William
Glibert, a physician to Queen Elizebeth-I later focused on amber problems, he
suggested that its another force, he named it ‘Electrica’ to refer to all such substances,
this word is Latin equivalent for Amber, ‘Electrum’ derived from the Greek word for
Amer.
So the above paragraphs give the reader enough information about the basics of
why and when’s of magnetism. Magnetism paved the way for electricity or electrical
systems. Now let us see how its existence came in to being.
55
A.2. Electricity In 1660, Otto Von Guericke built a glass ball turned by hand, when the ball gets
rubbed against a cloth creates sparks of electricity, its the first static electricity
generator.
Francis Hauksbec, year 1709 in London discovered that by placing a small
amount of mercury in the Von Guericke vacuumed ball, getting it charged and
placing a hand on it will glow and create enough light to read. St. Elmo’s fire, was a
concept of same kind, the name was given to the appearance of a strange glow around
ships during electrical storms, without having a clue that he has actually created Neon
light.
In 1731, Charles Dafey, a Frenchman discovered that statically charged
materials behave likes magnets, they either attract or repel each other like magnets.
He concluded that the two types of charged materials are either carrying ‘+ve’
electricity or ‘-ve’ electricity. The work further carried by Benjamin Franklin, he has
proved through his experiments that electricity is produced by storms and he flew a
kite to prove it. By flying kite he was able to generate charge from his kite. He flew
the kite not in a lighting storm since around same time a Russian scientist was killed
while holding a metal rod up during a lighting storm.
20th March 1800 came the greatest breakthroughs in electricity, based on a
professional disagreement over the results of an experiment, between Luigi Galvani
and Alessandro Volta. Volta successfully proved that when certain chemicals and
metals gets contact, results in generation of electricity. He proved his findings by
placing several pairs of Zinc and silver discs separated by paper, which is soaked in
salt water. The arrangement resulted in an electrical current, thus Volta produced the
first battery, a historical achievement, which made it possible for research community
to easily do experiments on electricity for further development.
In 1820, Hans Christian Oersted while teaching a class discovered that the
galvanometer (attached to a circuit) needle is moving without even turning on the
circuit, with small effort he came to know that a magnet placed nearby is causing that
effect. Thus he discovered by chance that there is a relationship between electricity
and magnetism. His discovery further influenced the work of Faraday and Henry.
56
A.3. Electromagnetism Micheal Farady in 1831 transferred current from a ring (paper cylinder) coupled
with wire to another of same kind having no physical contact among them, he called
it Induction, thus proving the concept that electricity causes magnetism and vice
versa. In 1832 Morse got the help from Henry to develop the first telegraph machine,
it got patented in 1838. Faraday also demonstrated his idea of achieving wireless
transmission via conduction, using water as the medium.
The development was carried on and Marconi built the first wireless receiver
and received letter ‘s’ as the first successful wireless transmission across Atlantic.
Reginald Fessenden created radio wave band for human speech.
In order to understand the above concept, understanding electric field is
important, its defined as the force per unit charge and written as,
FE = q
r
(A.1)
The Fig. A.1 shows is a static electric field created by a static charged body, it
has been shown that moving negative charged particles results in a moving electric
field, which result in having different force on the charged particle along the line of
force. Thus a sinusoidal current will result in an electric field of sinusoidal form, Fig.
A.2.
Force on charged particle (F)
Charged Particle
Charged body creating electric field (E)
Force on charged particle (F)
Charged Particle
Charged body creating electric field (E)
Figure A.1: Field created by a charged body.
Force on charged particle (F)
Charged Particle
Charged body creating electric field (E)
Force on charged particle (F)
Charged Particle
Charged body creating electric field (E)
Figure A.1: Field created by a charged body.
57
Figure A.2: Changing electric field due to moving negative charged particlesFigure A.2: Changing electric field due to moving negative charged particles
Rays
Rays
Rays
Rays
Rays
Rays
1) Moving electron away from proton
2) Moving electron very near to proton
3) Electric field direction change
4) Moving electron very near to proton
Figure A.3: Formation cause of electromagnetic waves.
Rays
Rays
Rays
Rays
Rays
Rays
1) Moving electron away from proton
2) Moving electron very near to proton
3) Electric field direction change
4) Moving electron very near to proton
Figure A.3: Formation cause of electromagnetic waves.
A moving electric field also has an associated magnetic field, thus the waves
created due to the movement of negative charged particles are called electromagnetic
waves. In order to clearly understand this concept, let us consider the example of an
antenna radiating electromagnetic waves. In a wireless antenna electrons moves
against a stationary proton. When the moving electron is very close to proton, the
electric loop closes on itself and the electromagnetic field radiates away (See Fig.
A.3).
Since light is also an electromagnetic energy, it diverges as it moves away from
the source. These wireless waves induce current in a receiving antenna and hence the
information is received in wireless form. All the above information presented is
enough in my opinion to give essential technical information on the word “wireless”.
58
Appendix B
Overview of W-CDMA
Protocols developed for multiple access schemes can be classified into two
categories: scheduling (contention less) and random access (contention). The
scheduling protocols avoid two users accessing the channel at the same time. This is
achieved in one of two ways, either assigning a fixed channel access time unit for
each user or channel access time assignment based on demand of service. The FDMA
and TDMA use fixed channel access time whereas the GSM uses the combination of
the two schemes. In FDMA, the system capacity is divided into frequency channels
which are assigned to users, whereas in TDMA users are allocated time slots to access
the medium.
When using random access protocols, the users are not sure if the transmission
will be collision free (where a collision means failed communication). Protocols like
ALOHA waits for a random amount of time before re-transmitting the previously
collided message. The CDMA technology is different from the traditional schemes. It
is primarily a code based scheme where a unique code is assigned to each mobile
user. The CDMA (scheduling protocol) is based on Spread Spectrum technology,
which makes it possible for all the users to transmit successfully at the same time.
The problem arises only when SNR for a particular user exceed a certain threshold at
the RAKE receiver. The principles behind these technologies can be shown as in Fig.
A.4.
59
Frequency FrequencyFrequency
Time
Time
Time Code
Figure A.4. Multiple Access Schemes
Frequency FrequencyFrequency
Time
Time
Time Code
Figure A.4. Multiple Access Schemes
Original User Signal (single bit)
Spreading Code (N bits or chips)
Spread User Signal (N bits or chips)Original User Signal (single bit)
Spreading Code (N bits or chips)
Spread User Signal (N bits or chips)
Figure A.5: Spread Spectrum Technology: Signal Spreading
60
0 5 10 15 20 25 30 35 40 4530
35
40
45
50
55
60
65
Frequency
PSD
Figure A.6: Signal PSD before spreading
0 5 10 15 20 25 30 35 40 45-300
-250
-200
-150
-100
-50
0
50
100
Frequency
PSD
Figure A.7: Signal PSD after spreading
61
The principal behind the spread spectrum technology is to expand a narrow
band signal (user signal) to a much wider bandwidth using a particular spread code.
Fig. A.5 illustrates this, where a single user bit is expanded into N bits (also called
chips). In Fig. A.6 and A.7 we can see what happens before and after spreading a
single user bit.
Principles of Spread Spectrum Technology
1. The spread spectrum technology researched and developed by Hedy Lamarr (an
actress) and George Antheil (an engineer). The main focus at the time of
development is to have an anti jamming, signal transmission technology. Thus the
user signal is spread over a large spectrum using unique code making it very
difficult to jam the signal, since one has to know the code and the spread
bandwidth range and transmit jamming signal at higher power in all or partial
spectrum range. Fig. A.8 shows what we mean.
TRANSMITTER RECEIVER
JAMMER WFrequency
Power
TRANSMITTER RECEIVER
JAMMER WFrequency
Power
WFrequency
Power
Figure A.8. Jamming Operation in Spread Spectrum Technology
TRANSMITTER RECEIVER
JAMMER WFrequency
Power
WFrequency
Power
TRANSMITTER RECEIVER
JAMMER WFrequency
Power
WFrequency
Power
Figure A.8. Jamming Operation in Spread Spectrum Technology
Here we are going to get a brief overview on the conventional jamming
technique. In Fig. A.8 transmitter is transmitting signal (bandwidth W) at a
certain power (S) and bit rate (Rb), with Eb as energy per bit we can write
transmitted signal power as,
S= EbRb (B.1)
62
The jammer can generate interference either for partial band or broadband
(W). For generating white noise, W equals 1000GHz. For jamming signal, the
jammer power required can be written as,
J=NjW (B.2)
where Nj represents number of signal coordinates.
j
2WT (Coherent Signal)N =
WT (Non-Coherent Signal){ (B.3)
where T is the time required to send a basic symbol.
At the receiver the jammer to signal power ratio (J/S) can be written as,
j
b b
N WJ = S E R
which can be rewritten as
b
j
E PG = JN ( )S
(B.4)
PG (W/Rb) stands for performance gain and is a very important factor in
successful signal reception. We got a basic understanding that how a signal can
be jammed. So all required is enough signal strength at Nj coordinates
(frequencies) or its subset.
Avoiding signal jamming:
Fig. A.9 shows how this technology works, a signal d(t) first gets multiplied by a
channelization code (so different channels possible by only assigning unique codes),
which transforms single bit to signal x1(t) of 64-bits (W-CDMA) which means a bit
rate of 4.02 Mcps (chips per second), where as in CDMA its 3 Mcps, after that the
signal gets multiplied by a unique base station code transforming it to x2(t), which
gets transmitted. On the receiving side, the same base station and channelization
codes are used to decode the desired signal d(t). Spread spectrum technology can help
63
avoid signal jamming. The Fig. A.10 shows the performance gain (PG) of a signal at
the receiver side using spread spectrum technology. The PG of an undesired signal or
interference appears far less than the desired one, thus making it very easy to detect
the appropriate signal. This assures that without knowing the respective code used to
encode the signal, jamming is not possible. Spread spectrum technology uses
Orthogonal Codes.
SC / SF / Channelization Code: c(t) Scrambling / Base Station ID: s(t)
d(t)
d(t)x2 (t) x 1(t) Correlation Detection / Integration
s(t) c(t)
Figure A.9: Spread Spectrum Technology
64-bits (W=4.02 Mcps): x1 (t) 64-bits (W=4.02 Mcps): x2 (t)
SC / SF / Channelization Code: c(t) Scrambling / Base Station ID: s(t)
d(t)
d(t)x2 (t) x 1(t) Correlation Detection / Integration
s(t) c(t)
: Spread Spectrum Technology
64-bits (W=4.02 Mcps): x1 (t) 64-bits (W=4.02 Mcps): x2 (t)
SC / SF / Channelization Code: c(t) Scrambling / Base Station ID: s(t)
d(t)
d(t)x2 (t) x 1(t) Correlation Detection / Integration
s(t) c(t)
Figure A.9: Spread Spectrum Technology
64-bits (W=4.02 Mcps): x1 (t) 64-bits (W=4.02 Mcps): x2 (t)
SC / SF / Channelization Code: c(t) Scrambling / Base Station ID: s(t)
d(t)
d(t)x2 (t) x 1(t) Correlation Detection / Integration
s(t) c(t)
: Spread Spectrum Technology
64-bits (W=4.02 Mcps): x1 (t) 64-bits (W=4.02 Mcps): x2 (t)
+N (PG)
-NDesired Signal
Undesired Signal
Figure A.10: Performance Gain using Spread Spectrum10: Performance Gain using Spread Spectrum
+N (PG)
-NDesired Signal
Undesired Signal
Figure A.10: Performance Gain using Spread Spectrum10: Performance Gain using Spread Spectrum
64
Advantages:
Spread spectrum technology makes it possible to use single frequency band for all the
users. Thus a great solution to the problem of assigning a unique frequency for each
subscriber. Every subscriber has been assigned a unique code, which helps decode the
single for that subscriber. Some of the highlights on using Spread Spectrum
technology (SST) are,
o Resists intentional or non-intentional interference.
o Has the ability to eliminate the effect of multi-path interference.
o Has the ability to share the same frequency band with other users.
o Channels and base stations privacy due to PN codes.
o Parallel signals transmission due to OVSF codes.
The only major disadvantage of the SST is that the implementation is complex
(e.g. Synchronization).
W-CDMA radio access supports both FDD and TDD modes of operation.
Uplink (UL) and downlink (DL) channels are operated on different frequencies
having a certain guard band in FDD mode of operation, whereas in TDD mode same
frequency is used for both UL and DL but are operated in different time slots with a
certain guard period. Having discussed the overall working of W-CDMA system, now
let us talk about its physical architectural layout.
The W-CDMA system is composed of following channels. We will briefly get
an overview on the inner working architecture on these channels.
65
B.1 Transport Channels
Transport channels can be classified into two groups, Dedicated Transport
Channel (DCH) and Broadcast Channel (BCH). The DCH are used to carry
information between Mobile Station (MS) and UTRA networks and hence DCH are
bi-directional channels. On the other hand the BCH is used to transmit system and
cell specific information to all the mobile users in the cell. For example a mobile
stations initial transmit power on the Up Link (UL), cell specific scrambling code.
The other main transport channels are FACH (Forward Access Channel), PCH
(Paging Channel), RACH (Random Access Channel), CPCH (Common Packet
Channel) and DSCH (Downlink Shared Channel). Let us have a brief overview on the
working on these channels.
• FACH: It is a channel for downlink, and is used to carry smaller user data and
control information from BS to MT, provided that system knows the serving BS
of MT.
• PCH: Used on downlink to page MT in order to indicate a call for subscriber. This
channel is used when there is no information about the serving BS of a MT.
• RACH: Its an uplink channel and is used to carry small user data and control
information to the serving BS.
• CPCH: An uplink channel used for transmitting bursty data.
• DSCH: It is a DL channel shared by several users.
B.2 Physical Channels
UTRA physical channels in TDD mode of operation are composed of spreading
code, frequency and time slots. Fifteen time slots of two third milliseconds (2/3 ms)
duration makes one radio frame, which means a radio frame of 10 ms (2/3 x 15=10ms)
duration. Transmitting information using these frames differ not only for UL and DL,
but also if it is an FDD or TDD mode of operation. Fortunately the 10 ms radio frame
66
duration coincides with ITU’s G729 speech codec, it is a submultiple of several
frames in GSM system, possible convenient arrangement of H.263 videophone codec
and ease in varying SF (spreading factor). The Fig. A.11 give an overview on the
UTRA physical channels structure.
Figure A.11. UTRA Physical Channels
Radio Frame 1
Radio Frame 2
Radio Frame 3
Radio Frame 4
10 ms
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
2 /3 ms
Radio Frame 1
Radio Frame 2
Radio Frame 3
Radio Frame 4
10 ms
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
2 /3 ms
Figure A.11. UTRA Physical Channels
Radio Frame 1
Radio Frame 2
Radio Frame 3
Radio Frame 4
10 ms
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
2 /3 ms
Radio Frame 1
Radio Frame 2
Radio Frame 3
Radio Frame 4
10 ms
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
Time Slot 1
Time Slot 2
Time Slot 3
Time Slot 4
Time Slot 15
2 /3 ms
In UTRA FDD mode, spreading code and frequency constitutes a DL physical
channel, whereas in UL the modems I(in-phase) and Q(quadrature-phase) branches
delivers data and control information in parallel. UTRA physical channels are further
classified as dedicated and common channels. Let us get a brief overview on these.
B.2.1 Dedicated Physical Channel
The Fig. A.12(a, b) shows the structure of UTRA FDD UL and DL dedicated
physical channels. DPDCH and DPCCH are used to transmit DCH information
between MT and BS, and Layer-1 information respectively. Layer-1 information
includes TPC (Transmit Power Control), pilot bits, FBI (Feed Back Information) and
optional TFCI (Transport Format Combination Indicator). Both DPDCH and DPCCH
are bi-directional, and employ time multiplexing to form a single DPCH (Dedicated
Physical Channel) on the DL, whereas they are transmitted in parallel using I and Q
branches of modem on the UL.
The parallel transmission avoid EMC (Electromagnetic Compatibility)
problems arising due to DTX (Discontinuous Transmissions: occurring when no data
to transmit) of the DPDCH, creating high energy spikes which may effect user
equipment.
67
DATAUL DPDCH
PilotUL DPCCH TFCI FBI TPC
S field D field
2/3 ms, 10 x 2k bits (where k=0…6)
2/3 ms, 10 bits
(a) FDD mode uplink physical channel
Data1 TPC TFCI Data2 PilotDL DPCH2/3 ms, 10 x 2k bits (where k=0…7)
DPDCH DPCCH DPDCH DPCCH
(b) FDD mode downlink physical channel
DATAData
2/3 ms, 10 x 2k bits (where k=0…3)
PilotControl
2/3 ms, 10 bits
TFCI
Figure A.12. Structure of UTRA Physical Channels
(c) User message transmission, time slot configuration
DATAData
2/3 ms, 10 x 2k bits (where k=0…3)
PilotControl
2/3 ms, 10 bits
TFCI
Figure A.12. Structure of UTRA Physical Channels
(c) User message transmission, time slot configuration
68
The pilot bits are used for coherent detection both on UL and DL and are also
used for several performance enhancement techniques such as interference
cancellation and adaptive antennas.
The pilot bits are also used for frame synchronization between BS and MT,
since their sequence is known. TPC commands supports efficient power control to
address Near Far Problem [3], i.e. less power for MT which is near to BS and more
for those away from BS, this is required for successful mobile telecommunication
system operation in a CDMA environment. TFCI carries information about transport
channels, multiplexed on UTRA physical channels in a radio frame. FBI is composed
of two fields, the S and D field respectively, supports several transmission techniques.
S field is used for site selection diversity, which facilitates soft handover operation by
reducing interference due to multiple transmissions. D field is used for phase and
attenuation information in order to support closed loop transmit diversity.
Further details on channels used in FDD and TDD mode of operation in UTRA
can be found in [1].
69
Appendix C This project code simulates the following aspects of our projects work,
1) It simulates users mobility in a single 3G cell. Users mobility in one of eight directions, which changes every t-seconds.
2) Simulates users video streaming requests from an Internet server.
3) Video stream requests getting acknowledged by Internet server and sent in packets form of variable size and inter-arrival times.
4) Packets are queued using FIFO and the destination channel condition.
Project Code %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%Users Mobility Directions, Speed and Traffic Ranges%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% clear all; %Clear all variables, etc. MinimumDirections=1; %Directions MaximumDirections=8; %Directions MinimumSpeed=1; %Meters/Seconds MaximumSpeed=5; %Meters/Seconds MinimumVideoTraffic=100; %Kilo Bytes MaximumVideoTraffic=500; %Kilo Bytes MinimumLowPriorityTraffic=10; %Kilo Bytes; LowPriorityTraffic=>Web Pages,Email MaximumLowPriorityTraffic=90; %Kilo Bytes; LowPriorityTraffic=>Web Pages,Email MaximumPacketSize=3072; %Bytes MinimumPacketSize=1024; %Bytes MaximumArrivalTime=1000; %Milli Seconds MinimumArrivalTime=150; %Milli Seconds %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%Get Number of Users in a Cell%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% prompt={'Enter Cell Radius','Distance Between Two Rings'}; def={'1500','50'}; dlgTitle='Cell Radius'; lineNo=1; Cell_Radius=inputdlg(prompt,dlgTitle,lineNo,def); drawnow
70
%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%Define Cell Area/Axis%%% %%%%%%%%%%%%%%%%%%%%%%%%%%% Bottom_X1=int16(-1*str2double(Cell_Radius(1))); Bottom_Y1=int16(str2double(Cell_Radius(1))); Width_X1 =int16(-1*str2double(Cell_Radius(1))); Height_Y1=int16(str2double(Cell_Radius(1))); AXIS([Bottom_X1 Bottom_Y1 Width_X1 Height_Y1]); h1=line(0,0,'color','w','LineWidth',.1,'MarkerFaceColor','b','Marker','O'); %StopHandle=uicontrol('Style','pushbutton','String','Stop','Units','Normalized','Position',[20 20 60 20]); %StopHandle=uicontrol('Style','pushbutton','String','Stop','Callback','break'); %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%Get Input for Mobile Users%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% [X,Y]=ginput; h=line(X,Y,'color','w','LineWidth',.1,'MarkerFaceColor','r','Marker','o'); xlabel('Distance in meters'); ylabel('Distance in meters'); Title_Text='Total Mobile Users in the Cell\rightarrow'; title(Title_Text); set(gcf,'doublebuffer','on'); text(str2double(Cell_Radius(1)),str2double(Cell_Radius(1)),int2str(length(X))); %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%Define User Class/Structure%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Randomizer=0; for i=1:length(X) Cell_User(i).PositionX=X(i); Cell_User(i).PositionY=Y(i); Cell_User(i).DistanceFromBasestation=sqrt(Cell_User(i).PositionX^2+Cell_User(i).PositionY^2); %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% Select Mobile User Speed%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% RandomSpeedSelect=round(MinimumSpeed+(MaximumSpeed-MinimumSpeed)*rand); Randomizer=Randomizer+1; if(Randomizer>8) Randomizer=1; end switch(Randomizer) case 1 Cell_User(i).Speed=RandomSpeedSelect; case 2 Cell_User(i).Speed=RandomSpeedSelect; case 3 Cell_User(i).Speed=RandomSpeedSelect; case 4 Cell_User(i).Speed=RandomSpeedSelect; case 5 Cell_User(i).Speed=RandomSpeedSelect; case 6 Cell_User(i).Speed=RandomSpeedSelect; case 7 Cell_User(i).Speed=RandomSpeedSelect; case 8 Cell_User(i).Speed=RandomSpeedSelect; end
71
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%% Select Mobile User Direction of Travel. %%%%%%%%%%%%%%%%% %% 1=>East, 2=>West, 3=>North, 4=>South, 5=>NE, 6=>NW, 7=>SE, 8=>SW%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); switch(Randomizer) case 1 Cell_User(i).Direction=RandomDirectionSelect; case 2 Cell_User(i).Direction=RandomDirectionSelect; case 3 Cell_User(i).Direction=RandomDirectionSelect; case 4 Cell_User(i).Direction=RandomDirectionSelect; case 5 Cell_User(i).Direction=RandomDirectionSelect; case 6 Cell_User(i).Direction=RandomDirectionSelect; case 7 Cell_User(i).Direction=RandomDirectionSelect; case 8 Cell_User(i).Direction=RandomDirectionSelect; end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% Users Download Traffic Request%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% RandomHighPriorityTrafficSelect=round(MinimumVideoTraffic+(MaximumVideoTraffic-MinimumVideoTraffic)*rand); RandomLowPriorityTrafficSelect=round(MinimumLowPriorityTraffic+(MaximumLowPriorityTraffic-MinimumLowPriorityTraffic)*rand); Cell_User(i).VideoDownloadRequest=RandomHighPriorityTrafficSelect; Cell_User(i).LowPriorityTrafficRequest=RandomLowPriorityTrafficSelect; %%%%%%%%%%%%%%%%%%%%%%%%% %%% Packetize Traffic %%% %%%%%%%%%%%%%%%%%%%%%%%%% % Conversion in Bytes. HighPriorityDownloadRequest=Cell_User(i).VideoDownloadRequest*1024; LowPriorityDownloadRequest=Cell_User(i).LowPriorityTrafficRequest*1024; j=1; % Counter for user packets. while((HighPriorityDownloadRequest>=1024) | (LowPriorityDownloadRequest>=1024)) % High Priority Packets formation. if(HighPriorityDownloadRequest>=1024) HighPriorityRandomPacketSize=round(MinimumPacketSize+(MaximumPacketSize-MinimumPacketSize)*rand); HighPriorityRandomArrivalTime=round(MinimumArrivalTime+(MaximumArrivalTime-MinimumArrivalTime)*rand); HighPriorityDownloadRequest=HighPriorityDownloadRequest-HighPriorityRandomPacketSize; if(HighPriorityDownloadRequest<1024) HighPriorityDownloadRequest=HighPriorityDownloadRequest+HighPriorityRandomPacketSize; Cell_User(i).HighPriorityPackets(j).Size=HighPriorityDownloadRequest; Cell_User(i).HighPriorityPackets(j).ArrivalTime=HighPriorityRandomArrivalTime; HighPriorityDownloadRequest=HighPriorityDownloadRequest-HighPriorityRandomPacketSize; else Cell_User(i).HighPriorityPackets(j).Size=HighPriorityRandomPacketSize; Cell_User(i).HighPriorityPackets(j).ArrivalTime=HighPriorityRandomArrivalTime; end
72
end % Low Priority Packets formation. if(LowPriorityDownloadRequest>=1024) LowPriorityRandomPacketSize=round(MinimumPacketSize+(MaximumPacketSize-MinimumPacketSize)*rand); LowPriorityRandomArrivalTime=round(MinimumArrivalTime+(MaximumArrivalTime-MinimumArrivalTime)*rand); LowPriorityDownloadRequest=LowPriorityDownloadRequest-LowPriorityRandomPacketSize; if(LowPriorityDownloadRequest<1024) LowPriorityDownloadRequest=LowPriorityDownloadRequest+LowPriorityRandomPacketSize; Cell_User(i).LowPriorityPackets(j).Size=LowPriorityDownloadRequest; Cell_User(i).LowPriorityPackets(j).ArrivalTime=LowPriorityRandomArrivalTime; LowPriorityDownloadRequest=LowPriorityDownloadRequest-LowPriorityRandomPacketSize; else Cell_User(i).LowPriorityPackets(j).Size=LowPriorityRandomPacketSize; Cell_User(i).LowPriorityPackets(j).ArrivalTime=LowPriorityRandomArrivalTime; end end j=j+1; end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% Simulating Users Channel Conditions (1=>Good and 0=>Bad)%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Cell_User(i).ChannelCondition=1; % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% Other Users Traffic Parameters%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Cell_User(i).CurrentRing=0; Cell_User(i).RequestServedSoFar=0; Cell_User(i).RequiredDataRate=0; Cell_User(i).DataRateServed=0; End %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% Simulating Mobile Users Packets Arrival at Basestation%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% for i=1:length(X) Cell_User(i).TotalNoOfHighPriorityPackets=length(Cell_User(i).HighPriorityPackets); Cell_User(i).CopyTotalNoOfHighPriorityPackets=length(Cell_User(i).HighPriorityPackets); Cell_User(i).TotalNoOfLowPriorityPackets=length(Cell_User(i).LowPriorityPackets); Cell_User(i).CopyTotalNoOfLowPriorityPackets=length(Cell_User(i).LowPriorityPackets); Cell_User(i).HighPriorityPacketsArrivalSequence=randperm(Cell_User(i).TotalNoOfHighPriorityPackets); Cell_User(i).TempHighPriorityPackets=Cell_User(i).HighPriorityPacketsArrivalSequence; Cell_User(i).LowPriorityPacketsArrivalSequence=randperm(Cell_User(i).TotalNoOfLowPriorityPackets); Cell_User(i).TempLowPriorityPackets=Cell_User(i).LowPriorityPacketsArrivalSequence; end
73
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% Filling the Queues %%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TotalNoOfUsers=length(X); for i=1:TotalNoOfUsers MobileUsers(i)=i; TempMobileUsers(i)=i; end QueueCounter=1; while(length(TempMobileUsers)>0) SelectIndex=round(1+(length(TempMobileUsers)-1)*rand); % Select Index, to later pick user. RandomUserSelect=MobileUsers(SelectIndex); % Randomly pick a user. if(Cell_User(RandomUserSelect).ChannelCondition==1) % Users Channel Condition GoodChannelsPacketsQueue(QueueCounter).UserID=RandomUserSelect; % Packet Received for User # GoodChannelsPacketsQueue(QueueCounter).PacketsPriority=1; % Packets Priority GoodChannelsPacketsQueue(QueueCounter).PacketsSequenceID=Cell_User(RandomUserSelect).TempHighPriorityPackets((Cell_User(RandomUserSelect).CopyTotalNoOfHighPriorityPackets+1)-Cell_User(RandomUserSelect).TotalNoOfHighPriorityPackets); else BadChannelsPacketsQueue(QueueCounter).UserID=RandomUserSelect; % Packet Received for User # BadChannelsPacketsQueue(QueueCounter).PacketsPriority=1; % Packets Priority BadChannelsPacketsQueue(QueueCounter).PacketsSequenceID=Cell_User(RandomUserSelect).TempHighPriorityPackets(1); end Cell_User(RandomUserSelect).TotalNoOfHighPriorityPackets=Cell_User(RandomUserSelect).TotalNoOfHighPriorityPackets-1; if(Cell_User(RandomUserSelect).TotalNoOfHighPriorityPackets==0) MobileUsers(SelectIndex)=0; TempMobileUsers(1)=0; TempMobileUsers=find(TempMobileUsers); % Eliminate users selection, with complete request submition in queue. for i=SelectIndex:length(MobileUsers)-1 MobileUsers(i)=MobileUsers(i+1); end end QueueCounter=QueueCounter+1; end %%%%%%%%%%%%%%%%%%%%%%%% %% Draw Circles %%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%% alpha=0:360; angle=alpha*pi/180; hold on; Distance_Between_Two_Rings=str2double(Cell_Radius(2)); for i=Distance_Between_Two_Rings:Distance_Between_Two_Rings:round(sqrt(2*str2double(Cell_Radius(1))^2)), X1=i*cos(alpha); Y1=i*sin(alpha); plot(X1,Y1,'.');
74
drawnow; end %%%%%%%%%%%%%%%%% set(h,'XData',X,'YData',Y); %set(h,'XData',Cell_User.PositionX,'YData',Cell_User.PositionY); drawnow; %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% Mobile Users Movement in Cell %%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TotalTime=0; while(TotalTime<30) tic; Read_Timer=toc; while(Read_Timer<2) Read_Timer=toc; end TotalTime=TotalTime+Read_Timer; for i=1:length(X) switch (Cell_User(i).Direction) case 1 X(i)=X(i)+Cell_User(i).Speed; if(X(i)>=str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==1) | (RandomDirectionSelect==5) | (RandomDirectionSelect==7)) X(i)=X(i)-Cell_User(i).Speed; Cell_User(i).Direction=8; else X(i)=X(i)-Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 2 X(i)=X(i)-Cell_User(i).Speed; if(X(i)<=-1*str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); Cell_User(i).Direction=RandomDirectionSelect; if ((RandomDirectionSelect==2) | (RandomDirectionSelect==6) | (RandomDirectionSelect==8)) X(i)=X(i)+Cell_User(i).Speed; Cell_User(i).Direction=7; else X(i)=X(i)+Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 3 Y(i)=Y(i)+Cell_User(i).Speed; if(Y(i)>=str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==3) | (RandomDirectionSelect==5) | (RandomDirectionSelect==6)) Y(i)=Y(i)-Cell_User(i).Speed; Cell_User(i).Direction=8; else Y(i)=Y(i)-Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 4 Y(i)=Y(i)-Cell_User(i).Speed; if(Y(i)<=-1*str2double(Cell_Radius(1)))
75
RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==4) | (RandomDirectionSelect==7) | (RandomDirectionSelect==8)) Y(i)=Y(i)+Cell_User(i).Speed; Cell_User(i).Direction=5; else Y(i)=Y(i)+Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 5 X(i)=X(i)+Cell_User(i).Speed; Y(i)=Y(i)+Cell_User(i).Speed; if(X(i)>=str2double(Cell_Radius(1)) | Y(i)>=str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==5) | (RandomDirectionSelect==3) | (RandomDirectionSelect==1)) X(i)=X(i)-Cell_User(i).Speed; Y(i)=Y(i)-Cell_User(i).Speed; Cell_User(i).Direction=4; else X(i)=X(i)-Cell_User(i).Speed; Y(i)=Y(i)-Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 6 X(i)=X(i)-Cell_User(i).Speed; Y(i)=Y(i)+Cell_User(i).Speed; if(X(i)<=-1*str2double(Cell_Radius(1)) | Y(i)>=str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==6) | (RandomDirectionSelect==2) | (RandomDirectionSelect==8)) X(i)=X(i)+Cell_User(i).Speed; Y(i)=Y(i)-Cell_User(i).Speed; Cell_User(i).Direction=1; else X(i)=X(i)+Cell_User(i).Speed; Y(i)=Y(i)-Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 7 X(i)=X(i)+Cell_User(i).Speed; Y(i)=Y(i)-Cell_User(i).Speed; if(X(i)>=str2double(Cell_Radius(1)) | Y(i)>=-1*str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==7) | (RandomDirectionSelect==4) | (RandomDirectionSelect==8)) X(i)=X(i)-Cell_User(i).Speed; Y(i)=Y(i)+Cell_User(i).Speed; Cell_User(i).Direction=3; else X(i)=X(i)-Cell_User(i).Speed; Y(i)=Y(i)+Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end case 8 X(i)=X(i)-Cell_User(i).Speed; Y(i)=Y(i)-Cell_User(i).Speed; if(X(i)>=-1*str2double(Cell_Radius(1)) | Y(i)>=-1*str2double(Cell_Radius(1))) RandomDirectionSelect=round(MinimumDirections+(MaximumDirections-MinimumDirections)*rand); if ((RandomDirectionSelect==8) | (RandomDirectionSelect==4) | (RandomDirectionSelect==7)) X(i)=X(i)+Cell_User(i).Speed;
76
77
Y(i)=Y(i)+Cell_User(i).Speed; Cell_User(i).Direction=2; else X(i)=X(i)+Cell_User(i).Speed; Y(i)=Y(i)+Cell_User(i).Speed; Cell_User(i).Direction=RandomDirectionSelect; end end end end set(h,'XData',X,'YData',Y); drawnow; end
top related