Enhancing Throughput of Multihop Wireless Networks using Multiple Beam Smart Antennas Vivek Jain (Bachelor of Technology (E&C), Indian Institute of Technology.

Enhancing Throughput of Multihop Wireless Networks using Multiple Beam Smart Antennas

Vivek Jain(Bachelor of Technology (E&C), Indian Institute of Technology at Roorkee, India, 2002)

Thesis Title: On-Demand Medium Access with Heterogeneous Antenna Technologies in Multihop Wireless

NetworksThesis Advisor: Dr. Dharma P. Agrawal

Computer Engineering/VLSI Seminar

PhD CandidateOBR Center for Distributed and Mobile Computing

ECECS Department, University of [email protected]

Outline Introduction

Multihop Wireless Networks Antenna Technologies Medium Access Control Protocols

Analytical Framework IEEE 802.11 DCF based Protocols for MBAA ESIF Mechanism AMD for Beamforming Antennas HMAC for MBAA Summary of the Research Work Future Work

Introduction – Wireless Network

Infrastructure-based – Devices communicate with central Access Point (AP). Also, referred to as Wireless Local Area Network (LAN).

Peer-to-peer – Any two devices can communicate, when in range. Also, referred to as Personal Area Network (PAN) or an Ad hoc Network.

Introduction – Multihop Wireless Network (MWN)

Intermediate nodes act as routers or relay nodes

Multihop forwarding to ensure network connectivity

Extends coverage of single hop wireless networks

Multihop wireless networks offer Scalability Reliability Adaptability Easy deployment in rough terrains

Introduction – Mobile Ad hoc Network (MANET)

Set of mobile station (MSs) Lack of fixed infrastructure

relay nodes Dynamically changing

topology Applications

Military - Combat Systems, reconnaissance, surveillance

Disaster management Medical emergency Virtual navigation Distance education

Introduction – Wireless Mesh Network (WMN)

A combination of infrastructure-based and peer-to-peer networks

Set of mobile and immobile stations

Dynamically changing topology

Applications Intelligent transport systems Public safety Public internet access Residential broadband access Distance education

Introduction – Wireless Sensor Network (WSN)

Usually a set of small immobile nodes referred as motes

Generally static topology Cheap alternative to monitor inaccessible or

inhospitable terrains Applications

Medical Applications – wireless bio-sensors Nuclear and chemical plants Environmental monitoring Ocean monitoring Battlefields

Introduction – Challenges in MWN

Medium access protocols Routing protocols Transport Protocols Cross layer optimization Network capacity utilization Security Network lifetime in WSN Co-existence of several types of MWN

MANET,WMN,

andWSN

Network layer and Medium

access layerSmart Antennas and

MIMO

Introduction – Antennas

Omnidirectional Antenna – Low Throughput in Wireless Ad hoc networks due to poor spatial reuse.

Omnidirectional Communication

A B

C

D

E

F

G

H

Directional Communication

Directional Antenna – Better Spatial reuse. But a node still unable to fully utilize “spatial bandwidth”.

A B

C

DF

G

H

X

Nodes in Silent Zone

E

Introduction – Multiple Beam Smart Antennas

Also referred as Multiple Beam Antenna Array (MBAA) – Exploits spatial bandwidth fully.

A node can initiate more than one simultaneous transmissions (or receptions).

DATA

DATA

DATA

A

B

C

D

E

F

G

DATADATA

DATA

2

34

6

7

8

10

11

12

5

9

1

Adaptive array Switched array

Introduction – Multiple Beamforming Antennas

top view (horizontal)

Interferer 1

User 1

User 3

User 2 Interferer 2

Interferer 3

Applications

Military NetworksCellular Communication NetworksMultihop Wireless Networks

2

34

6

7

8

10

11

12

5

9

1

Switched array

top view (horizontal)

Interferer 1

User 1

User 3

User 2 Interferer 2

Interferer 3

Adaptive array

top view (horizontal)

Interferer 1

User 1

User 3

User 2 Interferer 2

Interferer 3

Introduction – Antenna System Phased Array Antenna

0 1 2 3 4 5 6 7

d

Incident Wave

8 Element Linear Equally Spaced

Antenna Array

0

1

2

3

4

56

7

8 Element Equally Spaced Circular Antenna Array

Greater the number of elements in the array, the larger its directivity

Introduction – Beamforming

… …

Direction of Arrival Estimation Beam Formation

As all antenna elements are used for beamforming, a node can either transmit or receive simultaneously, but

not both.

… …

Introduction – Medium Access

On-Demand or Contention-based

Scheduling or Contention-free

Channel Allocation Dynamic Pre-defined

Topological Change Adaptation

Good New Schedules Required

Time Synchronization No Yes

Energy Utilization Uncontrolled Controlled

Concurrent Receptions or Transmissions

Local Synchronization Required

Inherent

On-demand vs. scheduling medium access control protocols

MBAA Model Assumptions A wide azimuth switched-beam smart

antenna Antenna array has M elements that

forms non-overlapping sectors spanning an angle of 360/M degrees

Beam shape is assumed as conical Benefits of nulling or the impact of side-

lobe interference are not considered Carrier sense is performed directionally A collision occurs only if a node

receives interfering energy in the same beam in which it is actively receiving a packet

Range of omnidirectional and directional beam is the same

The Antenna Model

1

23

4

M

M-1

Directional Coverage Area

Omni-directional Coverage Area

Analytical Framework

Can we develop an analytical framework to: Calculate throughput of on-demand medium

access control protocols? Calculate concurrent packet reception capability

of medium access control protocols? Calculate upper bounds of throughput for the

ideal MAC in a multihop wireless network that can provide as a benchmark to compare with the proposed protocols?

Slotted Aloha

Slotted Aloha throughput with N=50, a=0.01 and p=0.03

CSMA

CSMA throughput with N=50, a=0.01, p=0.03 and f =0.03

Concurrent Packet Reception Bounds

Percentage of CPR for asynchronous on-demand receiver-initiated protocols

Concurrent Packet Reception Bounds

Percentage of CPR for asynchronous on-demand transmitter-initiated protocols

Throughput Bounds – Ideal MAC

sec/_**},...,,max{

1,min_

21

packetsDurationCommhwww

RxPacketsh

Source Destinationw1 w2 wh

otherwise ;

ioncommunicat ldirectiona 2 ;2

ioncommunicat ionalomnidirect 3 ;3

h

h

h

h

where, h is hop-length

Comm_Duration is communication time taken by a packet on each hop

length-hop effective is h

source of rate generationpacket is

Also,

IEEE 802.11 DCF for MBAA

Does the existing IEEE 802.11 DCF based MAC protocols for single beamforming antennas yield optimal results for MBAA also?

If not, then what are the features that are needed in a protocol to leverage the benefits of MBAA?

IEEE 802.11 DCF De-facto medium access control for wireless LAN and ad hoc

networks Originally designed for omnidirectional communication, its

virtual carrier sensing (VCS) mechanism is enhanced for directional communication to include directional of arrival also.

IEEE 802.11 DCF for Multiple Beam Antennas

Random Backoff after DIFS wait

Beam-based Node-based

Transmission Control Packets (RTS/CTS)

Directional Omnidirectional

All nodes employ IEEE 802.11 DCF with directional virtual carrier mechanism (DVCS).

MMAC-NBMDMAC-NBMDMAC-BB MMAC-BB

Performance Evaluation

1

23

4

8

7

Directional Coverage Area

Omnidirectional Coverage Area

5

6

The Antenna Model

Packet generation at each source node is modeled as Poisson process with specified mean arrival rate

Each packet has a fixed size of 2000 bytes and is transmitted at a rate of 2Mbps

Each node has maximum buffer of 30 packets Each packet has a lifetime of 30 packet durations Each simulation is run for 100 seconds.

Gains from spatial reuse only are

considered

Performance Evaluation None of the protocols are able to extract throughput of more

than 33% of the maximum possible value This implies only one route is active on an average and hence

concurrent packet reception is not occurring at node D.

A

B

C

D

E

G

F

ESIF

Can we have an on-demand medium access protocol that can yield nearly optimal results in multihop wireless networks with MBAA?

If yes, then Is the protocol synchronous or asynchronous or

a hybrid of both? Does the protocol support differentiated service

classes?

MAC – IssuesConcurrent Packet Reception with IEEE 802.11 DCF

Conclusion: Eradicate the backoff after DIFS duration

RTS

RTS

RTS

RTS

RTS

RTS

RTS

RTS

A

B

C

D

E

F

G

DATA

DIFS

DIFS

DIFS

CTSACK

RTS

DIFSC

TS

MAC Issues – Backoff Removal

Multiple transmitters, located in the same beam of common receiver, always get the same receiver schedule and thus initiate communication at the same time - collision

A node with very high data generation rate will overwhelm its receiver, without giving latter a chance to forward this traffic - fairness issue

All classes of service get same priority – QoS issue

C

A

B

DIFS

DIFS

XRTS

RTS

BADIFS

RTSCTSDATAACKDIFS

C

Use p-persistent

CSMA

Hold the transmitting node

ESIF – ENAV

Every node maintains an ENAV: The beam a neighbor falls within Neighbor’s schedule - the duration until this

neighbor is engaged in communication elsewhere Whether a neighbor’s schedule requires

maintaining silence in the entire beam Number of data packets outbound for the

neighbor The p-persistent probability to use when talking

to this neighbor

ESIF – Cross Layer Data Management

Using network layer information along with ENAV a node determines:

Whether a beam contains an active route The number of potential transmitters in each beam Until what time the node needs to maintain silence in a

particular beam Each node has a store-and-forward buffer for

relaying data packets Available buffer is used dynamically to form

different queues for each beam - prevents head-of-the-line blocking

ESIF – Design ESIF piggybacks feedback onto control messages; RTS with

Intelligent Feedback (RIF), and CTS with Intelligent Feedback (CIF), Schedule Update with Intelligent Feedback (SCH)

SCH identifier allows a neighbor to adjudge whether to defer transmission for only this node or for the entire beam

buffer-threshold to control priorities between receiver and transmitter modes

Reception gets priority as long as the buffer size remains under the threshold

If a node cannot actually initiate transmitter mode, the receiver still gets the priority

Priority switch solves problems of an overwhelmed receiver. This also provides a mechanism to control the contribution

of a node to end-to-end delays

ESIF – Basic Operation

Performance Evaluation Removal of contention window based backoff in ESIF does

not affect long-term fairness Both the transmitters get equal opportunity to transmit

A B

Performance Evaluation ESIF enhances throughput by the priority switch between

transmission and reception modes ESIF is able to achieve concurrent data communications

between node pairs A-B and C-D

B

A

C

D

Performance Evaluation ESIF is able to achieve CPR at common intermediate node D Dynamic priority switch ensures data packets just received are

transmitted (concurrently) in the next cycle, thus, maximizing throughput and minimizing delay

A

B

C

D

E

G

F

Concurrent Packet Reception Bounds

Percentage of CPR for IEEE 802.11 over four and eight

beam antennas Percentage of CPR for ESIF over four and eight beam

antennas

QoS over ESIF Mechanism

Multilevel Queue Organization in each beam of the sender in SS-

MQO

Multilevel Sender Queue (MSQ) at the receiver in RICS

Performance Evaluation Each node generate class 0, class 1 and class 2 packets with

probabilities 0.2, 0.3 and 0.5, respectively, while they are selected with respective probabilities of 0.5, 0.3 and 0.2

Prioritized flow selection is enforced more strictly in RICS as QoS parameters are applied at two ends – sender and receiver

E

BA

D

C

BeamformingAdvantages Longer Range

Better connectivity and lower end-to-end delay

Spatial Reuse Increased capacity and

throughput

Limitations Deafness and hidden terminal

problems Node is unaware of ongoing

communication in the neighborhood regions where it not currently beam-formed

1

23

4

8

7

Directional Coverage Area

Omnidirectional Coverage Area

5

6

Deafness Problem

Nodes X and Y do not know the busy state of node A and keep transmitting RTSs to A

RTS

RTSB

Y

X

DATAA

Deafness – Consequences

At transmitter Increases retransmission attempts after

doubling contention window for every unsuccessful attempt

At receiver Can increase collisions due to interference

with active RTS or data receptions Overall Network

Reduces throughput and increases end-to-end latency

Deafness – Proposed Solutions (Single Beam Antennas)

Omni-directional transmission of control messages Asymmetry in gain of

directional and omni-directional nodes leads to deafness

Circular sweeping of control messages Increases end-to-end delay

due to sweeping

Deafness – Proposed Solutions (Multiple Beam Antennas)

Proactive approach A node transmits control

messages in all free beams

Reactive approach A node transmits control

messages in all beams that are free and have potential transmitters

Proposed AlgorithmHybrid Approach Uses DVCS mechanism to

dynamically maintain two parameters for every beam

isRTSReceived: Set to true when a node receives a RTS intended for itself

isCTSReceived: Set to true when a node receives a CTS not intended for itself

Transmit control messages in all unblocked beams whose isRTSReceived is set to true

Transmit control messages in all unblocked beams if isCTSReceived is true for the beam engaged in actual data communication

SCH

SCH

SCH

CTS

Performance Evaluation Throughput obtained in MMAC-NB is low due to collisions

occurring at node D from transmissions by nodes A and B The topology has no effect on ESIF as control messages are

sent only in routes with potential transmitters

A

BC

D

HMAC Can we have a medium access control

protocol that can support both omnidirectional and beamforming antennas including MBAA?

If yes, then Is the new protocol backward compatible

with IEEE 802.11 DCF? Does this protocol reach ideal throughput

upper bounds? Can we have cost-effective mesh

network architecture?

HMAC - Features Can we have a medium access control

protocol that can support both omnidirectional and beamforming antennas including MBAA?

If yes, then Is the new protocol backward compatible

with IEEE 802.11 DCF? Does this protocol reach ideal throughput

upper bounds? Can we have cost-effective mesh

network architecture?

HMAC – Enhancements over ESIF Employs Algorithm for Mitigating Deafness

while transmitting control packets Exploits a Run-time Sender Estimation

Algorithm Does not rely on neighbor feedback Maintains same packet formats for control

messages as in IEEE 802.11 DCF and is thus compatible with it

Simpler implementation as compared to ESIF

HMAC – Packet Formats

HMAC/IEEE RTS packet format HMAC SCH packet format for RTS

HMAC/IEEE CTS and ACK packet format HMAC SCH packet format for CTS

HMAC/IEEE Frame control field for control messages (RTS/CTS)

Performance Evaluation With basic access mechanism similar to ESIF, HMAC is able

to deliver optimal performance

A

B

C

D

E

G

F

Performance Evaluation Run-time sender estimation algorithm employed in HMAC

delivers comparable performance to ideal p-persistent mechanism exploited in ESIF

DB

A

E

C

Mesh Network Architecture

Access Point

Wired internet backbone

Wired internet backbone

Wireless Routers

Mobile Users

1

2

5

10

6

8

4

13

149

15 16 1718

19

3

11

7

20

12

Summary – Analytical Framework

Evaluated the performance of slotted Aloha and basic CSMA Slotted Aloha gives better throughput than basic CSMA at

lower loads At heavier loads, CSMA gives better performance

Asynchronous on-demand protocols like CSMA gives stable results for MBAA at higher loads

Throughput in CSMA can be enhanced by using localized synchronization and properly adjusting transmission

probability p of nodes such that Np>2, where N is the number of contending nodes.

However, as number of beams increases, throughput per beam in CSMA falls due to synchronization losses

Developed analytical framework for calculating concurrent packet reception capability of asynchronous protocols

Calculated throughput upper bounds for an ideal MAC

Summary – IEEE 802.11 DCF Based MAC Protocols for MBAA

Concurrent packet reception in multiple beam antennas is highly improbable with IEEE 802.11 DCF based protocols

Asynchronous protocols thus cannot leverage the benefits of MBAA

A new MAC protocol based on the formulated guidelines is required.

Summary – ESIF ESIF is the first attempt to achieve concurrent packet

reception with on-demand protocols for MBAA ESIF removes the contention window based random

backoff in IEEE 802.11 DCF based protocols and uses embedded feedback to synchronize neighboring nodes

Allows nodes to receive or transmit multiple packets simultaneously in different beams

Cross layer information is used to guarantee long-term fairness

ESIF is a hybrid of synchronous and asynchronous on-demand medium access controlTwo protocols, SS-MQO and RICS to support QoS

over ESIF mechanism, are proposed recently.

Summary – HMAC HMAC is the first attempt to allow

coexistence of mesh and ad hoc networks with heterogeneous antenna technologies

HMAC is backward compatible with IEEE 802.11 DCF

Exploits Algorithms for Mitigating Deafness and Run-time Sender Estimation to achieve optimal performance

Future Work – MIMO Antennas

Source: Benjamin K. Ng and Elvino S. Sousa, “SSSMA for Multi-User MIMO Systems”, IEEE Microwave Magazine, vol. 5 , pp. 61-71 , June 2004

Single-user MIMO Spectral efficiency is

increased by supporting multiple data streams over spatial channels.

Spatial diversity is exploited to enhance the detection performance.

Multi-user MIMO MIMO channel is evenly

divided and allocated to multiple users.

Each user channel has access to the space domain over entire transmission channel and frequency bandwidth.

Future Work – MIMO Antennas

Source: http://www.airgonetworks.com/pdf/Farpoint Group 2003-242.1 MIMO Comes of Age.pdf

Outline of Proposed Work

Work completedWork in progressCo-authored

On-Demand Medium Access Control (MAC)

Multihop Wireless Network (MWN)

Antenna Technologies

SS-MQO;RICS

Energy Efficient Reliable

MAC

MIMO MAC

ReliableMAC

Hybrid MACESIF

Unified MAC

MDMAC-BB;MDMAC-NB;MMAC-BB;MMAC-NB

MSC AMD

Single Beam Antenna

Multiple Beam Antenna Array (MBAA)

Omnidirectional AntennaSmart Antenna Multiple Input Multiple Output (MIMO) Antenna

Wireless Mesh Network (WMN)

Wireless Sensor Network (WSN)

Mobile Ad hoc Network (MANET)

Publications Dhananjay Lal, Vivek Jain, Qing-An Zeng, Dharma P. Agrawal, “Performance

Evaluation of Medium Access Control for Multiple Beam Antenna Nodes in a Wireless LAN,” in IEEE Transactions on Parallel and Distributed Systems, Vol.15, No. 12, pp. 1117-1129, 2004.

Vivek Jain, Nagesh S. Nandiraju, and Dharma P. Agrawal, “Mode Selection Criteria in Mobile Ad hoc Networks using Heterogeneous Antenna Technologies,” in Proceedings of OPNETWORK 2005, Aug 2005.

Vivek Jain, Anurag Gupta, Dhananjay Lal, and Dharma P. Agrawal, “IEEE 802.11 DCF Based MAC Protocols for Multiple Beam Antennas and Their Limitations,” in Proceedings of IEEE MASS, Nov. 2005.

Vivek Jain, Anurag Gupta, Dhananjay Lal, and Dharma P. Agrawal, “A Cross Layer MAC with Explicit Synchronization through Intelligent Feedback for Multiple Beam Antennas,” in Proceedings of IEEE GlobeCom, Nov. 2005.

Vivek Jain and Dharma P. Agrawal, “Mitigating Deafness in Beamforming Antennas,” in Proceedings of IEEE Sarnoff Symposium, March 2006.

Ratnabali Biswas, Vivek Jain, Chittabrata Ghosh, and Dharma P. Agrawal, “On-Demand Reliable Medium Access in Sensor Networks,” in IEEE WoWMoM 2006 (accepted).

Anurag Gupta, Vivek Jain, and Dharma P. Agrawal, “Differentiated Service Classes Over Multiple Beam Antennas,” manuscript submitted.

Vivek Jain and Dharma P. Agrawal, “Concurrent Receptions with On-Demand Medium Access Protocols for Multiple Beam Antennas,” manuscript submitted.

Vivek Jain, Anurag Gupta, and Dharma P. Agrawal, “On Medium Access in Multihop Wireless Networks with Heterogeneous Antenna Technologies,” manuscript submitted.

Thank You!!!

Questions ???

For more information http://www.ececs.uc.edu/~jainvk