7/28/2019 cross layer design to achieve QOS
1/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 1
1. INTRODUCTION
Cross-Layer design has been the focus of several recent research efforts. Due to the highly
variable nature of the links used in wireless communication systems and the resource-poor nature of
the wireless mobile devices, there have been multiple research efforts to improve the performance of
the protocol stack by allowing cross-layer interaction in wireless systems. Cross-layer interaction
means allowing communication of a layer with any other possibly non-adjacent layer in the protocol
stack. Several issues related to the cross-layer design paradigm need to be addressed before it can
achieve its promises. One of these issues is to have a well defined framework that manages the
interaction between the different layers of the protocol stack, such that the modularity of the stack is
preserved while still achieving the flexibility and adaptability which cross-layer design promises.
This seminar addresses this issue by proposing a cross-layer coordination framework for next
generation wireless systems. The proposed framework enables the interaction between non-adjacent
layers in a systematic organized way while preserving the modularity of each layer. We believe that
the existence of such a framework will ease the development of cross-layer design schemes.
Figure : OSI Protocol Stack
7/28/2019 cross layer design to achieve QOS
2/22
7/28/2019 cross layer design to achieve QOS
3/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 3
Bandwidth limitation and fluctuations:
It is known that the throughput of a wireless channel may be reduced due to multipath fading,
co-channel interference, and noise disturbances. The capacity of a wireless channel may fluctuate
with the changing distance between the base station and the mobile host. In the meanwhile, the
network provides best-effort service and it does not provide Quos (Quality of Service) guarantee for
services. Specifically, network conditions and characteristics such as bandwidth, packet loss ratio,
delay, and delay jitter vary from time to time. Considering the bandwidth fluctuation, it is important
to estimate the available bandwidth dynamically. Throughput calculation, bandwidth probing,
packet pair are several popular techniques for bandwidth measurement .
Considering the bandwidth limitation, especially for wireless channel, it is essential to
improve the bandwidth utilization. It is known RTP/UDP/IP and TCP/IP have the problem of the
large header overhead on bandwidth-limited links. Header compression has been proven to be
efficient for using those protocols. Unfortunately, existing header compression schemes do not work
well on noisy links, especially the one with high bit error rate and long roundtrip time.
Low performances for traditional transport-layer protocols:
It is known that traditional transport layer protocol assumes congestion in the network to be
the primary cause for packet losses and unusual delay. It will decrease the transmitting rate in the
case of packet lost. Unfortunately, packets are lost in wireless channel due to channel error rather
than congestion, thereby resulting in an unnecessary reduction in end-to-end throughput. For wireless
network itself, both the high BER and frequently occurred fading make packet loss ratio very high
during a TCP/UDP connection. Many works have been made to overcome the drawbacks of
transport protocols over wireless networks.
Different Quos requirements for different types of data:
In general, different kinds of media have different characteristics. Real-time media such as
video or audio is delay sensitive but capable of tolerating certain degree of errors. Non-real-time
media such as Web data is less delay sensitive but requires reliable transmission. Consequently,
unequal error control and priority-based scheduling schemes are needed for different types of media.
Cross layer communication framework helps different kinds of protocols to communicate
with each other without knowing the protocol or system architecture specification. It maintains a
parameter repository to manage the data format that these layers accept or produce. Cross layer
7/28/2019 cross layer design to achieve QOS
4/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 4
manager manages and co-ordinates the communication among different protocols by sending and
receiving the event messages between different types of protocols. The below figures such a cross
layer framework representation where the cross layer manages the communication between different
types of protocols by sending and receiving the event messages and using the state variables.
Figure : cross layer design for different transmission protocols
While the standard TCP/IP stack has worked well for wired links, it suffers from badperformance when used over wireless links. When compared with wired links, wireless links in
general have lower bandwidths available, higher transmission delays, and higher BERs, and suffer
from channel fading. There is not much that can be done at the protocol stack level to work around
the first two problems, and users learn to live with these limitations. Unfortunately, transport
protocols suffer severely from the consequences of the last two problems. Again, the widely used
TCP protocol is the primary example of this situation. On one hand, erroneous datagrams are
automatically dropped by the link layer, while TCP always interprets losses as a congestion signal.
Thus, the TCP congestion control algorithm decreases (usually by half) the congestion window and
enters the congestion avoidance state, where the congestion window grows linearly. The net result is
a significant reduction in effective performance, which the user does not understand, knowing the
advertised bandwidth of the network interface card. On the other hand, wireless links often
experience channel fading effects, consisting of fluctuation of the channel capacity over time. We
can distinguish between slow and fast fading according to its duration, which is usually related to the
speed of a mobile node's movement. Although fast channel fading has little impact on the
performance of TCP, when slow channel fading occurs several consecutive TCP packets are
7/28/2019 cross layer design to achieve QOS
5/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 5
dropped. Thus, TCP congestion control will inevitably and quickly lower the congestion window to
its minimum value. Unfortunately, due to the multiplicative decrease additive increase property
of congestion avoidance the congestion window is very slow to return to its original value therefore
fading occurred. Once more, the net result is a decrease .in effective throughput and, consequently,
underutilization of resources.
The future of mobile communication is heading towards ubiquitous Next Generation (NGN)
heterogeneous or 4G Networks. The 4G networks are intended to cater to seamless and fast handover
between applications to be used anywhere, at any time, using different access technologies. There is
an increased demand for faster, seamless and cheaper multimedia delivery over the wireless Internet.
The Next Generation wireless Networks will have an all IP-based architecture to support this
heterogeneity. Currently, the foundation of these heterogeneous networks is based on a strict layered
architecture referred to as the Open Systems Interconnect (OSI) Protocol stack. However, this OSI
Protocol stack presents various bottlenecks to the performance of real-time applications over the
Internet. Consequently, a different approach, called Cross Layer Design has been introduced to
optimize the performance of the Quality of Service (QOS) of these applications. This seminar
discusses the basic OSI protocol stack and its need for modification.
Cross-Layer design has been the focus of several recent research efforts. Due to the highly
variable nature of the links used in wireless communication systems and the resource-poor nature of
the wireless mobile devices, there have been multiple research efforts to improve the performance of
the protocol stack by allowing cross-layer interaction in wireless systems. Cross-layer interaction
means allowing communication of a layer with any other possibly non-adjacent layer in the protocol
stack. Several issues related to the cross-layer design paradigm need to be addressed before it can
achieve its promises. One of these issues is to have a well defined framework that manages the
interaction between the different layers of the protocol stack, such that the modularity of the stack is
preserved while still achieving the flexibility and adaptability which cross-layer design promises.
This seminar addresses this issue by proposing a cross-layer coordination framework for next
generation wireless systems. The proposed framework enables the interaction between non-adjacent
layers in a systematic organized way while preserving the modularity of each layer. We believe that
the existence of such a framework will ease the development of cross-layer design schemes.
7/28/2019 cross layer design to achieve QOS
6/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 6
3. KEY CHALLENGES
3.1 Real Time Application Requirements
Real-time applications have strict Qos constraints on bandwidth, delay and jitter. Although
the OSI protocol has been in existence for over 20 years, advancing technology and use of multiple
applications on the same protocol stack has led to increased bottlenecks in the performance of the
protocol model. Some of the bottlenecks experienced are: increased energy consumption,
unnecessary encapsulation and decapsulation, increased jitter and delay, unnecessary
retransmissions, reduced throughput and inability to meet different Quos requirements of different
applications running on same protocol. This seminar discusses the implementation of the cross layer
architecture to help speed up communication and provisioning of Quos requirements between the
layers of the OSI protocol stack. It proposes an extension to the design of the existing internet OSI
architecture that allows interaction between non-adjacent layers of the protocol stack. In particular, it
focuses on end-to-end user delay, prioritization and per-user throughput for a VoIP application.
3.2 Cross Layer Design Architecture
Figure : overview of cross layer coordination framework
Cross layer clients are added to each layer of the protocol stack to enable interaction with the
server (Manager). Parameters that characterize each individual layer are relayed to the client which
in turn communicates across the other layers through the cross layer server. When an event occurs
through the signaling scheme, a request is made to the server for particular information. The server in
turn relays relevant information to the particular layer making the request. This way, information can
be transferred across non-adjacent layers through the cross layer server.
7/28/2019 cross layer design to achieve QOS
7/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 7
Figure : Details of cross-layer coordination framework
In order to fully address all aspects of the framework, the following issues have to be
discussed in detail.
A. Cross-layer Client (to be added to each layer of the traditional protocol stack to enable thecross-layer coordination operation)
B. Cross-Layer ServerC. Signaling Scheme (Event Messages)D. Adaptation Algorithms (reside in the cross-layer client)
Breaking the cross-layer framework into these four main blocks makes it easier to handle each
required aspect of the framework independently and focus on its solution.
7/28/2019 cross layer design to achieve QOS
8/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 8
A. The Cross-Layer Client
The cross-layer client is the part which is added to each layer of the protocol stack to
facilitate the cross-layer coordination functionality. The client communicates with other clients in
other layers through the cross-layer server to achieve the required functionality. The cross-layer
client consists of two major parts, namely:
a. The Adaptation ModuleThe adaptation module could be divided into three main parts:
The adaptation algorithm itself. This is the logic and the implementation which solves acertain problem. It receives events from the server and sends events to it and can change the
state variables of the layer it resides in.
It communicates events to other layers through the server. The conversion of the receivedparameters from other layers into the form that the algorithm needs to operate on.
The parameters from other layers which might be necessary for the proper operation of thealgorithm.
b. Abstracted Layer StateEach layer in the protocol stack could be viewed as a set of parameters. Depending on
the value of these parameters, one can determine the overall state of the layer and determine
its behavior (i.e. each layer could be abstracted in a set of parameters).
B. The Cross-Layer Server
The cross-layer server resides outside the protocol stack to facilitate the cross-layer
coordination functionality. It could be viewed as a service or part of the operating system. The cross
layer server consists of two major modules, namely:
a. Control ModuleThe Control Module is divided into two parts:
1. The Action Module which takes the actions towards other layers, i.e. sends events to themin the form of Event Messages, or takes internal actions in response to a certain event
received from a client.
7/28/2019 cross layer design to achieve QOS
9/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 9
2. The Event Management Module which manages concurrent events and schedules whichevent to be handled first in the case of several events occurring at the same time. This form of
scheduling should be easy and not time consuming; otherwise it will affect the performance
of the system.
b. Parameter Management ModuleThis Module consists of the Parameter Repository which is responsible for saving the
parameters (abstracted state) of each layer in a suitable form for the other layers to access easily.
C . Signaling Scheme (Event Messages)As mentioned previously, the communication between different non-adjacent layers in the
protocol stack happens through the cross-layer server. When an initiating layer wants to send a
certain event to another target layer, the client of the initiating layer sends this event to the server,
which forwards it to the target layer. The event message should be expressive enough to carry the
necessary information from one layer to another. Different events could achieve different tasks, for
example an event could be used to inform the server of a change of a parameter and report its new
value. Another event could be used to request a certain parameter value from the server. The server
could send events to the clients to request a certain action to be performed or request from the client
a certain parameter value. An event is transmitted from a client to the server via an Event Message.
An Event Message could contain one event or several events with or without associated
parameters. Each event should have a priority to facilitate scheduling it among other events. Event
parameters are optional (i.e. there could be an event without a parameter). Events should be encoded
in a TLV (Type, Length, and Value) to facilitate the existence of several events per message.
3.3 What the Cross-Layer Designer needs to Specify
In order to utilize the proposed framework, the cross-layer protocol designer needs to specify
following parts in the cross-layer framework:
1. The adaptation algorithm which will be implemented inside the client of a certain layer orseveral layers of the protocol stack.
2. The necessary events, their numbers, types, their parameters and the associated action witheach event, which will be sent from a client to the server and vice versa.
7/28/2019 cross layer design to achieve QOS
10/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 10
3. A priority policy for differentiating between different events in case they are all present at thesame time at the server side. This policy should define which one to execute first when
several events of different types are present
3.4 Cross layer design for voice and video applications.
Cross layer can be used for the bandwidth hungry application like voice, video or multimedia
transmission where the jitter, delay many condition decide the Quos.
A. VOICE OVER INTERNET PROTOCOLVoIP is an interactive voice application that facilitates routing of voice conversations over the
internet or any other internet protocol (IP)-based protocol. These protocols carry telephony signals as
data packets. They are reduced in data rate using speech data compression techniques and
encapsulated in a data stream over the Internet protocol. VoIP is location independent. Only an
Internet connection to the VoIP provider is required.
Cross layer implementation is especially necessary when the channel is congested. This is
when all the nodes are competing for the available bandwidth. As the sending rate-per-node
increases, the available bandwidth in the channel decreases and congestion occurs. This design
consists of an IEEE 802.11bWLAN network supporting various VoIP applications using a UDP/IP
transport protocol. It involves interaction between the application, transport and MAC layers of the
protocol stack. This proposed architecture is used to optimize two Quos requirements of real-time
applications. These are: per-user throughput and end-to-end delay.
Cross layer interaction is created between the transport and the MAC layer of the VoIP
application. The application layer relays to the transport layers its specifications (i.e. sending rate and
required throughput). The transport layer directly contacts the MAC layer to relay the VoIP
applications specifics, the type of packets being sent (CBR packets) and its priority status. At the
MAC layer, a procedure is implemented such that if it detects that the sending rate of the priority
application is much greater than the throughput received at the base station, the MAC layer increases
its packet size by a pre-defined set number of bytes. When the sending rate of this application is
much greater than the per-user throughput at the base station, this is an indication that the application
is not being adequately served. It is therefore an indication that the channel is experiencing
congestion. The MAC layer continues to increase the packet size after a set interval within bound
7/28/2019 cross layer design to achieve QOS
11/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 11
until the transport layer relays to the MAC layer that the priority nodes throughput is at the required
level. To exemplify the effectiveness of this design, consider a number of applications running over
an 802.11b application using CSMA/CA. Each of the\ applications is waiting and listening for a slot
to transmit its information.
Figure : Cross Layer Interaction between the MAC a Transport Layer.
Consider that each packet has 50 bytes of information. As the application using the cross
layer realizes that the channel is getting congested, the transport layer of the VoIP application alerts
the MAC layer of the need for prioritization to send information with minimal delay. The MAC
layer, on receiving this information, increases its packet size to 100 bytes.
Hence, 100 bytes of VoIP are sent for half the contention period of the other competing
applications. Therefore, more information is sent for half the time. The other applications will have
to wait and listen two contention periods in order to send the same amount of information (50 bytes
per contention period). By increasing the packet size of the application, the MAC layer gives this
7/28/2019 cross layer design to achieve QOS
12/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 12
VoIP application increased priority. In addition to this, the CSMA/CA contention period of this
application shows a marked decrease. More bytes of information are being sent to the base station for
a shorter period of time than the applications using the normal OSI protocol layered stack
mechanism. Accordingly, the per-user throughput of the VoIP application is improved despite the
congested nature of the IEEE802.11b WLAN network.
VoIP data travels on a Real-time Transport Protocol (RTP) over User Datagram Protocol
(UDP). VoIP packets are very small with a payload of about 20 to 150 bytes. An RTP/UDP/IP
header is exactly 40 bytes (IP = 20 bytes, UDP = 12 bytes, RTP = 8 bytes). Due to the high relation
between the header size and the payload size, transmission of VoIP is inefficient.
It is important to note, when transferring information using VoIP, the Quality of Service
requirements of the application. The main constraints specified by VoIP are: delay, jitter, packet loss
and throughput.
CARRIER SENSE MULTIPLE ACCESS/COLLISION AVOIDANCE
Link adaption is a technique used to handle the effects caused by the changes in the channel
condition. It is performed at the link or Media Access Control (MAC) layer. This technique is used
to automatically adjust a number of radio/MAC parameters, so that optimal throughput is achieved.
Cross layer signaling can be used to pass the information across the layers. The 802.11b family uses
a MAC layer protocol called Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA).
It is a peer-to peer Ethernet protocol that needs no master station. In CSMA/CA, if a node wants to
transmit, it performs the following sequence
A node desiring to transmit listens to the channel. If the channel is busy (i.e. another node is
transmitting), the node waits until transmission stops and then waits a further contention period (A
contention period is a random period waited after every transmission on every node. It is
approximately 20-50 ms).When it senses that the channel is free for a specified time (called the
Distributed Inter Frame Space (DIFS)), the node is allowed to transmit. The DIFS is the Inter Frame
Space used for a node that is willing to start a new transmission. It is usually 128 microseconds long.
SIMULATION SETUP
Using a Network Simulation tool called NS-2, the topology for the cross layer interaction is
set up. The various layers and their individual protocols are included in the nodes and cross layer
7/28/2019 cross layer design to achieve QOS
13/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 13
interaction between the layers is effected as explained below.The basic topology consists of eleven
wired nodes connected to a base station (access point). A base station is a wired gateway between
wired and wireless domains. To measure the throughput at the base station, we attach a wired node to
the base station using a large capacity lossless link. This wired node acts as the sink for the packets
being sent from the wireless nodes to the base station. The wireless nodes represent the different
real-time applications. Each wireless node represents a different user connected to the network using
a specific application.
To show the effect of cross layer architecture on the performance, a specific node is identified
as the priority node. This priority node represents the end-user that requires priority service over
the other users. It is given all the characteristics of a VoIP application. In this node, cross layer
design method is implemented. The rest of the nodes in the network are also given all the
characteristics of real-time applications. However, these non-priority nodes use the ordinary OSI
Protocol stack adjacent layer communication.
Figure : Simulation Topology Setup
An Agent is created at each wireless node. Each node is identified as a UDP Agent since real-time applications are throttled when TCP is used. The traffic is specified to be CBR traffic. The
sending rate and packet size of each of the wireless nodes is then set to the same value. Each node
gets the same initial characteristic. Packets are then sent from each of the eleven wireless nodes to
the base station. This approach is used to ensure that all the nodes are given the same basic
environment and channel conditions. It also ensures that there is no bias on measurement of results.
The aggregate and per-user throughput of each node is measured at the base station and recorded
using trace files in NS-2. The average end-user delay is also recorded.
7/28/2019 cross layer design to achieve QOS
14/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 14
Assumptions
A lossless wired link with a very large capacity is used to connect the base station to the
wired node. The assumption is made that the throughput at the base station exactly matches the
throughput at the wired node. The per user throughput, aggregate throughput and end-user delay are
hence measured at the wired node. Also, as the congestion increases, the number of packets lost or
dropped increases too. It is assumed that the number of packets dropped does not adversely affect the
throughput o f any of the nodes.For accuracy of simulation results, eleven nodes are simulated and
tested. However, for ease of presentation of results in this paper, only four nodes have been
represented in the graphs. These include: three randomly chosen non-priority nodes and the sole
priority node.
USECASE
For effective evaluation of the simulation results, they have been divided up into two
different scenarios: before and after congestion. In each scenario, performance of the priority node
that uses cross layer design is compared to other non-priority nodes that use the strictly modular OSI
protocol stack. This helps to nullify any bias to the results.
Per-user Throughput
Scenario 1: Before Congestion
In this scenario, there is no congestion detected in the link. The channel conditions are
capable of supporting all the traffic passing through the link. All the nodes connected to the base
station are able to transmit their packet information with ease. The sending rate of each of the nodes
is set at 10 000 bps (0.01
Mbps).
Figure: Throughput vs. time before saturation
7/28/2019 cross layer design to achieve QOS
15/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 15
Figure above shows that all the nodes are able to attain maximum through put within minimal
time of 30 seconds. No node requires or is given priority over the other nodes. The per-user through
put of all the nodes is relatively equal to the sending rate i.e.0.01 Mbps. The low sending rate of each
application ensures that the bandwidth is fairly shared by all the applications. There is also enough
bandwidth left over to cater for other bandwidth hungry applications that may join the network.
Since there is no congestion, the cross layer procedure at the link layer is not necessary and the VoIP
priority application receives the same service as the rest of the nodes.
Scenario 2: High congestion
In this scenario, the sending rate of all the nodes is now set to 0.5 Mbps. The channel is
battling to cater to all the applications and their requirements. The end-to-end delay of the
participating applications increases drastically and the throughput per-user reduces considerably.
Figure: Throughput vs. time after saturation
Figure is a clear illustration of the improved performance of the priority node over the non
priority nodes. It is evident that the priority nodes through put increases to a value of 0.58 Mbps
well above the sending rate of the application. However, the other non-priority nodes are greatly
affected by the high congestion levels. In addition to this, they have to cater for the increased
throughput of the priority node. The through put of the other nodes lowers to an average value of
7/28/2019 cross layer design to achieve QOS
16/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 16
0.41 Mbps. However, the congestion does affect the priority node to some extent as its time to reach
maximum throughput increases to a value of 150s.
End-to-end Delay
Scenario 1: Before Congestion
This scenario remains the same as explained above. The sending rate of each node is set to a
value of 10 000 bps (0.01 Mbps). At this rate, all the nodes are freely transmitting their information.
There is enough bandwidth on the link to cater for\ all the nodes simultaneously. The total
bandwidth for all the nodes (0.11 Mbps) is much lower than the total capacity of the link. This is
approximated to be 4.5 Mbps. Figure 7 below shows the end-to-end user delays experienced by both
the priority nodes and non priority nodes collectively.
Figure: Delay vs. time before saturation
Fig above shows that the end-user delay for\ the priority node is relatively the same as the
average delay for the non-priority nodes. This is due to the fact that there is no contention for
bandwidth so all the nodes are transmitting data at the designated sending rate.
Scenario 2: After Congestion
In this scenario, the nodes are all competing for the available bandwidth. The sending rate ofeach node is set to a high value of 0.5 Mbps. The overall capacity of all the nodes is much greater
7/28/2019 cross layer design to achieve QOS
17/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 17
than the total capacity of the link. As contention for this bandwidth increases, the priority node states
its need for prioritization above the other nodes.
Figure : Delay vs. time with congestion
Fig. above shows that the end to end delay increases for the non-priority nodes. However, the
end-user delay for the priority node remains below 200 milliseconds according to the required QoS
requirements for end-to end delay stated in Table I above. This is an indication that the cross layer
approach suggested here is effective in areas of high congestion. There is a marked reduction in end-
user delay.
B. VIDEO APPLICATIONSThe following figures show the event messages and the flow chart to explain the
video communication using the cross layer design framework.
This section illustrates how the framework is used in the case of the transmission of video
over a wireless channel using cross layer operation. Seminar discuses a transmission scheme for real
time video over wireless which utilizes cross-layer interaction between the physical layer and the
video application layer. In their proposal, the physical layer determines the number of bits that it can
transmit each coherence period. This Information is supplied to the video application which adapts
its transmission based on that number to achieve the best performance.
7/28/2019 cross layer design to achieve QOS
18/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 18
Figure : Event message structure
Figure : Adaptation algorithm for video application
7/28/2019 cross layer design to achieve QOS
19/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 19
The technique makes use of combined Progressive Group of Pictures (PGOP) and
Fine Grained Scalable (FGS) video to adapt the video transmission to the wireless channel. The FGS
video is a two layered scheme comprising a base layer (BL) and an enhancement layer (EL). With
the PGOP scheme, the BL rate can be maintained close to a predetermined constant. The BL
guarantees an acceptable video quality. The EL can be received even partially. The reception of
additional EL bits can only increase the video quality. FGS is used in the proposal due to its
flexibility of arbitrary bit rate truncation at the EL . This is utilized as an advantage to adapt to the
fluctuating channel capacity. In the proposed cross-layer solution for rate control, the encoder does
not vary the transmission rate every coherence period. The encoder provides all the frame bits in a
buffer in a continuous fashion for the packetization process to truncate.
The packetization process will choose the maximal number of bits to transmit based on the
feedback from the physical layer in each coherence period and will truncate the remaining bits. The
packetizer does not have to know the semantics of the data that needs to be truncated as arbitrary
truncation of FGS video is possible. The packetizer selects only the necessary bits from the
beginning of the buffer and truncates the rest. The packetizer chooses only the transportable number
of bits minus the lower layer packet overhead (headers of lower layers) of the layers below it. The
total number of bits available at the physical layer for transport after including packetization would
be the exact number of transportable bits for that coherence period. This scheme provides real-time
rate adaptation for every coherence period.
We use this application as an example of how our cross-layer framework could be used to
allow cross-layer coordination. The requirement in this case is that the client of the physical layer
needs to inform the client of the video application the number of bits which it can transmit during
each coherence period. This corresponds to an event that the physical layer needs to send to the video
application via the cross-layer framework. This event is transported in an Event Message which
would be in the form shown in Figure . The message carries one event which is the number of bits
that the channel state allows to be transmitted during the next coherence period denoted by X. The
event priority is high, since if the video application does not receive this number it will not be able to
construct the video packet in the correct form. There could be a default value which is the minimum
value which corresponds to only accommodate the BL bits. The cross-layer server upon receiving
this message from the physical layer client determines its event type. The management module in the
server then decides to forward the event to the video application. Thus another message from the
server to the client of the video application is needed. This message has the same type and format as
7/28/2019 cross layer design to achieve QOS
20/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 20
the one between the physical layer client and the cross-layer server. The client in the video
application receives this event and starts performing the adaptation algorithm which is shown in
Figure. The operation of the algorithm is very simple, the algorithm drops a number of bits from the
buffer filled by the video encoder so that the remaining number of bits is equal to the number
supplied by the physical layer minus the overhead of lower layer protocols (RTP, UDP, IP and Link
layer headers). Note that the demonstration of the adaptation algorithm does not take into account the
occurrence of an error that prevents the event from reaching the video application. Several strategies
could be used to solve this issue, for example sending the BL bits only or sending the same number
of bits as the last packet, or having a weighted average of the sizes of several previously transmitted
packets.
In this case no policy is required for determining how this event should be handled in the case
of the existence of other events at the same time. An example of such a policy would be if the
physical layer sends the event fading_start, which indicates that a sudden fade has occurred in the
wireless channel, meaning that the channel is inaccessible, in this case it should be forwarded to the
video application which should in turn drop the packet it was constructing. The abstracted layer state
of the video application contains the encoding rate that the video is using and the number of bits in
the BL and EL and any additional information for example the encoding algorithm used.
The parameter repository in the cross-layer server in this case will be used to store the base
rate of the BL at which the video application is sending and the maximum rate corresponding to BL
+ EL. It also is used to store the values that the physical layer reported for the previous coherence
periods.
4.FURURE SCOPE
In this seminar we have presented an overview of recent advances in the area of cross-layer
design. We identified an essential issue that we believe is crucial to the success of the cross-layer
design scheme. This issue is the existence of an organized framework which defines how the non-
adjacent layers communicate.
Future work includes investigating the impact of implementing several other cross-layer
adaptation algorithms under the same framework and qualitatively assessing the performance of the
framework and applications under such operation.
7/28/2019 cross layer design to achieve QOS
21/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC, BAGALKOT Page 21
5. CONCLUSIONS
The results above reveal that this proposed Cross Layer architecture is indeed an effective
means of reducing the end-to-end delay experienced by giving priority to a VoIP or video or
multimedia application over the non-priority applications on request and reducing its contention
period. Reduction of this contention period facilitates a reduction in overall end-to-end delay
between the source and of the application. In addition, this design method is effective in increasing
the per-user throughput of the application. As more bytes per packet are transmitted during the
allocated transmission time, the overall per-user throughput of the prioritized application increases
greatly.
The effectiveness of this design is more evident during periods of high congestion when all
the applications are struggling to utilize the available bandwidth of the channel. In the event that all
the nodes: priority and non-priority used this proposed design, the overall throughput would be
increased and end-to-end delay per user decreased due to the increase in packet size and reduction of
contention period for each application. Cross Layer architecture is therefore a more effective and
easier means of provisioning Qos requirements to real-time applications.
REFERENCES
1. Qian Zhang, Fan Yang, and Wenwu Zhu, Cross-Layer Quos Support for MultimediaDelivery over Wireless Internet EURASIP Journal on Applied SignalProcessing. Volume
2005 , pp 207219, January 2005
2. ustavo Carneiro, Jose Ruela, and Manuel Ricardo, Cross Layer Design in 4G WirelessTerminals, IEEE Wireless Communications, Volume: 11, Issue 2, pp 7-13, Apr 2004.
3. James F. Kurose, Keith W. Ross Computer Networking. A top down approach featuring theInternet 3rd Edition 2005
4. Deeya Shakti Nursimloo, A two layered mobility support architecture: Fast Mobile IPv6 andSession Initiation Protocol, Masters thesis, University of Cape Town, 2006.
5. http://www.zytrax.com/tech/wireless/802_mac.htm - Retrieved on April 28, 20096. Karim M. El Defrawy, Magda S. El Zarki, and Mohamed M. Khai ry, Proposal for a Cross-
LayerCoordination Framework for Next Generation Wireless Systems, Proceedings of the
2006 international conference on Wireless communications and mobile computing, pp. 141-
146, Vancourver, 2006.
7. http://sss-mag.com/pdf/802_11tut.pdf Retrieved on 30 June, 2009
7/28/2019 cross layer design to achieve QOS
22/22
CROSS LAYER DESIGN IN HETEROGENEOUS NETWORK FOR QOS SUPPORT
DEPARTMENT OF COMPUTER SCIENCE BEC BAGALKOT Page 22
8. Arulsaravana Jeyaraj, Liang Cheng, Magda El Zarki. Proposal for a Cross Layer scheme forReal Time Wireless Video. In preparation, April, 2006.
9. Carneiro, G.; Ruela, J.; Ricardo, M. Cross-Layer Design in 4G Wireless Terminals. IEEEWireless Communications Magazine. Volume 11, Issue 2, Apr. 2004 Page(s):713.
10.Haitao Zheng. Optimizing wireless multimedia transmissions through cross layer design.IEEE Proceedings of International Conference on Multimedia and Expo, 2003(ICME '03).
Volume 1, 6-9 July 2003 Page(s):I - 185-8 Vol.1.