Document Sample
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME
                             TECHNOLOGY (IJCET)

ISSN 0976 – 6367(Print)
ISSN 0976 – 6375(Online)                                                       IJCET
Volume 4, Issue 4, July-August (2013), pp. 100-115
Journal Impact Factor (2013): 6.1302 (Calculated by GISI)                    ©IAEME


                 GK Srinivasa Gowda1, CV Srikrishna2 and Kashyap Dhruve3
                              (Professor, SSET, Ernakulum, Kerala, India)
                               (Professor, MCA, PESIT, Bangalore, India)
                      (Technical Director, Planet-i Technologies, Bangalore, India)


         In order to have maximum utilization of the resources as well as to enhance the throughput of
the network, its quality of service (QoS) plays a significant role. The decentralized characteristics of
Ad hoc network need a highly optimized and enhanced technique for optimizing the fundamental
performance parameters of network. An effective available bandwidth estimation approach and
throughput optimization mechanisms might be the optimum solution for increasing throughput as
well as QoS of network.The approach of estimating end-to-end delay in IEEE 802.11 multihop
network might be an effective way for bandwidth estimation and optimum utilization of resources. In
this paper, in order to estimate end-to-end delay in IEEE 802.11 network protocol, enhanced
mathematical expressions have been developed in the service time patterns. In this paper the queuing
theory approach called; M/M/1/K queue has been used and on the basis of it every consisting nodes
have been modeled. In this work, in spite of employing mentioned approach, few more optimizing
mechanisms for admission control and different delay estimation mechanisms have been considered
for coming up with an effective and the best solution for end-to-end delay estimation in 802.11
network protocol. Thus combining these noble approaches, an optimum technique called “End to
End Delay estimation (             ” in mobile wireless AdHoc Network has been developed, that
facilitates QoS optimization with optimum bandwidth estimation and resource utilization. The
proposed              mechanism has depicted great performances for bandwidth estimation and
resource utilization in decentralized network as compared to networks of centralized nature. The
simulation framework for proposed               has been developed on .net platform with C sharp
programming language for varied network size and results has been obtained for different network
parameters, where              has performed better in terms of lower packet drop ratio and lower
processing delay while with higher packet success rate and lower end to end delay and thus depicting
QoS optimization and optimum resource utilization in the wireless Adhoc network 802.11.

Keywords: End to End delay, decentralized network,              , Quality of Service, Adhoc network.

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME


        In recent times, the works done on the guaranteed quality of service (        ) in the Ad hoc
networks has grabbed attention throughout the world. IEEE 802.11 technology [1] is the main
technology used in these networks and the same is assumed by the most of the works done in this
area. The main reason for using this technology is that, it comes at a very low cost and is effective
and provides a well distributed radio medium access that can easily be implemented in the ad hoc
networks. This random radio medium access which is provided by the IEEE 802.11 standards gives
us a great control on the emission and makes it very much difficult to share it on a multihop context
[2]. There are several works which offers the quality of service to the ad hoc networks which is
based on the IEEE 802.11 either by providing a delay guarantee or throughput guarantees or may be
both. When we see this several studies, throughput guarantee is mainly provided by them we can see
them in [3], [4], [5], [6] and very few of them has concentrated on the delay guarantees. But if the
elucidations given for the throughput guarantee is not perfect, then they can’t be used in more
efficient manner. IEEE 802.11 technology which gives the complex radio medium sharing is much
better integrated in the       solution. In context to this the solutions provided can give us a very
much precise available bandwidth estimation and so it gives us the guarantee of throughput
efficiency. To provide an ensured delay is a much more daring task. It is very much tough to get the
exact delay (as mentioned in the [7]) because of the strong reliability among the flows in the
multihop setting in wireless network. In this piece of writing the authors explains that it is very hard
task to design an admission control protocol which is based on the measurement for the delay
parameter when compared to the parameter of throughput. In this write-up a new protocol for delay
guarantee has been proposed in the multihop networks which are wireless. With this study, we come
to know that it is likely to design a well organized admission control protocol which is based on the
measurement for delay limitations. This proposed protocol is known as              (End to End Delay
estimation) which depends upon the prior estimation of average end to end delay. The main model
behind the estimation is a very simple model of IEEE 802.11 nodes which provides the estimation
from a precise assessment of each and every link’s collision probability. When we add the two i.e.
this estimation and the precise admission control, a guaranteed estimation delay is provided when a
new flow starts. The guarantee mainly depends upon how the bandwidth available is correlated to the
estimated delay so as to provide efficient bandwidth estimation. This finally estimated by a protocol
known as          (Available Bandwidth Estimation) which also gives us the most precise estimation
[8]. A decentralized Ad hoc communication network is moreover used in the architecture of the
system proposed and after that there is a comparison between the several parameters has been done
which is transparent. The system proposed does not add any extra overhead as it mainly uses the
control packets which is provided by the             and is required for the estimation of available
bandwidth and thus is not very costly. In this article for estimating end to end delay to facilitate a
higher       in the network communication, strong protocol architecture is formulated. A comparative
study between the decentralized and centralized network Ad hoc network is carried out as well and
different results we get from the different parameters has been done in the paper with the delay
parameter and the significance of it over      optimization.
        The paper is organized like; the related works of the research domain have been described in
Section II whichis followed by Section III that contains the contribution made by author or the
research accomplished. The next section is Section IV thatdescribes the research implementation and
development of proposed protocol called End to End Delay Estimation in wireless AdHoc network
(         ). The research implementation and results obtained and are in section V. The last section,
the conclusion part of the research work which is followed by the references considered in research
work is presented in section VI.

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME


         Chaudhary, D.D. and Waghmare [9] intended a probabilistic model to make a comparison
between the transmission delays in these two models. The WSN model which is proposed was
evaluated by estimating energy consumption, end-to-end delay and packet drop ratio of both the
models. By minimizing the delay in packet delivery of the network The QoS was improved. It was
observed that similarly overall delay can be minimized significantly.
         Jae-Ho Lee [10] proposed a system called WTE-MAC, by using a new model which can
reduce the delivery delay of asynchronous MAC protocols in multi-hop environment and is called as
Virtual Tunnel (VT). In this model, without the special process, through the estimation of next
wakeup time of peer node, each node on the transmission path can recover end-to-end delay in multi-
hop topologies. And it becomes low power consumption by reducing unnecessary retransmissions.
         S.; Fohler, G. [11] proposed a notion which is generalized and allows under reliable condition
by providing some meaningful performance matrices, common in WSN. A probabilistic metric has
been developed by them which for the timeline performance capture the level of confidence without
restricting the applicability which consists of the limitation of the delay distribution in end to end
delay network by current information of the intermediate hops, and requires computational resources
and low memory.
         Baoliang Li [12] presented an end to end delay which is very important in the
performance evolution of the Network On Chip (NoC). No assumptions have been made on the NoC
topology, hardware implementation and the traffic pattern which makes it attractive for the fast
performance evolution. The results show the approach’s robustness.
         Rodoplu, V [13] proposed empirical system architecture for estimating end to end
Voice-over-IP delay estimation for a multihop wireless communication network. Initially, they do
characterizes the VoIP regime that represents a regime of network operation where a part of packets
are received by the gateway lying in the maximum VoIP’s networking delay and especially in the
circumstances the delay is always less than a maximum probability of outages. In their research
work, they have depicted that in defined VoIP regime, the upstream VoIP delay is well structured by
employing an exponential distribution that solely depends on the number of hops to the gateway.
Similarly, they illustrates that the coherence time of VoIP regime is large enough so that each
participating node can effectively estimate its parameters and end to end delay from its present
position and for accomplishing call admission decisions.
         Kataria, D et al [14] presented an enhanced delay accumulation mechanism that was even
considered in ATM Forum. The enhanced mechanism performs better as compared to the majority of
existing system. The proposed system is very simple and sophisticated to implement and even it
needs very few functional parameters than the other alternative, the asymptotic method. On the other
hand, the researchers illustrated that the enhanced method is backward compatible with the existing
         Dong Linfang [15] implemented a robust Markov chain model for analyzing the
probability of transmission at each node in an arbitrary slot, and then they do derive the mechanism
for channel access delay estimation. This proposed system was extended from analyzing the single-
hop average packet delay to estimating the end-to-end packet delay in multi-hop ad hoc networks
without considering any hypothesis that the traffic to be in a saturation state.
         Matta, J.M.; Takeshita [16] proposed a QoS measurement scheme for the VOIP which makes
it compulsory a less number of probe packets while they add the simple queuing delay estimation in
the core routers. The main concept behind that delay is that it provides jitters and available
bandwidth seen by the voice application and is caused by the queuing changes at intermediate hops. .
To obtain accurate QoS limitations collecting and combining queuing delay estimates from the core
routers is used.

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

        Despaux, F. [17] presented an observed support for an analytical approach for the end to
end delay estimation in multihop network by employing frequency domain analysis the simulation
validates the proposed analytical result of the distribution in end to end delay and compare it with
queuing based analysis using concrete scenarios. To provide the capable estimation of end to end
delay an analytical prediction scheme is insufficient and it requires to be mixed with detailed links
and nodes latencies distribution.


       The foundation or the building block of the research development and the assistance made by
the authors has been mainly presented in this section to accomplish the final aim of getting end to
end delay estimation in Ad hoc network and the optimization of QoS. The work for estimating the
available bandwidth is also mentioned in the section which is predeceased by mean delay estimation
in Ad hoc Network.

        To ensure the delay guarantees, our solutions depend upon the proper available bandwidth
estimation. After that we define the available bandwidth among the two neighbors so that the
maximum throughput can be transferred between the two peers without disrupting the ongoing flow
of the network. This term may not be confused with the link capacity (also called base bandwidth)
which designates the maximum throughput; a flow can be achieved between two neighbor nodes,
even t the cost of other flow level of service degradation.
For the estimation of available bandwidth the ABE (Available Bandwidth Estimation) has been
chosen which is first proposed in [8] and [6] refined it. ABE is the most accurate protocol when
compared to the other protocols of same goal while it requires a small overhead and is shown by the
authors in [6]. After considering the overlapping of the silence periods of both emitter and receiver
of a link, the collision probability that exists on the link and the back-off window size correlated to
this collision probability, ABE provides the accuracy which other protocols cannot provide. As our
delay estimation mainly depends upon this available bandwidth estimation, this section mainly
consists of the description of the ABE. We cannot include all the limitations of the ABE due to
limitation of space and for more detailed description one can refer to [8], [6]. To provide accurate
evaluation, some phenomena should be taken to the account when the IEEE 802.11 MAC protocol

       •   Carrier sense mechanism avoids two close emitters from transmitting at a same time.
           Therefore, an emitter has to share the channel bandwidth with all these close emitters. The
           channel utilization has to be observed to evaluate the capability of a node to emit a given
           traffic volume. In many protocols, the channel utilization is calculated by each node by
           observing the radio medium in its environment and measuring the total quantity of time that
           is redundant for emitting frames. Therefore, this scheme does not only take into account the
           bandwidth used in the transmission range of the nodes but also in the whole carrier sensing
       •   For a transmission to happen, emitter and receiver both suppose that there is no jamming
           takes place during the transmission. Therefore, the available bandwidth’s value depends on a
           link which further depends on both peers’ respective channel utilization ratios and also on the
           inactive period’s synchronization.
       •   Collision detection is not possible in the wireless environment. So, whenever collision
           happens, both colliding frames are completely get emitted, and maximizing the bandwidth
           loss. It is thus essential to incorporate this bandwidth loss to the available bandwidth

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

       estimation. We estimate of the collision probability on each link. This estimation combines
       two approaches:
           1. A on line approach that calculates the contact of the medium occupancy distribution
               at the receiver side with help from the collision probability on Hello packets. Many ad
               hoc networks used these Hello packets in routing protocols and are needed for
               calculating the previous estimation E(b) on each link.
           2. A off line approach that uses the size of the packets sent by the source thanks to an
               interruption. The main aim of this last approach is to calculate the collision
               probability that packets of known and fix size will undergo on a link from the
               collision probability of Hello packets figured out from the real measurements on the
               same link. This collision probability estimation denotes p, in the following, and
               depends on the size of packets that will be sent.
   •   At last, when collisions take place on unicast frames, the IEEE 802.11 protocol retries to emit
       the same frame automatically, drawing the back-off counter in a double sized contention
       window. The available bandwidth has a great impact on the time lost in the extra overhead.

         We calculate the mean back-off which depends on p the collision probability calculated in the
previous estimation. It is then possible to assume the proportion of bandwidth consumed by the back-
off scheme. This proportion is represented by K in the following.
These diverse estimations are then mixed to approximate the available bandwidth on the wireless
link, i.e. between an emitter s and a receiver r:

        Delay tells us the time to send a packet from a source to a destination node. Opposite to
bandwidth, delay is an additive metric. Thus, the delay alongside a path is equivalent to the sum of
the delays on one-hop links of the same path. For this study, we suppose that the clocks of all the
mobiles are absolutely synchronized.
By using IEEE 802.11, the mean packet delay on a definite one-hop link, denoted by D, can be
alienated into three parts:

   •   The mean queuing delay which corresponds to the interval between the time the packet
       comes in the queue of the link’s emitter and when the packet becomes the head of line packet
       in this node’s queue. We represent it by .
   •   The mean contention delay is the period between the time the packet arrives at the head of
       line and the time when the packet is sent to the physical medium. We represent it by . This
       interval reveals that actually a node may contend to access to the channel due to other
       transmissions in its carrier sensing area.
   •   The mean transmission delay is defined as the time to transmit the whole packet which
       includes possible retransmissions in case of collisions. We represent it by

Thus, we have the relation that would be like,

In the remaining part of this section, we made some postulations in order to make things easier for
the analysis and to give an analytical expression for        and .

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

   A. Assumptions In System Implementation
       We represent an IEEE 802.11 node as a discrete time M/M/1/K queue. The properties of this
queue are:

   •   The packet arrival follows by an exponential law of parameter λ.
   •   The service rate also follows an exponential law of parameter µ.
   •   The size of the queue is restricted by the value K. When a new packet arrives and if there are
       already K packets in the queue, then this one is dropped.
   •   The queue is a standard FIFO (First in First Out).

       We suppose no use of RTS and CTS messages. The studycan be easily being extended for the
cases where such messages arethere. The parameter λ denotes the number of packetswhich arrives in
the queue per second and depends on theapplication throughput (if such an application presents on
thenode) and the traffic routed by this node. The service time µdenotes the number of packets
leaving the queue per second.

    B. General Idea
        Our early aim is to provide guaranteed delay to delaysensitive flows. For this, we need the
estimation of the meandelay that the packets of such a flow will attain beforetransmitting this flow.
Therefore, we need the estimation of theservice rate that can be offered to this flow on each of the
nodespassed through by the flow. It is also vital to be reminiscent thatthe reception of a new flow
may impact the delays of the existingflows. Our aim at that time was also to minimize such an
impactin order to get the guaranteed delay of existing delaysensitive flows. For the available
bandwidth estimation(see Section III), we may define the available service rateof a node as the rate
that can be proposed to a newflow without increasing the delay of any ongoing flow in thenetwork.In
a way to limit the impact on the mean delay of existingflows, a congestion control must be
comprehended. Thus, theservice rate that may be offered by a node to a new flow iscorrelated to the
residual bandwidth directly as seen by this node.
        This residual bandwidth is the same to the medium occupancy which isseen by this node
(including its own transmissions) multipliedby the capability of this node. This value confines the
effectthat after the queuing procedure, a packet which comes at the head of line of the MAC layer
should remain idle until the channel is free to gain the access. More accurately, we model         the
service rate that can be offered to a new flow, as the available bandwidth computed by the node
which is rescaled in packets per second.

    C. Estimating The Mean Queuing And The Contention Delay
        In this section, we evaluate + . When µ > λ, the service rate of the node is greater than
the arriving process and the queue will not increase which involves a queuing and a contention delay
which are void

Whenever        .

Let’s suppose the probability to have n packets in the queue is denoted by     where

The transmitted packet approaches with a rate    and exits with rate .

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME


Now implementing

The sum of the probabilities being equal to 1, the value       can be simply presented as a function
of variables p and q.

Thus, the average number of packets Q can be presented as follows:

Now, implementing queuing principle and considering little’s law, it can be found that the parameter
        isequal to the mean waiting time and it would be presented as

Finally, it can be presented as follows

Here, it can also be found that, since the queue size is bounded,        is bounded by a maximum
value Dmax.


             , therefore

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

        In this estimation, the contention delay considers only the time spent until the medium is free
to gain the access to the radio medium to send the packet for the very first time. We do not consider
the time that can be required to retransmit the packet. This time is incorporated in our transmission
delay, is described in the next sub-section.

    D. Estimating the mean transmission delay
         The mean transmission delay is defined as the time to transmit thewhole packet. When this
operation issuccessful, in IEEE 802.11 DCF a positive acknowledgement is sent back to theemitter.
Still, there is a chance, even for a single framethat when a packet is emitted, then the medium is not
inactive atthe receiver’s side, infuriating a collision. These collisionsinclude retransmission of the
same packet and raise thecontention window size, all these phenomenon resulting in anraise of the
mean transmission delay.

    1) Modeling the exponential back-off mechanism
       It is considered that a random wireless link suffers from collision with a probability p. We
consider transmission is successful at the first attempt for every frame, with probability       . It
again succeeds in the second attempt with probability p*(1-p). After C unsuccessful attempts, C
which depends on the frame size, the IEEE 802.11 standard specifications that the frame should be
       If we represent the random variable representing the number of efforts shown for the correct
transmission of a given frame by X, we have:

The n number of retransmission which can be expected as follows:


        Now, we require calculating the expected back-off that impacts the delay transmission. First
we assume that there is no collision takes place, and then the back-off is drawn according to a
uniform law in the interval                          which is being determined by the specification of
the MAC protocol. On a large surveillance window, the back-off can be calculated by its average
value        . When the collisions take place, the exponential back-off scheme is triggered. After the
unsuccessful transmissions, thecontention window size doubled up to a maximum valuerepresented
by        . In these circumstances, the average back-off value go way above          and it is essential
to model the time used by the exponential back-off process.
        The expected number of back-off slots decreased until the end of transmission attempts for a
single frame can be represented as:

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

To simplify the expression. Let us suppose that

   2) Mean Transmission Delay Computation
       The different points stated above can be mixed to calculate the mean transmission delay on a
wireless link, i.e. through an emitter to a receiver. To summarize, the mean transmission delay
among two neighbor nodes can be calculated by the following formula:

where Tm represents the time to successfully transfer a whole packet of m bytes with IEEE 802.11,
T c represents the collision duration, where n is the mean number of retransmissions which depends
on collision probability, back off is the estimated number of back off slots and      represents the
duration of a slot. To sum up, the mean delay of a one-hop link contains:

      •   The mean delay occurred by a packet on the link’s emitter is          It communicates to
          the waiting time before the first transmission of the packet.
      •   The mean delay occurred by a packet during the transmission is            It involves the
          potential retransmissions provoked by collisions.


        As our aim is to ensure delay for delay sensitive flows, we incorporate the previous
assessment technique of the mean delay into a protocol, and this protocol is called             for
Delay Estimation in Ad hoc Networks. The protocol part, i.e. the setting up and upholding of
reservations, does not contain any new or specific feature. It is based on the broadcasted route
request messages, access control at each transitional node and plain reservation by an uncast route
reply message issued by the destination. Our delay evaluation requires available bandwidth
estimation. We use the method designed in the protocol ABE, since the results we get show a high
level of precision.
        The congestion control method required to minimize the impact on the delays of existing
flows will be based on the available bandwidth estimation of      and then will be performed via an
admission control on bandwidth.

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

A. Delay Estimation in proposed system
        In          , delay information is exchanged by neighboring nodes through Hello messages.
Every      seconds, each node nearby estimates its medium occupancy ratio and comprises this
information in a Hello packet, then the medium occupancy permits the                 to estimate the
available service time µ
        Hello-based methods generate additional expenditure responding on the Hello emission
frequency. If possible, the frequency of Hello packets should be modified to the nodes mobility
and/or to the flows dynamics. The larger , the more constant the measurements will be, hiding the
fast deviations in the medium load. However, should also be small enough to permit fast reactions
to long-term load deviation and to nodes mobility. In this protocol we decide, in order to have
significant comparisons, to fix the value of = 1 second. Similarly, all compared protocols are tuned
consequently to emit one information frame each second.

B. Admission control and          routing in
       The             routing protocol is based on the cross-layer routing protocol. The      layer of
each node approximates proactively and periodically the mean delay of the neighboring links and
makes the routing layer in charge of discovering QoS routes fulfilling the applications demands,
basing their results on the       layer. We try to offer routes for which the end-to-end delay denoted
by the application level is greater than the mean value estimated along the path. Let’s take a path
composed of hops. The delay limitation can be expressed by the following inequality:

       Where        the end-to-end delay is denoted by the application level            is the one-
hop delay between transitional neighbor nodes i and i + 1 on the path                 .The routing
process of DEAN is strongly encouraged by         and contains of two major parts: route discovery
and route maintenance.

Route discovery: The main aim of the route discovery method is to get a route between the sender
and the receiver that meets delay limitations specified by the application level. Therefore, two flows
with the same source and destination can track different routes which depend on the network state.
          performs an on-demand route discovery like in         . Whenever a source node has to send
data, it broadcasts a route request              to its neighbors. The            packet includes the
bandwidth and delay requirements at the application level, the destination address, a sequence
number, the address of the sender, and the cumulative delay calculated along the path. Each mobile
that receives such a RREQ executes three admission controls:

•   Delay calculated corresponds to the sum of the growing delay given in the             packet and the
    predicted delay on the link from which the            packet is received. To estimate this latter, we
    use λ the throughput requirement of the function and the service rate that can be presented to
    the application by the link’s emitter.
•   The service rate which will serve the application on this node corresponds to min (λ, µres). ]]
•    The second one makes sure that throughput of the flow to be emitted (figured out from the
    service rate calculated at the previous step) will not be decreased by close flows.
•   The third one makes sure that the release of this flow on this link will not decrease the throughput
    of close flows which are in hidden nodes arrangement. These two admission controls are
    executed in ABE and we re-use them in the               .
International Journal of Computer Engineering and Technology (IJCET), ISSN 0976      0976-
                                                         July August
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

Figure 1 presents the cross-layer scheme and the different admission controls performed in the

                         Fig. 1. Cross-layer and admission controls in

        The node adds its own address and its mean delay to the cumulative delay of the route if all
these controls are positive, and then forwards the                  ; otherwise it discards the message
automatically. When the destination will receive a first              , it will send a uncast route reply
(       ) to the initiator of the request along the reverse path finally. The resources in terms of service
rate proposed to this flow at each node are then reserved so that the new          flow can be sent.

Route maintenance: A route maintenance process is necessary, particularly in case of mobility. We
employed a simple detection and reaction scheme. The detection a broken route by our proposed
          is done by monitoring the Hello messages. If Hello packet is not received by a node from a
neighbor within a particular time interval (equal to the time which takes place to transmit 3 Hello
packets in the assessment part), or if one of its link fails to meet the reserved delay any more, then it
sends a route error (        ) to the source which later rebuilds its route.
       It is very much motivating to note that proposed                 , as it is described here it does not
only guarantees the mean delay to applications but with slight amendments on the admission control
phases, it is likely to guarantee both throughput and delay requirements.


        In this research work robust system architecture for estimating end to end delay in multihop
wireless Adhoc network has been developed. The ultimate goal of this work is to achieve an
optimized solution for         in Adhoc communication network. The overall system has been
developed for two kinds of system architectures one is for centralized network whil another was
developed for decentralized type of wireless Adhoc networks. The simulation framework was
established and developed with framework and with C Sharp language. The simulation
results were obtained for average packet end to end delay in wireless Adhoc network (for both
centralized as well as decentralized), overall network end to end delay, processing delay, packet drop
rate and packet success rate.In simulation different network size like with 500 nodes, 750, 1000,
1250 nodes and 1500 nodes have been considered for simulation.
        In order to compare the different protocols and illustrate the effectiveness of the proposed
system so as to provide end-to-end delay guarantees, here in these work random topologies have
been generated with random constant bit rate flows (random source, random destination and random

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

throughput with fixed 1000 bytes frames). For each of these protocols, similar scenarios (varied
nodes and number of flows) lead to similar behaviors. Therefore, for this section, we give the results
of one scenario and the presented results are obtained over 30 simulation runs with different random
seeds. Few of the results obtained have been presented as below:
        The below mentioned graph the network end to end delay has been presented for different
network sizes. The graph depicts that in decentralized network, the end to end delay is lower as
compared to centralized network and even in case of decentralized network the delay observed is
constant while in case of centralized network, it is increasing proportionately.

                                                              NE T WO RK EN D TO E ND DE LAY O BSE RV ED

                                                 900                DECENTRALIZED TOPOLOGY            CENTRALIZED TOPOLOGY

                                                        500              750              1000            1250               1500

                                                                                     TOPOLOGY SIZE

                                                               Figure 2. Network End to End Delay observed

        The below mentioned figure (Figure 3), depicts the average end to end delay observed in
centralized and decentralized network. Here we can find that the average packet end to end delay is
higher as compared to centralized because of the route delay introduced due to decentralized

                                                        AV ERAG E PACK E T E ND TO EN D D ELAY O BS ERV E D

                                                  8.9                DECENTRALIZED TOPOLOGY           CENTRALIZED TOPOLOGY
                                                        500               750                1000          1250              1500

                                                                                      TOPOLOGY SIZE

                                                                Figure 3: Average packet End to End Delay

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

                                          PACK E T D RO P RAT I O O BSE RV ED
                                            DECENTRALIZED TOPOLOGY             CENTRALIZED TOPOLOGY





                                    500           750                1000           1250              1500
                                                              TOPOLOGY SIZE

                                           Figure 4: Packet drop ratio observed

        The above mentioned graph illustrates the packet drop ratio in centralized as well as
decentralized network environment. Here it can be found that in case of decentralized (Proposed
network) the packet loss is much higher as compared to existing or centralized network. This
signifies that in this case when there is higher success rate, there is no requirement of retransmission
and hence the delay caused will be much smaller as compared to lossy network.

                                          PAC K ET SU CESS RAT I O O BS ERV E D

                          0.99986          DECENTRALIZED TOPOLOGY             CENTRALIZED TOPOLOGY

                                    500           750                1000           1250              1500

                                                              TOPOLOGY SIZE

                                              Figure 5: Packet success Ratio

       Figure 5 represents the packet success ratio in developed system architecture with the
network size of varied dimension. Here we can find that in case of decentralized topology packet
success ratio is higher and therefore, the retransmission is negligible. The higher success rate make
the system capable of transmitting packets in least or minimum time and thus enhancing the
throughput or of course higher QoS.

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

                                                       PACKET PROCESSING DELAY MEASURED

                                                          DECENTRALIZED TOPOLOGY            CENTRALIZED TOPOLOGY
                                                 500            750             1000             1250              1500

                                                                            TOPOLOGY SIZE

                                                        Figure 6: Packet processing Measured

        The above mentioned graph illustrates the processing delay caused in different network
topology. Here, it can be noticed that in case of centralized network the packet processing delay is
higher. On the other hand the decentralized topology exhibits comparatively less processing delay
over a wide range of network size. Thus, considering these obtained factors and results, it can be
stated that the developed system architecture has exhibited a tremendous performance in optimizing
     in wireless Adhoc network. Thus if fulfills most of aspects for QoS optimization and enhanced
resource estimation and utilization.


        The high paced increase in wireless communication demand has ignited a revolution for
optimization in optimal resource utilization and hence QoS optimization of communication
protocols. On the other hand, the effective resource utilization becomes very critical factor in
networks of decentralized nature in Adhoc 802.11 Protocols. An enhanced approach for end-to-end
delay estimation and hence the effective available bandwidth utilization might be an effective
solution for       optimization in multihop wireless Adhoc network. In order to enhance the
throughput as well as to optimize resource utilization in Adhoc network, here in this research paper,
a noble approach called “End to End delay Estimation                in wireless Adhoc network” has
been developed. This developed approach               for     optimization in decentralized networks
might be considered as a hybrid technique of               queuing, admission control, QoS routing,
route discovery and maintenance optimization and available bandwidth estimation schemes. The
incorporation of these techniques has been made only for achieving the goal of enhanced end to end
delay estimation in IEEE 802.11 Adhoc network protocol. Initially, the           C queuing approach
has been used for queuing of nodes incorporating the network and that has been followed by
available bandwidth estimation         step measuring the end to end delay and available bandwidth.
At       layer the admission control and QoS routing approach has been enhanced to deliver the best
outputs. The presented                 approach has also enhanced the route discovery and its
maintenance, thus coming up with higher throughput and ultimately with enhanced QoS of
considered IEEE 802.11 Adhoc network protocol. The simulation framework has been designed of

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

varied network size (500 to 1500 nodes) for centralized Adhoc as well as our proposed decentralized
wireless Adhoc network. The proposed                  technique has depicted higher efficiency in
proposed decentralized wireless Adhoc network as compared to centralized in terms of lower end to
end delay, packet drop ratio and overall packet processing time. On the other hand the proposed
         approach in decentralized wireless Adhoc network has depicted higher packet success rate
with minimum maintenance packets and thus facilitating higher network throughput. The overall
result analysis proves that the developed             technique may play a vital role in available
bandwidth estimation, optimum resource utilization and QoS optimization in IEEE 802.11 Adhoc
decentralized network protocol.


 [1]    IEEE Standard for Information Technology Telecommunications and Information Exchange
        between Systems. Local and Metropolitan Area Network – Specific Requirements – Part 11:
        Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications,
 [2]    C. Chaudet, D. Dhoutaut, and I. Gu´erinLassous. “Performance issues with IEEE 802.11 in
        Ad Hoc Networking”. IEEE Communication Magazine, 43(7), July 2005.
 [3]    Y. Yang and R. Kravets. Contention Aware Admission Control for Ad Hoc Networks. IEEE
        Transactions on Mobile Computing, 4(4):363–377, 2005.
 [4]    C. Chaudet and I. Gu´erinLassous. BRuIT - Bandwidth Reservation under Interferences
        influence. In Proceedings of European Wireless 2002 (EW2002), Florence, Italy, Feb 2002.
 [5]    H. Badis and K. Al Agha. QOLSR, QoS routing for Ad Hoc Wireless Networks Using
        OLSR. European Transactions on Telecommunications, 15(4), 2005.
 [6]    C. Sarr, C. Chaudet, G. Chelius, and I. Gu´erinLassous. Improving Accuracy in Available
        Bandwidth Estimation for 802.11-based Ad Hoc Networks. Technical Report 1, INRIA, June
 [7]    Y. Yang and R. Kravets. Achieving Delay Guarantee in Ad Hoc Networks by Adapting IEEE
        802.11 Contention Windows. In Infocom, Anchorage, USA, May 2007.
 [8]    C. Sarr, C. Chaudet, G. Chelius, and I. Gu´erinLassous. Improving Accuracy in Available
        Bandwidth Estimation for 802.11-based Ad Hoc Networks. Technical Report 1, INRIA, June
 [9]    Chaudhary, D.D.; Waghmare, L.M. “Quality of service analysis in wireless sensor network
        by controlling end-to-end delay” Industrial Electronics and Applications (ICIEA), 2012 7th
        IEEE Conference on Digital Object Identifier: 10.1109/ICIEA.2012.6360816 Publication
        Year: 2012, Page(s): 703 – 708
 [10]   Jae-Ho Lee; KyeongHur; Doo-SeopEom “WTE-MAC: Wakeup time estimation MAC for
        improving end-to-end delay performance In WSN” MILITARY COMMUNICATIONS
        CONFERENCE,2011              MILCOM         2011        Digital      Object      Identifier:
        10.1109/MILCOM.2011.6127793 Publication Year: 2011, Page(s): 902 – 907.
 [11]   R.S.; Fohler, G. “Probabilistic estimation of end-to-end path latency in Wireless Sensor
        Networks Oliver” Mobile Adhoc and Sensor Systems, 2009 MASS '09 IEEE 6th
        International Conference on Digital Object Identifier: 10.1109/MOBHOC.2009.5336970
        Publication Year: 2009 Page(s): 423 – 431
 [12]   Baoliang Li; Jie Zhao; Junhui Wang; Wenhua Dou Semantics “A Max-Plus Algebra
        Approach for Network-on-Chip End-to-End Delay Estimation Knowledge and Grids (SKG)”,
        2012 Eighth International Conference on Digital Object Identifier: 10.1109/SKG.2012.6
        Publication Year: 2012 , Page(s): 217 – 220

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-
6367(Print), ISSN 0976 – 6375(Online) Volume 4, Issue 4, July-August (2013), © IAEME

 [13] Rodoplu, V.; Vadvalkar, S.; Gohari, A.A.; Shynk, J.J. “Empirical Modeling and Estimation
      of End-to-End VoIP Delay over Mobile Multi-Hop Wireless Networks” Global
      Telecommunications Conference (GLOBECOM 2010), 2010 IEEE Digital Object Identifier:
      10.1109/GLOCOM.2010.5684031 Publication Year: 2010, Page(s): 1 – 6.
 [14] Kataria, D.; Logothetis, D.; Elwaid, A. “An enhanced method for the estimation of end-to-
      end cell delay variation for real-time services” Global Telecommunications Conference,
      1999.      GLOBECOM           '99     Volume:      2     Digital     Object      Identifier:
      10.1109/GLOCOM.1999.829998 Publication Year: 1999, Page(s): 1367 - 1372 vol.2
 [15] Dong Linfang ; ShuYantai ; Chen Haiming ; Ma Maode, “Estimation and application of end-
      to-end delay under unsaturation traffic in wireless ad hoc networks Mobile Technology”,
      Applications and Systems, 2005 2nd International Conference on Digital Object Identifier:
      10.1109/MTAS.2005.207203 Publication Year: 2005 , Page(s): 6 pp. – 6.
 [16] Matta, J.M.; Takeshita, A. “End-to-end voice over IP quality of service estimation through
      router queuing delay monitoring Global Telecommunications Conference, 2002.
      GLOBECOM            '02.    IEEE      Volume:       3    Digital     Object      Identifier:
      10.1109/GLOCOM.2002.1189072 Publication Year: 2002, Page(s): 2458 - 2462 vol.3
 [17] Despaux, F.; Ye-Qiong Song; Lahmadi, A. “Combining Analytical and Simulation
      Approaches for Estimating End-to-End Delay in Multi-hop Wireless Networks” Distributed
      Computing in Sensor Systems (DCOSS), 2012 IEEE 8th International Conference on Digital
      Object Identifier: 10.1109/DCOSS.2012.31 publication Year: 2012 , Page(s): 317 – 322.
 [18] Kapil K Shukla, Kaushik I Manavadariya and Deven J Patel, “Comparative Study of
      Bluetooth, 802.11 and Hiperlan”, International Journal of Computer Engineering &
      Technology (IJCET), Volume 4, Issue 3, 2013, pp. 455 - 463, ISSN Print: 0976 – 6367,
      ISSN Online: 0976 – 6375.
 [19] Venkatesh Kumar.P, Vallikannu A.L and Kavitha B.C, “Effective Broadcasting in Mobile Ad
      Hoc Networks using Grid Based Mechanism”, International Journal of Computer
      Engineering & Technology (IJCET), Volume 2, Issue 1, 2011, pp. 39 - 46, ISSN Print:
      0976 – 6367, ISSN Online: 0976 – 6375.
 [20] T.Priyadarsini, B.Arunkumar, K.Sathish and V.Karthika, “Traffic Information Dissemination
      in Vanet using Ieee-802.11”, International Journal of Electronics and Communication
      Engineering & Technology (IJECET), Volume 4, Issue 1, 2013, pp. 294 - 303, ISSN Print:
      0976- 6464, ISSN Online: 0976 –6472.


Shared By: