MedicineLakex

Quality of service aware mac based on ieee 802.11 for multihop ad-hoc networks - wireless communications and networking conference, 2004. wcnc. 2004 ieee

Quality of Service Aware MAC Based on IEEE 802.11 for Multihop Ad-Hoc Networks Center for Communication Systems Research Center for Communication Systems Research University of Surrey, Guildford GU2 7XH University of Surrey, Guildford GU2 7XH G.Pavlou @eim.surrey.ac.uk Abstract—Real-time multimedia applications necessitate collisions, unbounded delay, and increased jitter. The time
predictable network resources. Quality of Service (QoS) support
required to resolve collisions is a function of the network load. for such applications in mobile ad hoc networks (MANETs)
In addition, the DCF makes extensive use of control packets as requires acceptable channel conditions, QoS-aware mechanisms
a handshaking mechanism in order to minimize hidden-node for channel access, identification of proper forwarding (transit)
and exposed-terminal problems [2]. This approach is not nodes, as well as measures for congestion prevention and
desirable, especially for periodic time-sensitive traffic, as it not management in those nodes. This paper proposes a new QoS-
only increases the collision rate, but also deteriorates the aware medium access control (MAC) protocol that takes the
overall efficiency of the channel and the system [7][8][9]. above requirements into consideration. This novel protocol is
based on the legacy IEEE 802.11, and thus can be easily
Besides the DCF, the IEEE 802.11 also incorporates an integrated into existing systems without much difficulty. alternative access method known as the point coordination
Simulation results confirm that our approach results in improved
function (PCF) [4][5][6]. This access method is similar to a throughput for real-time periodic traffic, while providing
polling system, and uses a point coordinator (PC) to determine deterministic delay performance.
which station has the right to transmit. The PCF falls under demand assignment access schemes, and as such it is more Keywords – Ad hoc Networks, Multiple Access Control, QoS-
suitable for an environment that requires QoS guarantees [16]. aware MAC, IEEE 802.11.
The PCF operation, however, needs a centralized node such as an access point (AP), and hence is normally used in WLAN environments. In our approach, spread spectrum techniques Given that real-time applications will be used in ad hoc and collision avoidance multiple access protocols are combined networks, efforts for QoS support are under way. However, in to form a new MAC protocol for multimedia traffic over order to facilitate QoS support, a clear understanding of the MANETs [2][10]. This protocol is based on a hierarchical difficulties and issues in provisioning QoS in MANETs is approach consisting of two sub layers. The lower sub-layer of necessary. Since ad hoc networks lack fixed infrastructure, the protocol provides a fundamental access method using the there is no dedicated agency to manage the channel resources DCF to support asynchronous data traffic, and to enable time- for the network nodes. Quality of service is possible only if sensitive traffic to reserve bandwidth using a two-way supported by the underlying medium access technology. In handshake mechanism. The upper sub-layer is designed to other words, the network-level QoS mechanisms cannot work support real-time periodic traffic. Our novel smart MAC, in MANETs, unless the MAC ensures orderly access to the which thus consists of both random (contention-based) and shared wireless medium, playing a crucial role in the efficient regulated (contention-free) access to the medium, provides and fair sharing of the scarce wireless bandwidth [1]. The applications with enough resources in order to improve QoS. nature of the wireless channel requires that different layers, in The rest of the paper is organized as follows. Section II particular the network-layer and MAC sub-layer, interact reviews IEEE 802.11 and related work on MAC-level QoS, constantly in order to provide an overall QoS. Also, there must and presents our motivation. The proposed MAC protocol is be mechanisms available to minimize or recover efficiently described in section III. Section IV presents the evaluation of from packet collisions. However, most of the network-layer the proposed scheme through simulation, and demonstrates that QoS work is tailored to the distributed coordination function our MAC approach leads to improved-QoS performances. (DCF) of IEEE 802.11a/b as the underlying MAC [11]. In the Section V presents our conclusion and future work. legacy IEEE 802.11, an ad hoc network is named Independent Basic Service Set (IBSS) [6]. An IBSS is based on the DCF that utilizes a random access method of carrier sense multiple II. PREVIOUS WORK AND OUR MOTIVATION access with collision avoidance (CSMA-CA). Since the latter is Given that our approach is based on IEEE 802.11, a basic mainly meant for best-effort traffic, the present DCF-based description of its working mechanism is necessary here. Since MANET cannot support QoS at MAC-level, and subsequently the main focus of our work on the MAC sublayer is on PCF, overall end-to-end QoS guarantees [3][7][8][9]. The contention we review the IEEE 802.11with an emphasis on the PC mode. from multiple users to access the common medium using a random access technique often results in unavoidable packet WCNC 2004 / IEEE Communications Society 0-7803-8344-3/04/$20.00 2004 IEEE A. IEEE 802.11 mind. MAC protocols can be classified based on the mode of operation into random access, guaranteed access, and hybrid access protocols [16]. Random access schemes are typically used for data traffic, and cannot support QoS. Guaranteed and hybrid access schemes normally require central nodes. Most of the works on QoS-enabled MAC have been based on guaranteed and hybrid access schemes, hence targeting infrastructure-based networks such as WLANs. Figure 1. Timing Diagram for IEEE 802.11 MAC Operation. A multiple access scheme based on Time Division The DCF mode is the fundamental access method of the Duplexing (TDD) for single hop MANET is proposed in [7]. In 802.11 MAC [6]. The time-period during which the network this approach, the channel is time-slotted, and a slotted system operates in the DCF mode is known as the contention-period requires network-wide time synchronization, which is (CP). Access priority to the medium is controlled through the relatively easy to achieve in infrastructure-based networks by use of inter frame spaces (IFS). There exit four types of IFS: using the base station as a time reference. This task becomes the Short IFS (SIFS), the Point coordination function IFS extremely difficult in distributed networks such as multihop (PIFS), the Distributed coordination function IFS (DIFS), and MANET environments [16]. Also this work considers a the Extended IFS (EIFS). The SIFS is the shortest interval, and network, where all the nodes are assumed to be within radio is used for transmission of acknowledgements (ACK), station range of each other, and only a limited number (maximum 12) responding to polls from the PC, and between fragments. As of multimedia sessions can be supported at a particular moment such transmissions that are required to wait only for SIFS have within the considered network. This is unrealistic, as MANETs the highest priority over the medium. The AP uses PIFS (> tend to be multihop, and should support as many sessions as SIFS) to initiate the CFP. The DIFS (> PIFS) is used by possible. This scheme further requires that a source, after ordinary nodes during the CP. The shorter the period that a having successfully reserved a time-slot, send a busy-indication transmission has to wait for, the greater the access priority it packet until the end of its session. This approaches increases has over the medium. The DCF mode consists of a four way the number of exposed nodes. Another similar scheme, known exchange: request-to-send (RTS) - clear-to-send (CTS)- as Soft Reservation Multiple Access with Priority Assignment DATA-ACK. RTS is used for a node to acquire the medium (SRMA-PA), is presented in [8]. It is a Time Division Multiple after waiting for a minimum period of DIFS. The receiving Access (TDMA) frame based MAC protocol that allocates node (destination) responds with CTS after a SIFS, indicating stations to different time-slots. This scheme does not take that it is ready to receive data. The sender then completes the asynchronous data traffic into consideration, as all data packet transmission. On the other hand, in case the sender transmissions are required to reserve slots irrespective of cannot access the medium after DIFS due to the medium not whether they are real-time or best-effort traffic. Also there is a being idle, the transmission is deferred until the end of the possibility for higher priority traffic to starve lower priority current transmission. A random interval in the range of zero to traffic, as any higher order traffic can snap the slots already Contention Window (CW) is then computed by the node to reserved by lower priority traffics. A MAC approach that initialize its backoff timer. In addition to physical medium combines an allocation-based (TDMA) protocol and a sensing, virtual medium sensing is achieved by using time contention-based (CSMA-CA) protocol is proposed in [9]. In fields in the packets, which indicate to the other nodes the this scheme, the number of slots in each frame is dependent on duration of the current transmission. the number of nodes in the network, and hence each slot belongs to a single node only. The higher the number of nodes The PCF mode provides contention-free frame transfer and in the network, the larger the frame size would be. This leads to the time-period in which the LAN operates in the PCF mode is unbounded delay for time-sensitive applications. Similar known as the Contention-Free Period (CFP) [6]. The AP approach is followed in reservation CSMA-CA [3]. In this performs the function of the PC by gaining control of the scheme, CP and CFP alternate, and the CFP is based on medium in the beginning of the CFP, after sensing the medium TDMA. Since there is no node to regulate the common to be idle for PIFS. During the CFP, nodes that are CF-Pollable medium, this scheme may lead to a "stretching" problem [4]. It are polled by the AP. On receiving a poll, a node transmits its also requires proper time-synchronization, and each node is data after a SIFS. In order to poll the nodes, an AP must supposed to maintain a "slot-table" that indicates whether each maintain a polling-list. The CFP must alternate with the CP. slot is "reserved" or "available". Another MAC protocol that The sum of the two periods is called the "super-frame" and is considers multiple channels is proposed in [10]. It combines shown in Fig.1. The AP initiates the CFP by transmitting a code division multiple access (CDMA) or frequency division Beacon frame, and ends it by transmitting a CF-End frame. The multiple access (FDMA), and TDMA to create a contention- contention-free repetition interval (CFPPeriod) is the reciprocal free MAC, termed the sequenced neighbor double reservation of the rate at which the AP initiates the CFP. To support error (SNDR). Since it mainly considers time-slot allocation to make correction, positive ACKs are used in both the DCF and PCF it contention-free, it fails to support asynchronous data traffic and requires complex slot-synchronization. B. Related work on MAC-LevelQoS forMANET Although the IEEE 802.11 DCF is meant for best-effort traffic, there have been some efforts that investigate It is difficult to compare different MAC protocols. Each has differentiated services at MAC-level in infrastructure-based been developed with a different architecture and application in WCNC 2004 / IEEE Communications Society 0-7803-8344-3/04/$20.00 2004 IEEE networks [11][12]. A similar idea can also be applied to MANETs. Service-differentiation is achieved by setting different values for CW – values of minimum (CWmin) and max) – for different traffic classes. Two different service classes such as high priority and best effort are considered, and the traffic packet with the smaller value of CW is more likely to be transmitted first [11]. There is, however, no explicit guarantee of the level of service differentiation. There have also been some proposals to make DCF to be "per-stream- fair", as the DCF of legacy IEEE 802.11 tends to be unfair due to the "capture-effect" [13]. With these schemes, different sessions are allowed to gain access to the shared wireless Point Coordination Function medium equally. Fairness is achieved by dynamically modifying the CW of each traffic type by the source. The fairness approach does not, however, guarantee QoS support. Distributed Coordination Similarly each work presented above has its own drawback(s), and does not have the capability to provide MAC- level QoS for multimedia traffic in multihop MANETs. The Unique Receiver-Based Code next section describes how our approach tries to achieve improved throughput for real-time periodic traffic, while Figure 2. Structure of Our QoS-Aware MAC. providing deterministic delay performance. III. PROPOSED APPROACH Having taken the common deficiencies of other approaches into consideration, our approach tries to support both asynchronous and time-sensitive multimedia traffic based on a hierarchical approach. In our scheme, both the DCF and PCF of the IEEE 802.11 are used after being modified to accommodate MAC-level service differentiation. Although the PCF does require a centralized node, we describe next how this can be introduced for the first time in multihop MANET with novelty. The motivation for this work comes from the observation that the PCF mode offers a "packet-switched connection-oriented" service which is well suited for voice as well as multimedia traffic. The "connection-oriented" aspect of the PCF mode would allow the network to provide namely throughput, delay, and jitter guarantees [4]. In order to accommodate simultaneous transmission of several data traffic, multiple parallel media (channels) are created with receiver-based spread-spectrum technology [2][10]. In this scheme, each node has its unique code, and hence its unique medium, on which it has to receive packets Figure 3. Working Mechanism of Our MAC. from others. In addition, there is a common medium, which all nodes can use to disseminate and acquire neighbor and routing In our work we have considered two different service related information. These codes are assumed to be orthogonal classes, high-priority and best-effort. Our MAC's mode of to each other, and assigned to nodes dynamically in a conflict- operation on each unique medium switches between pure DCF free manner using the common medium. In our approach, mode and combined (DCF + PCF) mode depending on traffic transmission by any node A to another node B has to be on the types, and hence adapts. Each unique medium supports only receiver's (B's) spreading code. In order to accommodate the pure DCF mode of operation as long as all traffic types are situation in which any node can receive multiple transmissions best-effort. On the other hand, whenever a high-priority traffic initiated by different sources, IEEE 802.11 (both DCF and needs to be transmitted, the source node A has to send an PCF) is used on top of each unique medium as depicted on Fig. "Association Request" (AReq) frame to the forwarder (transit) 2. The common medium, however, can support only the DCF. node B selected by the routing protocol [6]. This AReq frame Each node is expected to regulate and schedule its own unique is normally sent on the CP of a transit node's (B's) own unique medium. Also each node has to maintain constant CP and CFP medium (see Fig. 3). As soon as node B receives the Areq on its own medium in order to minimize or completely avoid frame, it has to send Association Response (ARes) frame to the the "stretching" effect [4]. This is important in order to originating node A on the CP of the latter's (A's) own medium. minimize the delay jitter experienced by applications. At the same time, node B has to create a polling-list and include node A in it. At the start of CFP on B's medium, node WCNC 2004 / IEEE Communications Society 0-7803-8344-3/04/$20.00 2004 IEEE B has to begin polling node A. In this way, any node (B) best-effort data traffic share the CP of a unique medium. Since should be able to emulate the functionality of PC, and in our the RIFS is shorter than the DIFS, the high-priority traffic class approach such a node is referred to as a virtual PC (VPC). has priority over the best-effort traffic, which uses DIFS. ARes Since node B is an intermediate node, it has to forward the frame is sent by any VPC, after SIFS when operating in the packet to its destination or the next forwarding node. DCF mode. ARes frame is sent only when admission control Accordingly, it would soon send the AReq frame to node C, module has analyzed the current load as expressed by equation which is here assumed to be the destination, on C's own (2) below. Since there is a maximum limit for high-priority medium. After sending ARes frame on node B's medium, node traffic, the probability for AReq frames of high-priority traffic C has to be ready to poll node B at the start of CFP on C's to starve the best-effort traffic during CP is minimized. This unique medium. If node C were to send packets back to node achieves the "fairness" in our scheme [13]. When a collision A, then it would follow the same process as node A has happens in the reservation process, the back-off time is performed, but in the opposite direction. In this way, nodes calculated using the following modified equation [12]: along a particular path (or route) become polling-list members of each other. It is thus important that whenever a node (A) Back-off-time = [ p 2+i × rand ()]× Slot _ time (1) transmits to another node (B), it has to be on the latter's (B's) medium. If the traffic type is of high priority, the source (A) Where p is the priority-factor with p=2 for high-priority traffic can transmit when the node (B) polls. This is the case even and p=4 for best-effort traffic, i is the transmission attempt when a node transmits an ACK for the packets it receives number, and rand() is a random function with a uniform correctly on its own medium. If, however, the routing protocol distribution in [0,1]. This ensures that the high-priority traffic is unable to provide the next hop address (probably during the class still enjoys priority over best-effort traffic during the route discovery process in the case of an on-demand routing collision-resolution period [6][12]. Although this type of protocol), the proposed MAC will use the common channel for prioritization is an important enhancement, it is not enough to disseminating the packet. In all other circumstances, our MAC provide effective traffic protection and QoS guarantees. This is protocol tries to minimize flooding on the common medium, achieved with our polling-based scheme introduced in MANET unless it is required depending on relative velocities as in a novel way as described below. The maximum number (Np) explained later. Because of the way in which transmissions are of high-priority traffic that can be supported in CFP, given a performed, our approach can completely eliminate the constant super frame size TSF, is given by equation (2)[4]. In exposed-terminal problem. The hidden-terminal problem is this case, the high-priority traffic is assumed to be a time- minimized to a greater extend with a use of unique media, and sensitive periodic interactive voice service, which is generated with the adoption of DCF and especially PCF mode of using a constant bit rate (CBR) source for convenience. operations. In addition, each node maintains its own polling-list T − T − T dynamically in order to use it on its own medium. If a polling- ovhd (2) node finds that it has not received any transmission from one of its polling-list members for time period greater than POLLING_TIME_OUT, then that node address will be deleted Where Tcp, Tovhd, and Tv are duration of CP, overhead involved from the former's polling-list immediately. This is how a for beacon and CF_END transmissions, and time to send a "disassociation" process is performed in our scheme [6]. This voice packet generated over a TSF [4]. In other word, the VPC approach leads to efficient bandwidth management, and this can poll to a maximum of Np number of times (or nodes) occurs whenever nodes move out of each other's range or have within a CFP on its own medium. Depending on the intensity finished their transmission. With this approach, a source node of the high-priority traffic load, any node can request a VPC to does not have to predict and inform others as to how long its poll it for more than once within each super frame period (TSF) transmission is going to last, which is often difficult in practice. of VPC. The MAC-level QoS-mapping module of a particular node calculates the number of times it has to be polled by any Our QoS-aware MAC protocol has three components: VPC. This calculation is based on the bandwidth requested by admission control, QoS-mapping and resource reservation as the network-level QoS mechanism. Any node can inform any shown in Fig. 2. Provisioning of network resources uses two VPC as to how many times it has to be polled by that VPC techniques such as resource reservation in PCF mode, and prioritization in the DCF mode of operation, as explained SF of the latter through the AReq frame – the AReq frame format is modified to accommodate this in our below. The objective of a priority-based approach is to provide scheme. Whenever a VPC receives an AReq frame from its service-differentiation by allowing faster access to the medium neighbors, its admission control module will check whether its to traffic classes with higher priority [11][12]. Like in the IEEE CFP period is fully utilized (ie. whether the N 802.11 DCF, priority access to the wireless medium is been reached). If not, the VPC is required to send the ARes controlled through the use of an IFS. A new IFS termed frame, and allocates the required bandwidth (here allocation Reservation IFS (RIFS) is defined and its value is selected such means how many times the requesting node has to be polled that SIFS < PIFS < RIFS < DIFS. To initiate new data transmission, RIFS or DIFS is used to contend for access to the SF of a VPC). If the maximum number has already been reached, then the VPC should not respond to any medium depending on the traffic type. A high-priority real- AReq. In this case, the requesting node should look for another time (periodic) traffic uses RIFS before sending the AReq, appropriate forwarding node, after waiting for a period of while DIFS is used to gain access right for best-effort ASSO_PROC_THERESHOLD_TIME_OUT. asynchronous traffic as in the IEEE 802.11 DCF. In our approach, the transmissions of AReq and ARes frames, and WCNC 2004 / IEEE Communications Society 0-7803-8344-3/04/$20.00 2004 IEEE Since the common medium is necessary for every node to In our initial evaluation, we consider two performance disseminate and acquire routing related information and metrics: throughput and MAC delay. We performed our possibly to perform dynamic code assignment, each node may simulations using the GloMoSim [15] simulation package, in receive transmission from its neighbors. On the other hand, as which we implemented our MAC scheme, and compared it explained above, nodes are expected to use receiver-based against the DCF mode operation of the IEEE 802.11. Nodes' unique media for data transmission. If, however, a node movement was modeled by the random waypoint mobility decides that its relative velocity with respect to its neighbors model. Nodes move at a speed between 0 and 10 ms-1. The increases beyond a certain threshold within a short time-period, pause time takes a constant value of 30 seconds. Each run is then it cannot rely on PCF-based operation. Only in this executed for 300 seconds of simulation time, and models a circumstance, the node would use the common medium for network of 20 nodes placed randomly in a 500m X 500m area. data transmission. Each node calculates its relative velocity by Each node has a transmission range of 100m, and full duplex making power measurements from neighbors on its own operation is considered with two antennas per node (one for medium and the common medium as explained below. Under transmission and the other for reception). The propagation Friis' free space propagation model, the signal power detected, model is the free space model. The bandwidth is 2 Mbs-1, the say RxPr, at the receiving node is indicative of the distance data packet size is 512 bytes, and packets are sent at a rate of between the transmitting and receiving node pairs. Since it is 100 to 400 per second by each node. Other important very difficult to calculate the exact distance between two nodes simulation parameters are listed in Table 1. without wasting bandwidth, we try to use the MOBIC model Fig. 4 shows the total throughput as a function of offered that defines a relative mobility metric, M rel (Y ) , at a node X load for both our scheme and the DCF of IEEE 802.11. The with respect to node Y [14]: total throughput is defined here as the total number of packets actually delivered to their respective destinations within the whole network. From Fig. 4, it becomes obvious that our Y → X scheme leads to better throughput performance. The throughput  Rx PrY→X  of IEEE 802.11 continues to drop after a slight initial increase, due to increased collisions and the resulting binary exponential Every node X determines the above mobility metric for backoff (BEB) scheme. As it can be seen, the throughput in our each neighbor Y by making subsequent power measurements, scheme tends to increase and soon reaches a saturation point. given a constant transmission power. A negative value for This point is dependent on Np of (2), which again depends on M rel (Y ) indicates that nodes X and Y are moving away from the link bandwidth and the CFP repetition interval, which here each other, and a positive value indicates that they are moving takes the value of 70 milliseconds. In our simulation, for towards each other. For a node with m number of neighbors, convenience, a node determines the receiver-based code of its neighbor based on the latter's address. Fig. 5 depicts the each node X will have m such values for M . Each node X average MAC delay incurred for a high-priority packet in both determines the aggregate local mobility value by calculating schemes. The MAC delay of a node is the time between the the variance (with respect to zero) of the entire set of relative instant at which a packet comes to the head of the node's mobility samples M (Y ) , where Y transmission queue and the end of the packet transmission. As i is a neighbor of X: load increases, there would be increased contention, and hence M (Y ) m = E ( rel MAC delay tends to increase in any MAC scheme. However, i ]i= in our scheme this increase is only slight compared to the original DCF, and is dependent on the link bandwidth and the Each node X computes (3) and (4) in an attempt to calculate CFP repetition interval. Reducing the inter-poll period or T its relative velocity with respect to its neighbors. A low value can further reduce the delay in our scheme [4]. On the other for Mx indicates that node X is relatively less mobile with hand, in the DCF, the MAC delay tends to increase respect to its neighbors, while a higher value indicates that significantly with the number of sessions. This can be node X is highly mobile. Whenever Mx exceeds Mthreshold, node attributed to such factors as increased collision, and hence X has to rely on the common medium for data transmission. In this way, our QoS-aware MAC adapts depending on relative increased retransmission attempts and extended BEB delay. mobility information. IV. EVALUATION THROUGH SIMULATION IMPORTANT SIMULATION PARAMETERS Parameter Value
Duration of the Superframe (T Value of the CFP (Tcfp) 50,000 The SIFS interval The PIFS interval The RIFS interval Total Offered Load The DIFS interval Figure 4. Total Throughput as a function of Offered Load. WCNC 2004 / IEEE Communications Society 0-7803-8344-3/04/$20.00 2004 IEEE minimizes the need for re-transmissions. This fact will in turn conserve scarce resources such as battery power and bandwidth. As explained, the MAC functionality of a node is adaptive, depending on relative node velocities. Since this work is mainly based on the IEEE 802.11 standard, it can be relatively easily integrated into existing systems. We plan to extend this work in the future with the use of the recent IEEE 802.11e standard in order to support multiple traffic classes. [1] D.H.Cansever, A.M.Michelson, and A.H.Levesque, "Quality of Service Support in Mobile ad-hoc IP Networks", Proc. Military Average MAC Delay (milliseconds) Comm.unications Conf. (MILCOM1999), vol. 1, Nov. 1999, pp. 30 – 34. [2] M.Joa-Ng, and I-T.Lu, "Spread Spectrum Medium Access Protocol with Collision Avoidance in Mobile Ad-hoc Wireless Networks", Proc. 18th Annual Joint Conference of the IEEE Computer and Communications Number of Sessions Societies (INFOCOM'99), vol. 2, Mar. 1999, pp. 776 – 783. Figure 5. MAC Delay as a function of Number of Sessions. [3] I.Joe, and S.G.Batsell, "Reservation CSMA/CA for Multimedia Traffic over Mobile Ad-hoc Networks", Proc. Int'l Conf. On Communications. (ICC 2000), vol. 3, Jun. 2000, pp. 1714 – 1718. [4] M.Veeraraghavan, N.Cocker, and T.Moors, "Support of Voice Services in IEEE 802.11 Wireless LANs", Proc. 20th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 2001), vol. 1, Apr. 2001, pp. 488 – 497. [5] J-Y.Yeh, and C.Chen, "Support of Multimedia Services with the IEEE 802.11 MAC Protocol", Proc. Int'l Conf. on Communications. (ICC 2002), vol. 1, May. 2002, pp. 600 – 604. [6] IEEE Draft International Standards, "Part 11:Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications", ISO/IEC 8802-11, IEEE P802.11/ D10, Jan. 1999. [7] V.N.Muthiah, W.C.Wong, "A Speech-Optimised Multiple Access Scheme for a Mobile Ad Hoc Network", Proc. 1st Annual Workshop on Mobile and Ad Hoc Networking and Computing (MobiHOC 2000), Aug. 2000, pp. 127 – 128. [8] C.W.Ahn, C.G.Kang, and Y.Z.Cho, "Soft reservation multiple access
Figure 6. Normalized Throughput as a Function of Maximum Speed and with priority assignment (SRMA/PA): a novel MAC protocol for QoS- Number of Sessions. guaranteed integrated services in mobile ad-hoc networks", Proc. 52nd IEEE Vehicular Technology Conference (VTS-Fall 2000) , vol. 2, Sep. Fig. 6 shows the performance of our scheme for different 2000, pp. 942 – 947. mobile speeds. In this simulation, the normalized throughput [9] I.Chlamtac, A.Farago, A.D.Myers, V.R.Syrotiuk, and G.Zaruba, performance of our scheme is observed by varying both the "ADAPT: A Dynamically Self-Adjusting Media Access Control Protocol for Ad Hoc-Networks", Proc. Global Telecommunications maximum speed of each mobile node and the offered load. The Conference, 1999, GLOBECOM '99, vol. 1A, pp. 11 – 15. normalized throughput is defined here as the total number of [10] Z.Cai, and M.Lu, "SNDR: a new medium access control for multi- packets actually delivered to their respective destinations channel ad hoc networks", Proc. 51st IEEE Vehicular Technology divided by the total number of packets generated within the Conference (VTS-Spring 2000) , vol. 2, May. 2000, pp. 966 – 971. whole network. The minimum speed and the pause time of [11] M.Barry, A.T.Campbell, and A.Veres, "Distributed Control Algorithms each node are kept at constant values of 0 ms-1 and 30 seconds for Service Differentiation in Wireless Packet Networks", Proc. 20th respectively throughout this simulation run. As it can be seen, Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 2001), vol. 1, Apr. 2001, pp. 582 – 590. the throughput is affected by the speed. Also it can be noted [12] I.Aad, and C.Castelluccia, "Differentiation Mechanisms for IEEE that our scheme leads to better throughput and delay 802.11", Proc. 20th Annual Joint Conference of the IEEE Computer and performance, when nodes move as groups in the same Com. Societies (INFOCOM 2001), vol. 1, Apr. 2001, pp. 209 – 218. direction. This is the case in battlefields and other similar [13] B.Bensaou, Y.Wang, and C.C.Ko, "Fair Medium Access in 802.11 Based Wireless Ad Hoc Networks", Proc. 1st Annual Workshop on Mobile and Ad Hoc Networking and Computing (MobiHOC 2000), Aug. 2000, pp. 99 – 106. V. CONCLUSIONS AND FUTURE WORK [14] P.Basu, N.Khan, and T.D.C.Little, "A Mobility Based Metric for In this paper we presented a QoS-aware MAC protocol for Clustering in Mobile Ad Hoc Networks", Proc. Int'l Conf. on multimedia traffic in MANETs and evaluated its performance Distributed Computing System, Apr. 2001, pp. 413 – 418. through simulation. The proposed protocol introduces a packet [15] X.Zengu, R.Bagrodia, and M.Gerla, "GloMoSim: A Library for Parallel switching concept based on the PCF in multihop MANET in a Simulations of Large-scale Wireless Networks", Proceedings of the 12th Workshop on Parallel and Distributed Simulations, May 1998. novel way. Simulation results confirm the performance [16] A.Chandra, V.Gummalla, and J.O.Limb, "Wireless Medium Access (throughput, delay) improvements of our scheme. In addition, Control Protocols", IEEE Communications Surveys & Tutorials, Second our proposed approach leads to fewer collisions and hence Quarter 2000, vol. 3, no. 2. WCNC 2004 / IEEE Communications Society 0-7803-8344-3/04/$20.00 2004 IEEE

Source: https://www.ee.ucl.ac.uk/~gpavlou/Publications/Conference-papers/Siva-04a.pdf