Design of MAC protocols to support quality of service for multimedia traffic in wireless LANs

Since the introduction of 802.11 MAC protocol standards for wireless local area network (WLAN) in 1999 by the IEEE task force group, WLAN has experienced tremendous growth. This can be observed by the increasing number of “hotspots” in public places such as airports, train stations, shopping complexes and private organisation such as universities and hospitals, and also at many homes where the use of wireless routers is very common. As more users embrace WLAN technology, the demand for more advanced services such as multimedia is increasing. Unfortunately, the current technology is still not able to provide adequate QoS support as demanded by these services. There are three versions of 802.11 MAC protocol available in the market today: 802.11a, 802.11b and 802.11g; but none of these can adequately support QoS for real-time traffic. Some enhanced protocols that have been proposed are 802.11e and 802.11n. The former enhances QoS support to a certain degree, although it still faces many weaknesses such as suppression of the best-effort (BE) service and packet collisions, while the latter is expected to increase the throughput by increasing physical channel rate to up to 600 Mbit/s. In addition, 802.11n adopts the MAC protocol used in 802.11b and 802.11g, which is exposed to packets collision and delay due to contention backoff. The aim of this research is to design WLAN MAC protocol to support QoS for real-time traffic in multimedia applications. Three novel designs are developed in this research. The first is the Segregation Hybrid Coordination Function (SHCF) MAC protocol. This protocol is based on traffic segregation, where real-time traffic is segregated from non real-time traffic. This enables the real-time traffic to avoid competition from the non real-time traffic to access the channel and provides easy management to support QoS for real-time traffic. In the SHCF, we introduce two service intervals (SI), one each for voice and video traffic. Each service interval caters for specific requirements of these traffic types reducing the probability of voice traffic queue being empty when polled. We carry out analytical modelling and simulation using Markov Chain (MC) analysis and NS2 respectively and show that with the SHCF protocol, significant improvements in performance over 802.11b and 802.11e in QoS support for real-time traffic are achieved. Our second design is an admission control algorithm for Ad-hoc wireless networks. We introduce a new technique based on a self-restraining mechanism in every node to prevent the network from congestion. Nodes or node wishing to join the network monitor the collision rates for a duration of time, known as Pre-admission Monitoring Time (PRAM). If the rate of packet collisions is below the Collision Threshold Level (CTL), it joins the network by sending an admission request packet. After sending this packet, the node again monitors the collision rates for a duration known as Post- Admission Monitoring (PAM). If collision rates are higher than CTL during PAM, the node drops itself from the network and starts PRAM again. We carry out analysis and investigate the behaviour of packet collision rates and contention window sizes in 802.11 protocols. We show that there is a correlation between channel congestion level and packet collision rates, and this relationship can be used to design an effective admission control mechanism. By tuning the value of the three parameters, PRAM, PAM, and CTL; we show that the self-restraining mechanism works well in blocking admission when the channel is congested and allowing admission when the channel is not congested. The third main contribution of this thesis is the development of the Contention Window Reservation Scheduling (CWRS) MAC protocol. This protocol eliminates packet collisions by scheduling contention window. With almost zero collisions, WLAN performance is enhanced and the channel utilisation is maximised. It is widely known that the random backoff algorithm implemented in the 802.11 MAC protocols does not provide an efficient medium access mechanism and has high collision rates when the number of nodes in the WLAN is large. By replacing this with scheduled reservation, nodes can access the channel in an orderly manner and avoid collisions. Here, when the nodes sense that the channel is idled for a DIFS period, all nodes reduce their contention window by one slot. A node which has its Contention Window reduced to zero (CW = CW0) transmits its packet. After transmission, it sets its CW to a fixed set value, CWn. The scheduler within each node, which keeps track of the position of all traffic flows in the channel, then shifts down the CW position to the next empty position. In this way, nodes transmit in an orderly manner and packet collisions are avoided as no node is having the same CW position. We analyse the CWRS mechanism using Markov Chain (MC) analysis and compare it with 802.11b and 802.11e. Using NS2, we run multimedia traffic with the CWRS mechanism and show that throughput performance for all traffic is maximised and delays are sustained to a level that supports QoS for real-time traffic. Fairness to the best-effort (BE) traffic is achieved in this mechanism, and comparison of performance with 802.11b and 802.11e shows that CWRS outperforms these two protocols