Study On Admission Control And Scheduling Mechanisms For MAC Layer In Wireless Local Network

The public's increasing demand for wideband mobile data is driving wireless communication networks toward the all-IP, wideband and mobile network. Now as a broadband IP access mode, WLAN faces the new challenges of high mobility support, more efficient use of the available scarce spectrum resources and QoS guarantees for different services. MAC layer in WLAN, the manager of wireless resources, controls user data to access wireless media and allocates the resources to users. In this paper, a prototype of MAC protocol for mobile WLAN system, which adopts TD-CDMA scheme and provide multi-rate support with OVSF for users, is proposed. In mobile WLAN system, the resources needed by a service are allocated dynamically. So the admitted ongoing service may be interrupted during hand-off process because of loss of resource. In addition, concurrent dropping may happens when an admitted service changes its state from off to on and the channel capacity can't accommodate its bandwidth requirement. So a user service may suffer three kinds of interruption -blocking, hand-off dropping and concurrent dropping. In this paper, the relationship between service's completion probability and these interruptions is analyzed. At the same time, based on probability analysis, an analytical relationship model for WLAN composed of homogeneous cellular networks is developed. Then assuming service's data source is modeled by on-off traffic model, a two dimensional Markov process is established to compute these blocking and dropping probabilities. The impacts of different new service arriving rate, service's traffic characteristic, user's mobility, service holding time and hand-off service's admission threshold on performance metrics are evaluated and compared through numerical examples. Furthermore, the two dimensional Markov model is expanded to an n-dimensional Markov models to provide an analytical model to analyze admission control of multiple kinds of service. These two analytical models can provide a basis for helping to set the service admission threshold. Because wireless channel quality changes with user's mobility, appropriate scheduling algorithms in MAC layer can improve whole system performance if they control user's access to channel by combining channel characteristic of time-varying and user data source characteristic of intermittence. Assuming parametric stabilization channel, the minimum power allocation condition, which supports multi simultaneous transmissions and can meet minimum BER requirements of these transmissions, is educed to schedule data packet efficiently, according to analysis of SINR of code channel in uplink and downlink. And the best transmission rules to achieve or approach the maximum transmission rate in a time slot are derived through the condition. In order to arrive at the allowed maximum transmission rate, both rate and power schedulings are needed. Then a time frame is considered. In order to allocate more user data in a time frame, users are classified according to their channel gain while slots in a frame are classified according to users arranged on them. A suitable slot can be chosen for a user by comparing levels between the user and slots of a frame. In addition, scheduling algorithms -BCFS, EDFS and LQFS are put forward by taking consideration of slot selection priority and user selection priority (i.e. best channel first, delay shortest first and longest queue first). Moreover, the problem of providing users with fair throughput is transformed into an optimization problem with restriction condition of weighted average throughput fairness and objective of maximization of total average throughput. Scheduling algorithm -CDFS (i.e. channel dependent fair scheduling) is designed by resolving this problem. For enhancing performance of short term fairness, CDFS is modified through update the user selection priority and scheduling algorithm CDFS-SF is obtained. At last, the feasibility of the MAC protocol is examined and performance of each scheduling algorithms are compared through many simulation tests. The simulation results show that CDFS can not only provide high system throughput, but also guarantee the fairness among service flows insensitive to delay. CDFS-SF can offer fairness among flows sensitive to delay.