A fuzzy bandwidth allocation controller to support real-time traffic over wireless network

In this paper, we propose a fuzzy bandwidth allocation controller (FBAC) to support two types of services: (i) restricted time-bounded service (RTBS), such as voice and video services, (ii) loose time-bounded service (LTBS), such as data service in wireless networks. Base on the FBAC, five different request assignment strategies are introduced. Simulation results demonstrate that the fuzzy controller with some proposed assignment strategies obtain high network performance, high fairness as well as an acceptable blocking probability. I . INTRODUCTION In wireless network, many MAC protocols have been proposed about how to guarantee the quality of service (QoS) recently. Typically, slotted based protocols, like R-ALOHA, PRMA [l], ISMA [2], RS-ISMA [3], BRMA [4], reserve slots in the next cycle by pervious contention results. Since each of them does not consider the traffic characteristics, it is hard to exactly match the service requirements or obtain the maximum throughput. Adaptive protocols, such as PRMA/DA [5] and DSAMA [6], dynamic change the slots of different kinds of services to provide better QoS. However. too many kinds of services make the algorithms complexity and need much computing time. To reduce the protocol complexity, we simplify service types into two basic classes: (i) restricted time-bounded service (RTBS), such as voice and video services, (ii) loose time-bounded service (LTBS), such as data service. Each kind of services has its special traffic characteristic, bandwidth requirement and acceptable blocking probability. In order to fully utilizing the bandwidth in wireless network, a flexible bandwidth sharing strategy has to be applied. Although an aggressive bandwidth sharing scheme can provide a better management of network resource, due to the unpredictable fluctuations of traffic flow, the blocking probability of restricted time-bounded service may be violated. To overcome the problem, we propose a fuzzy bandwidth allocation controller (FBAC) to deal with the bandwidth sharing problem in the wireless network. Based on the FBAC, five different request assignment strategies are proposed to dynamically select proper LTBS requests to serve. Among these strategies, two of them employ another fuzzy strategy switching controller (FSSC) to obtain the best results. The performance of the fuzzy controllers and strategies are evaluated and compared. The simulation results show that the proposed strategies can improve the network performance, bandwidth utilization, fairness. Most important, it has the ability to control the blocking probability of RTBS within an acceptable range as desired. Moreover, the proposed approach is very simple and can be implemented in hardware easily. The rest of this paper is organized as follows. The system model is given in section 11. In section 111, we present the functional block diagram of the FBAC. In section IV, five different strategies based on the FBAC for assigning requests are discussed. In section V, the performance measurements, simulation models and results are reported. Finally, some conclusions are given in section VI. 11. SYSTEM MODEL In wireless networks, channel is often divided into frames with fixed length and each frame is composed of C time slots (i.e., bandwidth is divided into C channels). In such timedivision multiple access (TDh/lA), each time slot can be used by either RTBS or LTBS. According to the characteristics of two types of services, RTBS has higher priority than LTBS and the LTBS may starve when RTBS is overloaded. To prevent LTBS from starvation, a small amount of .slots in frame, which is denoted as R, is particularly reserved for the LTBS and the remaining slot!; in frame can be allocated for either RTBS or LTBS. For simplicity, we denote the former and the latter as the reserved slots and the sharable slots, respectively. Therefore, the exact number of sharable slots is C-R. The channel's allocation of sharable and reserved slots within a frame in a wireless network is shown as Figure 1. As described above, when customer issues a request for LTBS, a particular parameter "due-time'' is also specified at the same time. This identifies that it is valid only if it is delivered before this specified time. For example, assume a LTBS with transmission time L arrives at time T. the latest timing T ' of starting to service is T+D-L if it can be delayed no later than T+D. 111. THE Fuzzy BANDWIDTH ALLOCATION CONTROLLER (FBAC) In this paper, the fuzzy approach is adopted to determine how many sharable slots can be temporarily borrowed to serve LTBS without degrading the quality of RTBS. There are two input linguist parameters considered for the fuzzitier in FBAC: the RTBS blocking probability, which is denoted as rbp, and the channel free proportion (the ratio of the number of free sharable slots to the total sharable slots). which is denoted as c$. For iriput parameters, we define the corresponding fuzzy term sets: T(rbp) = {Safe, 'Normal, Dangerous} and T(cfp) = {Small, Medium, Large}. The selected membership functions for T(rhp) and T(cfp) are the shape of Gaussianilike function (see Figures 2(a) and 3(b)). 0-7803-5668-3/99/$10.00