| With the development in wireless communication and network technologies, multimedia applications and services are widely used in mobile cellular networks recent years. The next generation wireless networks are expected to support multimedia traffic such as voice, data and video, which request different bandwidths and quality of service (QoS) requirements. How to guarantee QoS of multi-class services and maximize resource utilization at the same time are the main goal for network designers. Mobility of hosts, scarcity of the spectrum resource and channel fading make QoS provisioning more challenging task in wireless networks compared with the provisioning problem existing in wireline networks. Call admission control is very important technology in wireless resource management, and it is one of the key schemes to decrease congestion, guarantee QoS and improve resource utilization. Our research is mainly focus on call admission control policy in cellular wireless communication networks based on previous achievements. Imprecision, dynamic change and burst are features of the flow of wireless traffic. The fluctuation in the flow of traffic can not be predicted in advance, conventional call admission control policy can not adapt itself to the change of traffic. But intelligent control has lots of advantages to cope with complexity and uncertainty. Intelligent techniques include fuzzy logic systems, neural networks, genetic algorithm and expert systems. Without any math model developed in advance, these intelligent techniques can deal with uncertain and dynamic systems, and they can solve complex problems in communications and networks in most cases. We apply intelligent control and optimal theory to the research of call admission control policy in cellular wireless networks in this paper. Due to static guard channel policy can not adapt itself to the change of traffic, we use fuzzy logic for single-class service and multi-class services in cellular wireless networks. We propose new schemes of fuzzy dynamic guard channel (FDGC) and multi-class fuzzy dynamic guard channel (MFDGC). They adjust dynamically the number of guard channels according to the current number of the guard channels and one important QoS parameter——call dropping probability (CDP). Simulation results show that the fuzzy schemes have better robust performance. They keep the CDP and call blocking probability (CBP) low at the same time, guarantee the QoS provisioning and increase the resource utilization of the network. Since multimedia services request different bandwidths and QoS requirements, we consider different class and different priority level of different call type, then propose the new scheme of multi-class fuzzy call admission control (MFCAC) in mobile cellular networks. It can make decisions to admit or reject a call according to the number of available channels and CDP of different class services. Simulation results show that this scheme performs much better than conventional scheme does. Based on Semi-Markov Decision Process, we propose an optimal call admission control policy for multi-class services. After meeting the requirements of CDP constraints, the weighted linear function of CBP and CDP of each class service is the optimized object. Linear programming formulation is used to find the optimal solution. Simulation results show that the optimal scheme performs much better than conventional scheme does. When cell capacity and class of service are large, system state space and action space will be very large, and this will make it complicated to solve problem for SMDP approach. Since genetic algorithm has the ability in optimization computation, we propose a near optimal call admission control scheme using genetic algorithm. The weighted linear function of CBP and CDP of each class service is the optimized object. Simulation results are compared with the optimal solution obtained from linear programming for SMDP. The results show that the genetic algorithm approximates the optimal solution very well.
|
PDF Full Text Request