Study On Call Admission Control And Intelligent Resource Allocation For QoS Support In Wireless Mobile Networks

This dissertation addresses the problems of call admission control (CAC) and intellengent banswidth resource allocation for providing Quality of Service (QoS) to multimedia applications in the next generation wireless communication networks.On the provision of connection-level QoS measured by two metrics, which are named as the new call blocking probability and the handoff dropping probability, our multimedia service model consists of three service classes, i.e., handoff-guaranteed handoff (HGH) service, handoff-prioritized handoff (HPH) service and new call, with different QoS requirements. Thus radio resources are reserved in a different manner for each service class. In this paper, some famous existing call admission control schemes are discussed and analysised and we can come to a conclusion that most of them are queueing model-based approaches.To overcome the shortcomings of the existing CAC approaches, a novel measurement-based admission control algorithm for multi-class traffic is proposed. It can maximize statistical multiplexing gain (the number of users provided service in a network system at a time) while guaranteeing the predefined quality-of-service (QoS) requests such as handoff-dropping and new-call-blocking probability. The approach updates the reservation threshold through a closed-loop control by directly forecasting the next possible number of handoff-dropping calls using auto regressive moving average (ARMA) model. So the computational complexity of Markov chain is avoided. In addition, a concept named acceptance probability is additionally introduced to obtain equivalent dropping probabilities for different traffic classes. Similar to pre-distortion technology adopted in signal processing theory, the service classes with lower bandwidth requirements are accepted with lower acceptance probabilities that depend on the relative dropping probabilities, so as to pre-assign higher relative admission priorities to traffic classes with higher bandwidth requirements especially under high overload. The simulation results suggest that the proposed approach enables us to achieve a higher level of statistical multiplexing gain than existing schemes while reliably meeting the target handoff-dropping probability. Also, the algorithm provides equal handoff-dropping probabilities for different traffic classes under heavy overload.The next-generation multimedia wireless networks, such as Wideband Code Division Multiple Access (W-CDMA), define a range of bitrate values between guaranteed bitrate and maximum bitrate. In order to effectively provide better QoS to users, a Hopfield neural network (HNN) is used in this paper to allocate bandwidth resources adaptively and fairly among the admitted calls. Also, we propose a new QoS parameter for adaptive multimedia services: the satisfaction degree (SD), which is rarely discussed in previous literatures. Based on SD, a measurement-based call admission control framework is proposed to satisfy the application QoS requirements and to utilize the network resources efficiently. Detailed simulation experiments are conducted to validate the proposed scheme.