Adaptively increasing connection capacity of real time streaming by fuzzy neural intelligent admission control

Abstract

Wireless multimedia services have to rely on strict quality of service (QoS) requirements to reach the feasibility, including guaranteed available bandwidth, tight delay constraint, and hard jitter tolerance. WiMAX has emerged as one of the most promising broadband wireless access technologies to support these types of services. To guarantee the QoS of connections, CAC (Call Admission Control) can grant a new call only while the required bandwidth is available. With concern about connection capacity, the CAC seems rigid and turns into an obstacle instead while the bandwidth of some connections could be stolen temporarily. Thus, adaptive admission control, differing from the conventional CAC, can flexibly increase connection capacity by stealing bandwidth based on user streaming buffer (USB). A connection in sufficient USB state implies that the undergoing streaming application can perform smoothly even if its bandwidth is stolen temporarily. Therefore, how to determine the USB state of each connection is the most important topic in this study. In order to effectively decrease the risk of system overuse arising from stealing bandwidth, the adaptive admission control is extended by means of an intelligent mechanism to decide the USB states. The simulation results show the proposed intelligent mechanism utilizing fuzzy neural network (FNN) really outperforms than the adaptive admission control in terms of controlling the probability of poor application performance. With approximate connection capacity (or call blocking probability), the intelligent resource management indeed decreases the risk of system overuse.