A Novel Neural Network GA-Optimized Controller for QoS Support in Wireless MACs

As the multimedia applications such as voice and video streaming across the Internet emerge, there is a growing demand for quality of service (QoS) over different networks. Among different Internet-based networks, 802.11 wireless local area networks (WLAN) are among the most important. Two significant factors that affect multimedia quality over a wireless network are the packet loss and delay. While the packet loss is tolerable to some extent the delay plays a major role in the performance. In this paper an effective and computationally efficient neural network controller is proposed to optimize the IEEE 802.11 medium access control layer MAC behavior by adaptively adjusting the contention parameters. The neural network architecture and the weights in the network are optimized using genetic algorithm (GA) for high load situations. After this optimization, elaborate simulations are performed to show the effectiveness and efficiency of the method proposed