A Novel QoS Mapping Algorithm for Heterogeneous Home Networks Using General Regression Neural Networks

This paper presents a novel QoS management algorithm for heterogeneous home networks using the general regression network (GRNN). The algorithm first establishes profiles of the responses of users to various usages of different communication links for different services. By the employment of the GRNN, the prediction of the future responses of a service from a user could then be made based on the profiles. Among the bandwidth allocations predicted to receive positive feedback, the algorithm finds the bandwidth allocation consuming minimum total bandwidth in the network. The algorithm requires no offline training and achieves fast adaptation to the responses of users. Both analytical and numerical analyses of the proposed algorithm have been included in the paper to show the effectiveness of the proposed algorithm.