A GRNN -Based Traffic Reconstruction Algorithm to Surface Wave-Based Power Line Communications

In power line communications, network traffic has the highly dynamic inherent nature. This leads to the huge challenge to accurately predict and estimate them. Accordingly, this results in new problem for power communication networks' traffic engineering and power scheduling task. This paper proposes a novel reconstruction approach to estimate dynamic traffic in surface wave-based power line communications based on the Generalized Regression Neural Network (GRNN) theory. Firstly, we use the GRNN theory to describe time-varying network traffic features. Then by considering the spatio-temporal correlation feature of network traffic, we construct the dynamic GRNN model to describe network traffic. Thirdly, we propose a new traffic reconstruction algorithm to estimate and predict network traffic in power line communications, based on the proposed model. Thus based on the proposed method, we can quickly estimate network traffic and obtain accurate estimation values. Simulation results show that our approach is feasible.