L-PLC Channel Noise Prediction Based on Improved Weight One-Rank Local-Region Algorithm

In this paper a novel L-PLC (low-voltage power line communication) noise prediction algorithm is proposed based on the chaotic characteristics of L-PLC. The new algorithm is derived from weight one-rank local-region method by revising the Euclidean distance formula. With the revised formula, not only different relevance of phase points to the focal point but also the prediction contribution of each dimension of phase points for forecasting is used to select neighboring points of predicted model. Moreover the distance calculated from the revised formula can be made as the weight in predicted model. So the model reflects the dynamic characteristics of chaotic systems more accurately. The effectiveness of new algorithm is confirmed by applying it to the prediction of noise time series measured from L-PLC channel.