Fractal Prediction Algorithm of L-PLC Channel Noise Based on Phase Space Reconstruction

Abstract

In this paper, the fractal characteristics including self-similarity and long-dependence of the L-PLC (Low-voltage power line communication) noise signal are investigated by the R/S analysis firstly. And then, according the weak chaotic and statistical self-similarity of the noise signal, a new fractal prediction algorithm is proposed. In this algorithm, the fractal interpolation based on phase space reconstruction sample selection method and fractional collage theory is applied to determine an iterated function system, whose attractor is similar with the primary noise signal. Finally, the fractal prediction model is set up by this iterated function system. The results show that, the average prediction error between the actual value and prediction value is smaller than 6.0%, comparing with the traditional fractal algorithm, the average reducing prediction error was close to 20%. The experimental results suggested that the proposed algorithm improves the prediction accuracy and more suitable for the L-PLC noise signal prediction.