Efficient Algorithms for Noise Estimation in Electrical Power Line Communications

Power Line Communication (PLC) has received much attention due to the wide connectivity and availability of power lines. Effective PLC must overcome the harsh and noisy environments inherent in PLC channels. Noise in power lines is modeled as a cyclostationary Gaussian process. In order to achieve reliable communication using power lines, effective measures including error control techniques need to be taken against this particular noise. Low-Density Parity-Check (LDPC) codes have excellent performance in power lines. This paper presents two new iterative algorithms for noise estimation on power lines based on Higherorder statistics and the Maximum-Likelihood (ML) estimation principle, respectively. The algorithm based on Higher-order statistics uses second, fourth, and sixth moments of the received noisy signal to provide a signalto-noise ratio (SNR) estimate. For the ML estimation algorithm, a derivation of the ML estimate of the amplitude of a Binary Phase-Shift Keying (BPSK) modulated signal is presented. Then, the proposed iterative search algorithm is developed. The proposed algorithms are especially favorable in cases of low SNR values, e.g., the ML estimation algorithm can achieve as large as 7.5 dB and 11.7 dB gains over conventional estimators at an SNR of -5 dB and -10 dB, respectively. Furthermore, since accurate SNR estimation is required for “good” (in terms of bit-error rate (BER)) decoding performance of LDPC codes, the performance of the proposed schemes is compared to some of the previously suggested SNR estimation algorithms. Finally, simulation results show that the proposed estimators perform close-to-optimum at a significantly lower computational complexity. Key-Words: Power Line Communications (PLC), Low-Density Parity-Check (LDPC) Codes, signalto-noise ratio (SNR) estimation.