A high speed low-noise equalization technique with improved bit error rate

This paper presents a low-noise adaptive equalization technique suitable for Gbps data transmission. The proposed equalization technique uses adaptive analog FIR filter as the equalization filter for high-speed operation. The FIR filter is trained with a random search algorithm on an error signal calculated by comparing the spectral energy of the equalized signal with the desired spectral energy. The proposed error detection scheme includes the comparator as an integrated part of the equalizer to avoid over-boosting the high frequency noise, thus it has less high-frequency noise amplification than typical equalizers. Bit error rate analysis comparing the noise performance of the proposed equalization method with typical equalization methods is presented; simulation shows that the proposed method has much improved bit error rate performance at the same noise level. The high-frequency noise amplification of the proposed equalization method is 27.38 dB lower than typical method for 125 Mbps data transmission over 200 m UTP5 cables