Investigation of Fast and Efficient Adaptation Algorithms for Linear Transversal and Decision-Feedback Equalizers in High- Bitrate Optical Communication Systems

Abstract:
We investigate fast and efficient adaptation algorithms for linear transversal feed-forward (FFE) and decision feedback (DFE) equalizers in intensity modulated optical communication systems with direct detection (IM/DD). First, we employ the well-known minimum mean square error (MMSE) adaptation criterion and give a benchmark with respect to minimal bit error rate (BER). We further take the least-mean-square algorithm (LMS) as stochastic implementation of the MMSE scheme and derive some simplified versions which use coarsely quantized signals. Consequently, the complexity of a digitally implemented adaptation circuit is reduced and application to bitrates of 40-100 Gbit/s becomes more feasible. The performance of these simplified versions is assessed with respect to adaptation speed and accuracy in comparison to the MMSE equalizer. It is shown that sufficient fast and almost accurate adaptation of the equalizer can be achieved.