A parallel architecture for adaptive channel equalization based on Kalman filter using MMAlpha

We apply the square-root covariance Kalman filter to solve the ill-posed problem of signal reconstruction specifically for adaptive channel equalization. The computation latency and the throughput of this algorithm are significantly improved with the derivation of a systolic architecture using MMAlpha which is a tool dedicated to automatic synthesis of systolic architectures. The proposed architecture is validated by a VHDL simulation in the case of a time varying channel impulse response. The performance evaluation is based on a 20-bits wordlength and is synthesized in a 0.5 /spl mu/m CMOS technology.