Parallel-Pipelined Architecture for the Kalman Based Adaptive Equalizer

The requirement of many data communication systems is to employ an adaptive equalizer to minimize the inter-symbol interference. Several adaptive Kalman equalizers have been reported in literature. In these works either the least mean square (LMS) or the recursive least squares (RLS) or the Kalman algorithm have been adopted for channel estimation. The Kalman estimation method can lead to significant improvement in the receiver bit error rate (BER) performance. The use of a Kalman filter for channel estimation leads to a state model of size 2ntimes2n, where n is the number of filter taps. These solutions are computationally intensive and follow a nonlinear relation in the observation equation. New methods have to be followed to solve the nonlinear model resulting in complex parallel structures. This paper proposes a new approach for the real time implementation of the adaptive Kalman equalizer by providing two Kalman filters that run concurrently to perform the estimation and detection. Thus the Kalman estimator operates in parallel with the Kalman filter based equalizer following a linear model and the size of the state matrix reduces to ntimesn. Parallel-pipelined architectures are proposed to perform the time update and measurement update equations of the Kalman equalizer and Kalman estimator. The functionality of the proposed architecture has been verified through VHDL simulation. The synthesis results are presented. It is shown that the convergence performance of the proposed approach is superior to that of the Kalman-RLS and Kalman-LMS adaptive equalizers