The ASIC design of an LMS-based decision feedback equalizer for TDMA digital cellular radio

Abstract

In a TDMA mobile communication system, an adaptive decision feedback equalizer (DFE) is the common device to compensate the intersymbol interference caused by the multipath effect. Due to the fast fading characteristic of channels, most conventional DFEs use the recursive least square (RLS) algorithm. However, it is known that the RLS algorithm is computationally intensive and not easy to implement. The LMS is well known for its simple structure, however, due to its poor tracking property, it cannot be used. Recently, a multiple-training and multiple-tracking LMS (MTT-LMS) algorithm was developed [7]. It was shown that the DFE trained by the MTT-LMS algorithm can yield a satisfactory result at moderate mobile speed. The MTT-LMS algorithm remains the simple structure of the LMS algorithm and is suitable for hardware implementation. In this paper, we consider the ASIC design of the MTT-LMS DFE for application to North American IS-54 systems. We have designed an efficient architecture not only has small chip area but low clock rate. We also develop a new decision device to provide the desired signal for the MTT-LMS algorithm and the implementation of the device is simpler than the conventional one.