A New Bi-Directional Soft Decision Linear Equalizer

A novel low complexity soft-input soft-output equalizer is proposed for reducing the effects of intersymbol interference channels. The equalizer uses a bi-directional multipass algorithm to successively improve a soft interference canceler/adaptive linear filter combination. The resulting transmitted symbol estimates are calculated with computational complexity of cubic order in channel memory length. Traditional turbo equalization is a joint equalizer/decoder scheme which relies on estimates from the decoder to improve equalizer performance. In contrast, the results presented here show the equalizer can achieve performance gains without the presence, or use of coding/decoding. We show significant improvement is achieved with each pass and the equalizer has near identical performance to the optimal symbol-by-symbol maximum a posteriori detector. Furthermore, we present an even lower complexity version of this scheme that offers similar performance with complexity of linear order in channel memory length.