A low complexity approach to Turbo Equalizer

The turbo equalizers (TEQ) proposed in literature utilize equalizers based on trellis, soft Wiener filters. The resulting complexity of these equalizers is exponential and cubic in terms of the sampled channel impulse response (CIR). The interference cancellation based decision feedback filter based equalizers requires adaptation of two filters simultaneously. In this paper, a low complexity equalizer is proposed that neither uses a trellis nor a Wiener filter. The proposed equalizer utilizes a soft interference cancellation (SIC) technique that uses the log likelihood ratio (LLR) available at the matched filter (MF) using all the coded bits in a given block of data. The MF output is justified as Gaussian distributed and the LLRs are computed accordingly. This is fed as the apriori to the decoder after suitable deinterleaving. The soft estimates for the bits are used to form an estimate of the interference with the help of perfect channel tap knowledge at the decoder output. This estimate of interference is subtracted from the MF output giving the SIC framework. We call it a soft decision feedback equalizer (SDFE).The SDFE bypasses the filters completely resulting in a linear complexity in CIR. Simulation results over four different channels show that the receiver performance improves with iterations and a gap of 1-3 dB is observed from the coded AWGN bound depending on the channel type. Two different TEQs based on namely soft output Viterbi algorithm (SOVA) and the Wiener filter respectively are compared with the SDFE. Key-Words: SIC, Wiener Filter, LLR, SDFE, SOVA