Turbo Equalization With Optimized Linearly Encodable LDPC Codes

Abstract — Turbo equalization is a joint iterative equalization and decoding technique known to produce superior performance for Inter-Symbol Interference (ISI) channel, comparing to non-iterative methods. We analyze turbo equalization that employs linearly encodable LDPC codes. The codes are irregular and their degree profiles are optimized in order to achieve capacity-approaching performance. To optimize the code, we use Ex-trinsic Information Transfer (EXIT) charts and introduce an optimization algorithm that searches for the code degree profile to maximize the minimum distance between the two curves in the EXIT chart, one corresponding to the channel equalizer and one to the decoder. Our simulation results reveal that turbo equalization with the optimized codes can approach the channel capacity to within 0.28 dB at BER of 10−6 and the frame size of 106 bits. I