On coding for Faster-Than-Nyquist signaling

International audienceIn this paper, we study the design of sparse graph based codes for Faster-Than-Nyquist (FTN) signaling. Using an asymptotic approach based on EXIT charts, we show that good low-density parity check codes can be designed that perform well under iterative detection and decoding and that have better performance than a FTN scheme using a code optimized for the additive white Gaussian noise channel