A Low-Complexity Receiver for Multicarrier Faster-Than-Nyquist Signaling over Frequency Selective Channels

In this letter, we propose a low-complexity receiver for multicarrier faster-than-Nyquist (MFTN) signaling over frequency selective channels. By introducing a time-frequency trellis state, we are able to reformulate the MFTN detection problem into a linear state-space model. Building on this, a vector-form Forney style cycle-free factor graph is constructed. Then, Gaussian message passing technique is employed to derive the parametric message updating expressions. Accordingly, only the mean vectors and covariance matrices have to be calculated, which significantly reduces the computational complexity of MFTN detection. Simulation results demonstrate that the proposed receiver outperforms the symbol-by-symbol receiver by about 1 dB in terms of bit error rate (BER) and approaches the BER performance of the maximum a posteriori probability detector. Meanwhile, 39% more spectral efficiency can be obtained at a BER performance loss of less than 0.3 dB.