ICI Cancellation Based on MIMO Decoding for FTN Non-Orthogonal FDM Systems

In this paper, we investigate multiple-input multiple-output (MIMO) decoding in faster-than-Nyquist non-orthogonal frequency-division multiplexing (FTN-NOFDM) for inter-carrier interference (ICI) cancellation. The MIMO decoding is based on depth-first tree search or breadth-first tree search. The sphere decoder (SD) is a commonly used decoding algorithm based on the depth-first tree search. Maximum-likelihood detection with QR decomposition and M-algorithm (QRM-MLD) is a commonly used decoding algorithm based on the breadth-first tree search. By adding the initial radius to both SD and QRM-MLD, the computational complexity of these can be further decreased without the loss of the ICI cancellation performance. In the experiment, SD, QRM-MLD, and their corresponding simplified versions are used for the ICI cancellation in a 28-Gb/s FTN-NOFDM system with transmission over a 10-km standard single-mode fiber. The 3-dB equivalent bandwidth of this system is only approximately 5.5 GHz. Under serious high-frequency distortion caused by limited bandwidth, the bit error rate performance of the FTN-NOFDM can still reach the forward error correction limit due to the bandwidth compression and effective ICI cancellation. FTN-NOFDM with 20% and 30% bandwidth savings has approximately 1.3- and 1.1-dB improvements in receiver sensitivity compared to FTN-NOFDM with 10% bandwidth savings, respectively. To the best of our knowledge, this paper is the first experimental demonstration of the depth-first tree search and breadth-first tree search for FTN-NOFDM in a bandwidth-limited optical transmission system. The experimental results show that FTN-NOFDM is a promising scheme for bandwidth-limited systems.