Abstract : In remote rural areas, it is not possible to employ massive multiple-input multiple-output (MIMO), small cells, and ultra-dense networks (UDNs) with the aim of increasing throughput. A solution is to improve the waveform spectral efficiency, integrating faster than Nyquist (FTN) signaling with generalized frequency division multiplexing (GFDM). However, this presents high self-interference in the time and frequency domains, requiring dedicated detectors for performance loss mitigation. Hard decision detection schemes primarily designed for MIMO have been adapted to detect FTN-GFDM signals without degradation of the uncoded bit error rate (BER), but these schemes are suboptimal in terms of capacity as they do not provide all the information contained in log-likelihood ratios (LLRs). We design and evaluate in this paper a soft sphere detector (SD) algorithm for FTN-GFDM that can be integrated with state-of-the-art forward error control (FEC) decoders for good BER performance over mobile channels. The SD detector is combined with polar codes, and the BER and complexity are evaluated for different channel models. The results show that FTN-GFDM can provide high spectrum efficiency gains without significant coded BER losses and with affordable complexity on the receiver side, which makes this waveform an interesting candidate for mobile networks in remote areas.
Index terms : multicarrier systems , tree search algorithm , soft detection , non-orthogonal waveform , polar coding , spectral efficiency