Iterative Detection Algorithms for High Spectral Efficiency Wireless Communication Systems

Publication Type:
Issue Date:
Full metadata record
With the ever-growing of the number of devices and new applications (e.g., in industry automation, intelligent transportation systems, healthcare) and given the severe bandwidth congestion observed at the sub-6GHz frequency bands, it is critical to develop high spectral efficiency transmission techniques. However, the detection of new transmission signalling becomes more challenging compared to that of its conventional counterpart. In this thesis, advanced iterative detection algorithms for index modulation-aided and faster-than-Nyquist (FTN) signalling-aided communication systems are investigated. First, factor graph-based message passing algorithms are proposed for the joint phase noise (PHN) estimation and signal detection in orthogonal frequency division multiplexing with index modulation (OFDMIM) systems. The solutions are superior to conventional extended Kalman filter and variational approaches in terms of the robustness to severe PHN, as well as the realistic imperfect channel state information and residual carrier frequency offset. Second, amalgamated belief propagation and mean field message passing methods based iterative detection algorithms are developed for satellite communication systems relying on the dual mode-aided index modulation (Sat-DMIM) over nonlinear dispersive satellite channels. The computational complexity of the proposed detector is further reduced by approximating the nonlinear messages using a Taylor series expansion technique. The bit error rate performance of Sat-DMIM is improved compared with the conventional linear equalizer which directly linearizes the nonlinear system model. Third, frequency-domain joint channel estimation and signal detection methods using the variational Bayesian framework for FTN systems over frequency-selective fading channels are designed. Taking into account the structured inter-symbol interference imposed by FTN signalling and dispersive channels, reliable estimates of channel coefficients and FTN symbols are obtained by minimizing the variational free energy without the aid of the cyclic prefix. Simulation results show that the FTN system relying on the proposed iterative detection algorithm can significantly improve the spectral efficiency compared to its conventional Nyquist counterpart.
Please use this identifier to cite or link to this item: