Enhanced performance and efficiency schemes for generalised spatial modulation.
Naidoo, Nigel Reece.
MetadataShow full item record
Generalised spatial modulation (GSM) is a relatively new multiple-input multiple-output (MIMO) technique, which conveys information via a spatial constellation, comprising groupings of active transmit antenna, and a conventional M-ary quadrature amplitude modulation (MQAM) or M-ary phase shift keying (MPSK) signal constellation. The first objective of this thesis is to propose a GSM scheme with improved bit error rate (BER) performance, termed generalised spatial modulation with constellation reassignment (GSM-CR). A framework for the design of GSM-CR systems is presented. An analytical bound for the average BER of GSM-CR over independent and identically distributed (i.i.d.) Rayleigh flat fading channels is also derived and the accuracy of this bound is verified by Monte Carlo simulation results. Moreover, Monte Carlo simulation results demonstrate that for 9 bits/s/Hz transmission, GSM-CR achieves gains of 5 dB and 4 dB, at a BER of 10-5, when compared to GSM and spatial modulation (SM), respectively. The maximum likelihood (ML) GSM-CR detector offers optimal BER performance at the expense of high computational complexity, particularly when high order digital modulation techniques are employed. The second objective of this thesis is to propose a new GSM-CR detector, termed zero-forcing maximum likelihood (ZF-ML). The proposed ZF-ML detector is characterised by low computational complexity, complexity that is independent of the order of digital modulation technique utilised and BER performance similar to that obtained by the ML detector. Performance and complexity comparisons between the ZF-ML GSM-CR detector and popular GSM and GSM-CR detection schemes reveal that for the 𝑁𝑡×4 64QAM configuration, where 𝑁𝑡 refers to the number of transmit antennas and 4 is the number of receive antennas: i) the ZF-ML GSM-CR detector reduces computational complexity at the receiver by up to 60% and exhibits similar BER performance when compared to the ML GSM-CR detection scheme; ii) ZF-ML configured with two candidate symbols per initial symbol estimate reduces the receiver complexity by up to 60% whilst ZF-ML configured with eight candidate symbols per initial symbol estimate achieves gains of up to 5 dB, at a BER of 10-5, when compared to the ML GSM detector; iii) for systems that encode more than one information bit in the spatial domain, certain ZF-ML configurations operate at a higher computational complexity of up to 235% when compared to the SV GSM detector. However, the higher computational complexity imposed by the ZF-ML GSM-CR detector is traded-off by significant gains of up to 5 dB, at a BER of 10-5, when compared to the SV GSM detection scheme. The GSM-CR scheme only permits two active transmit antennas during a particular timeslot. This limits the number of bits that can be encoded in the spatial domain and therefore the overall spectral efficiency attainable. The third objective of this thesis is to propose a generalised spatial modulation with dual constellation reassignment (GSM-DCR) technique, which is geared towards improving the overall spectral efficiency of GSM-CR. A framework for the design of GSM-DCR is presented. Moreover, an analytical bound for the average BER of GSM-DCR over i.i.d. Rayleigh flat fading channels is derived, and the accuracy of this bound is verified by Monte Carlo simulation results. A comparison between various GSM-DCR, GSM-CR and GSM schemes, which employ equivalent system configurations, reveals the following: i) for all configurations, GSM-DCR improves spectral efficiency by 1 bits/s/Hz as compared to both GSM-CR and GSM; ii) for the 6×4 64QAM configuration, GSM-DCR exhibits reduced performance of 2 dB, at a BER of 10-5, as compared to GSM-CR. However, the degraded performance is traded-off by the enhanced spectral efficiency of 1 bit/s/Hz offered by the GSM-DCR scheme; iii) for the 6×4 64QAM configuration, GSM-DCR can achieve gains of 2.5 dB, at a BER of 10-5, when compared to GSM.