Adaptive algorithms for sparse channel estimation for millimeter Wave hybrid MIMO systems

Abstract

With the increase in high-speed data demand and lower latency for upcoming fifth-generation (5G) and beyond communication systems, millimeter-wave (mmWave) multiple-input multiple-output (MIMO) has emerged as one of the promising physical layer technique for existing radio frequency (RF) communication systems. Although promising, it suffers from much greater attenuation compared to conventional cellular bands (sub-6 GHz band) due to penetration losses, reflection, and signal atmosphere. However, thanks to the short wavelength of mmWave signals, large antenna arrays can be packed into a small area. Hence, a large number of antennas can be adopted at both the transmitter and receiver to provide significant beamforming gains. However, the large number of antennas makes fully digital beamforming (in which each antenna is connected with a separate RF chain) impractical due to the huge power consumption caused by devices operating at radio frequency (RF). Therefore, hybrid (combination of analog and digital) architectures have been proposed, which can reduce hardware costs and power consumption with a reduced number of RF chains. Due to the hybrid architecture and the large number of antennas, it is difficult to obtain the channel state information (CSI), which is crucial for obtaining desirable beamforming gains. Further, due to high blockage and lower scattering, the mmWave channel is sparse, meaning that impulse responses are dominated by a small number of clusters of significant paths. Hence, this thesis develops a novel sparse adaptive online channel estimator (signifying that the estimator continually adapts to changes in the input data stream as it is received, rather than processing the entire dataset at once as a block update) based on the zero attractor least mean square (ZALMS) algorithm for mmWave hybrid MIMO systems. In this algorithm, l0 and l1 norm penalties are introduced in the least mean square (LMS) algorithms, which introduces a zero attractor in the LMS weight update recursion. This process shrinks the coefficients of inactive taps and hence reduces the steady-state mean square error (MSE) floor, consequently increasing the estimation accuracy and maximizing the overall spectral efficiency (SE) of mmWave hybrid MIMO systems.