Semi-Blind Detection in Hybrid Massive MIMO Systems via Low-Rank Matrix Completion

Second-Order Statistics-Based Semi-Blind Techniques for Channel Estimation in Millimeter-Wave MIMO Analog and Hybrid Beamforming

“…5(c) demonstrates the NMSE performance of both the mmWave-AB-ML and the proposed mmWave-AB-SB schemes versus the pilot length M P at different SNR values for both N D1 = 800 and for a large number of data symbols having perfect s 1 estimates. As expected, the performance of both the ML as well as of the [17] SBL [8] OMP [11] R−ALS [18] (a) proposed mmWave-AB-SB techniques can be seen to improve upon increasing M P . Fig.…”

“…However, the scheme of [17] has the following drawbacks: it i) does not exploit the statistical information in the data symbols; ii) requires the knowledge of a boundL on the number of spatial paths; iii) suffers from high computational complexity, which is on the order of O(N 2 t N 2 rL 2 ). To reduce the large pilot overhead, the authors of [18], using a low-rank matrix completion technique, proposed the semi-blind data detection and channel estimation methods for a sub-6 GHz multi-user (MU) hybrid massive MIMO systems. Two iterative solutions were presented therein: the regularized alternating least squares (R-ALS) and the bilinear generalized approximate message passing (BiG-AMP).…”

“…Fig. 7(a) compares the NMSE performance of the proposed mmWave-HB-SB and the state-of-the-art existing sparse techniques such as SBL [8] and OMP [11], also the atomicnorm-minimization (ANM) [17]-and R-ALS [18]-based semiblind channel estimation techniques for mmWave-HB MIMO systems. It can be observed that not only does the mmWave-HB-SB significantly outperform the OMP and SBL schemes, but is also robust to practical aberrations such as grid mismatch, where the AoDs/ AoAs do not coincide with the dictionary points.…”

“…This is due to the fact that i) the problem formulation in [18] based on a low-rank matrix decomposition is non-convex in nature, which is subsequently solved using the R-ALS scheme by decomposing it into two alternating convex sub-problems; ii) the low-rank matrix decomposition therein relies on the assumption N t << N r , which need not to satisfy in a typical hybrid MIMO system; and iii) unlike the channel estimation model formulated in Eq. (26)- (30), the scheme in [18] does not excite all the angular modes of the mmWave MIMO channel. Thus, the proposed scheme is ideally suited for mmWave MIMO channel estimation in practical settings.…”

5G Massive MIMO Signal Detection Algorithm Based on Deep Learning

“…e purpose of massive MIMO signal detection is to recover the original signal at the transmitter under noise and interference environments. It is an important link to improve the overall performance of the system and reduce the complexity of the system [6]. Low-complexity signal detection algorithms have become a research hotspot in recent years [7].…”