Efficient detection in uniform linear and planar arrays MIMO systems under spatial correlated channels

Abstract

Summary In this paper, the efficiency of various multiple‐input multiple‐output (MIMO) detectors was analyzed from the perspective of highly correlated channels, where MIMO systems have a lack of performance, besides in some cases, an increasing complexity. Considering this hard but a useful scenario, various MIMO detection schemes were accurately evaluated concerning complexity and bit error rate performance. Specifically, successive interference cancellation, lattice reduction, and the combination of them were associated with conventional linear MIMO detection techniques. To demonstrate effectiveness, a wide range of the number of antennas and modulation formats have been considered aiming to verify the potential of such MIMO detection techniques according to their performance‐complexity trade‐off. We have also studied the correlation effect when both transmit and receiver sides are equipped with uniform linear array and uniform planar array antenna configurations. The performance of different detectors is carefully compared when both antenna array configurations are deployed considering a different number of antennas and modulation order, especially under near‐massive MIMO condition. We have also discussed the relationship between the array factor and the bit error rate performance of both antenna array structures.

A Seysen's algorithm–based incremental lattice reduction

“…In order to further study how close the performance gaps between these ILR schemes can be, we enlarge the plots of average FLOPs against ranges from 9 to 20, which is illustrated in Figure 6. As we can see from it, the FLOPs of the constrained schemes and the unconstrained ones are almost the same when ∈ [9,14] for both CLLL-ILR and SA-ILR. Furthermore, the FLOPs of the SA-ILR and CLLL-ILR algorithms are almost constants when ranges from 10 to 14, and the former is just around 100 higher than the CLLL-ILR methods.…”

“…In this case, all plots reach their optimal BER performance with parameter = 11 while the SA-based algorithms have lower BERs. Intuitively, the SA-ILR algorithms are most obviously superior to the CLLL-ILR schemes when ∈ [9,14], which is called the feasible interval. We can see that this interval has a relatively broad range, which at least covers the optimal for SNR from 27 dB to 36 dB (see the optimal set in Section 4.3.2).…”

“…From the analysis above, we learn that, instead of using the fixed RA parameter, the unstrained algorithms can achieve the same performance as the constrained algorithms with a set of properly selected parameters. Therefore, a rough search for parameter over the integer set [4,18] for SNR ranges from 20 to 36 dB is conducted, and the corresponding optimal set for both SA-ILR and CLLL-ILR is [9,9,9,9,9,9,9,9,10,10,10,11,12,12,12,12,12]. They have the same optimal set because, on the one hand, this is the optimal set based on the rough search.…”

“…Observing this, the lattice reduction (LR) technique is introduced to improve their performance. [8][9][10][11] It has been proven that the LR-aided linear 12 and successive interference cancellation (SIC) 13 detectors are able to achieve the same diversity order as the ML detector, as the LR provides a near orthogonal channel matrix to reform the system model. 14 Moreover, a lot of research has been done to directly reduce the complexity of LR algorithms.…”

Graph traversal aided detection in uplink MBM massive MIMO based on socio‐cognitive knowledge of swarm optimization

“…Among several swarm and evolutionary algorithms, particle swarm optimization (PSO), 12 ant colony optimization (ACO), 13 gravitational search algorithm (GSA), 14 and firefly algorithm (FA) 15 are noteworthy and promising algorithms which are well-explored in the literature. PSO-based MIMO detection algorithms yield superior performance with lower computational complexity than zero forcing (ZF), 16 minimum mean square error (MMSE) 16,17 and MMSE with successive interference cancellation (MMSE-SIC) and MMSE with ordered SIC (MMSE-OSIC). 16,18 However, the computational complexity of PSO inspired symbol-detection algorithms [19][20][21] increases with the number of antennas, and hence, they are not suitable for mMIMO systems.…”