A Linear Bayesian Learning Receiver Scheme for Massive MIMO Systems

Abstract

Much stringent reliability and processing latency requirements in ultra-reliable-low-latency-communication (URLLC) traffic make the design of linear massive multipleinput-multiple-output (M-MIMO) receivers becomes very challenging. Recently, Bayesian concept has been used to increase the detection reliability in minimum-mean-square-error (MMSE) linear receivers. However, the latency processing time is a major concern due to the exponential complexity of matrix inversion operations in MMSE schemes. This paper proposes an iterative M-MIMO receiver that is developed by using a Bayesian concept and a parallel interference cancellation (PIC) scheme, referred to as a linear Bayesian learning (LBL) receiver. PIC has a linear complexity as it uses a combination of maximum ratio combining (MRC) and decision statistic combining (DSC) schemes to avoid matrix inversion operations. Simulation results show that the bit-error-rate (BER) and latency processing performances of the proposed receiver outperform the ones of MMSE and best Bayesian-based receivers by minimum 2 dB and 19 times for various M-MIMO system configurations.

A Bayesian Receiver with Improved Complexity-Reliability Trade-off in Massive MIMO Systems

“…APPENDIX A DERIVATION OF (8) The 𝑘-th symbol estimate from PIC scheme 𝑥 (𝑡) PIC,𝑘 can be expanded by substituting (1)…”

“…Considerable efforts have been made to develop detectors that can approach the MMSE performance with a much lower complexity 1 . In [13] and [14], the approximate matrix inverse (AMI) detectors based on the conventional inverse approximations such as Neumann series, Richardson, successive-over relaxation (SOR), Gauss-Seidel (GS), optimized coordinate descent (OCD), Jacobi, and conjugate gradient (CG) methods have been investigated.…”

“…A low complexity detector [16], referred to as the Huber fitting based alternating direction method of multipliers (HF-ADMM) detector, was shown to provide near MMSE performance. Several low complexity detectors [17]- [19], referred to as the PIC-DSC detectors, can achieve a near MMSE performance but offer a linear complexity with respect to the number of 1 Due to the space limitation, we do not provide the details of approximate MMSE detectors except for the PIC-DSC detector scheme, which we use to cancel the user interference in the proposed detector. However, interested readers could see their details in [13]- [16].…”

“…𝑇 are the estimated symbols in the (𝑡 − 1)-th iteration. At the first iteration, we initialise x (0) PIC,𝑘 = 0 indicating that the PIC is inactive and thus we have 𝑥 (1) PIC,𝑘 = h 𝐻 𝑘 y h 𝑘 2 , which is the expression of the matched filter [42].…”

“…, c[Θ]). Note that a random interleaver was used to obtain numerical results for Section VII-F. Then, a modulator maps groups of 𝑚 bits of the interleaved codeword A received signal block y[𝜃] corresponding to transmitted block x[𝜃] is described by (1), where the MIMO channel in the 𝜃-th time slot is characterized by the 𝑁 × 𝐾 matrix H[𝜃]. The signal blocks y[𝜃], 1 ≤ 𝜃 ≤ Θ, are independently processed by the B-PIC-DSC detector, which is illustrated in Fig.…”