A Bayesian Receiver With Improved Complexity-Reliability Trade-Off in Massive MIMO Systems

Kosasih, Alva; Miloslavskaya, Vera; Hardjawana, Wibowo; She, Changyang; Wen, Chao-Kai; Vucetic, Branka

doi:10.1109/tcomm.2021.3092042

Cited by 11 publications

(11 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section we briefly review the conventional MP detectors, specifically, the EP [13] and the BPIC [14] detectors. Both detectors consist of two modules: the observation and estimation modules.…”

Section: Conventional Mp Detectorsmentioning

confidence: 99%

“…In this section, we explain the BPIC detector [14]. The BPIC detector does not need to perform any matrix inversion in contrast to the EP detector, and therefore it has substantially lower computational complexity.…”

Section: B the Bayesian Parallel Interference Cancellation Detectormentioning

confidence: 99%

“…The latter case is referred to as a high MUI scenario. For instance, the approximate message passing (AMP) [10] and the Bayesian parallel interference cancellation (BPIC) [14] detectors perform very poorly in the presence of illconditioned channel matrices and high MUI. The problem of ill-conditioned channel matrices has been partially resolved by the orthogonal AMP (OAMP) detector [12] that integrates the AMP with the linear MMSE filtering.…”

Section: Introductionmentioning

confidence: 99%

“…We refer to the newly developed detector as the graph EP network (GEPNet) detector. We then demonstrate how the proposed GNN-based framework can also be used to improve the recently proposed low-complexity BPIC detector [14]. We refer to this detector as graph parallel interference cancellation network (GPICNet) detector.…”

Section: Introductionmentioning

confidence: 99%

“…The permutation equivariance paves a way to the robustness to the changes in the number of users, which allows the proposed detectors to be trained only once even in the systems with dynamic changes of the number of users. Simulation results show that the GEPNet detector is able to significantly outperform the EP [13], RE-MIMO [20], OAMPNet [16], GNN [21], OAMP [12], AMP [10], BPIC [14] detectors by more than 4 dB at the SER of 10 −4 for 64 × 64 MU-MIMO configuration and achieve a near ML performance for various system configurations. The GPICNet detector substantially improves the performance and doubles the multiplexing gain of its predecessor BPIC detector.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Graph Neural Network Aided MU-MIMO Detectors

Kosasih¹,

Onasis²,

Miloslavskaya³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Multi-user multiple-input multiple-output (MU-MIMO) systems can be used to meet high throughput requirements of 5G and beyond networks. A base station serves many users in an uplink MU-MIMO system, leading to a substantial multi-user interference (MUI). Designing a high-performance detector for dealing with a strong MUI is challenging. This paper analyses the performance degradation caused by the posterior distribution approximation used in the state-of-the-art message passing (MP) detectors in the presence of high MUI. We develop a graph neural network based framework to finetune the MP detectors' cavity distributions and thus improve the posterior distribution approximation in the MP detectors. We then propose two novel neural network based detectors which rely on the expectation propagation (EP) and Bayesian parallel interference cancellation (BPIC), referred to as the GEPNet and GPICNet detectors, respectively. The GEPNet detector maximizes detection performance, while GPICNet detector balances the performance and complexity. We provide proof of the permutation equivariance property, allowing the detectors to be trained only once, even in the systems with dynamic changes of the number of users. The simulation results show that the proposed GEPNet detector performance approaches maximum likelihood performance in various configurations and GPICNet detector doubles the multiplexing gain of BPIC detector.

show abstract

Section: Conventional Mp Detectorsmentioning

confidence: 99%

Section: B the Bayesian Parallel Interference Cancellation Detectormentioning

confidence: 99%