Low-complexity large MIMO detection via layered belief propagation in beam domain
| Autor: | Takumi Takahashi, Shinsuke Ibi, Antti Tolli, Seiichi Sampei |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2022 |
| Předmět: |
Signal processing
Computational complexity theory Computer science Applied Mathematics Detector MIMO 020206 networking & telecommunications 02 engineering and technology Belief propagation Computer Science Applications Interference (communication) iterative signal detection large MIMO detection 0202 electrical engineering electronic engineering information engineering Bit error rate beam-domain signal processing domain conversion Electrical and Electronic Engineering Algorithm Computer Science::Information Theory Communication channel belief propagation dimensionality reduction |
| Popis: | In large multi-user multi-input multi-output systems, the computational cost and circuit scale of base stations (BSs) are effectively reduced using two-stage signal processing consisting of a slow varying outer beamformer (OBF) based on long-term channel statistics and group-specific multi-user detection for instantaneous channel variations. However, the dimensionality reduction of the group-specific beam-domain channel based on the OBF causes significant performance degradation in the subsequent spatial-filtering detection. To compensate for this drawback, this paper introduces a novel layered belief propagation (BP) detector with a concatenated structure of beam- and antenna-domain BP layers for post-stage OBF processing. The proposed detector is designed for improving the convergence of iterative detection by suppressing intra- and inter-group interference in stages. The layered structure provides the advantages of both beam and antenna domains while maintaining low signal-processing complexity. Numerical results show the validity of our proposed method in terms of the bit error rate performance in both the uncoded and coded cases and the computational complexity. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|