Application of Reinforcement Learning and Deep Learning in Multiple-Input and Multiple-Output (MIMO) Systems

The current wireless communication infrastructure has to face exponential development in mobile traffic size, which demands high data rate, reliability, and low latency. MIMO systems and their variants (i.e., Multi-User MIMO and Massive MIMO) are the most promising 5G wireless communication systems technology due to their high system throughput and data rate. However, the most significant challenges in MIMO communication are substantial problems in exploiting the multiple-antenna and computational complexity. The recent success of RL and DL introduces novel and powerful tools that mitigate issues in MIMO communication systems. This article focuses on RL and DL techniques for MIMO systems by presenting a comprehensive review on the integration between the two areas. We first briefly provide the necessary background to RL, DL, and MIMO. Second, potential RL and DL applications for different MIMO issues, such as detection, classification, and compression; channel estimation; positioning, sensing, and localization; CSI acquisition and feedback, security, and robustness; mmWave communication and resource allocation, are presented.

Limited Feedback Techniques in Multiple Antenna Wireless Communication Systems

<p>Multiple antenna systems provide spatial multiplexing and diversity benefits.These systems also offer beamforming and interference mitigation capabilities in single-user (SU) and multi-user (MU) scenarios, respectively. Although diversity can be achieved without channel state information (CSI) at the transmitter using space-time codes, the knowledge of instantaneous CSI at the transmitter is essential to the above mentioned gains. In frequency division duplexing (FDD) systems, limited feedback techniques are employed to obtain CSI at the transmitter from the receiver using a low-rate link. As a consequence, CSI acquired by the transmitter in such manner have errors due to channel estimation and codebook quantization at the receiver, resulting in performance degradation of multi-antenna systems. In this thesis, we examine CSI inaccuracies due to codebook quantization errors and investigate several other aspects of limited feedback in SU, MU and multicell wireless communication systems with various channel models. For SU multiple-input multiple-output (MIMO) systems, we examine the capacity loss using standard codebooks. In particular, we consider single-stream and two-stream MIMO transmissions and derive capacity loss expressions in terms of minimum squared chordal distance for various MIMO receivers. Through simulations, we investigate the impact of codebook quantization errors on the capacity performance in uncorrelated Rayleigh, spatially correlated Rayleigh and standardized MIMO channels. This work motivates the need of effective codebook design to reduce the codebook quantization errors in correlated channels. Subsequently, we explore the improvements in the design of codebooks in temporally and spatially correlated channels for MU multiple-input single-output (MISO) systems, by employing scaling and rotation techniques. These codebooks quantize instantaneous channel direction information (CDI) and are referred as differential codebooks in the thesis. We also propose various adaptive scaling techniques for differential codebooks where packing density of codewords in the differential codebook are altered according to the channel condition, in order to reduce the quantization errors. The proposed differential codebooks improve the spectral efficiency of the system by minimizing the codebook quantization errors in spatially and temporally correlated channels. Later, we broaden the scope to massive MISO systems and propose trellis coded quantization (TCQ) schemes to quantize CDI. Unlike conventional codebook approach, the TCQ scheme does not require exhaustive search to select an appropriate codeword, thus reducing computational complexity and memory requirement at the receiver. The proposed TCQ schemes yield significant performance improvements compared to the existing TCQ based limited feedback schemes in both temporally and spatially correlated channels. Finally, we investigate interference coordination for multicell MU MISO systems using regularized zero-forcing (RZF) precoding. We consider random vector quantization (RVQ) codebooks and uncorrelated Rayleigh channels. We derive expected SINR approximations for perfect CDI and RVQ codebook-based CDI. We also propose an adaptive bit allocation scheme which aims to minimize the network interference and moreover, improves the spectral efficiency compared to equal bit allocation and coordinated zero-forcing (ZF) based adaptive bit allocation schemes.</p>