References & Further Reading

References

T. L. Marzetta, Noncooperative Cellular Wireless with Unlimited Numbers of Base Station Antennas, 2010
The foundational massive MIMO paper. Marzetta introduces the concept of using unlimited numbers of BS antennas with TDD operation, establishing the pilot contamination limit. The paper explicitly argues that FDD is impractical for massive arrays — the overhead bottleneck analyzed in Section 1 of this chapter.
T. L. Marzetta, E. G. Larsson, H. Yang, H. Q. Ngo, Fundamentals of Massive MIMO, Cambridge University Press, 2016
The definitive textbook on massive MIMO fundamentals. Chapters 1–2 provide the TDD vs FDD comparison and pilot overhead analysis. Chapter 7 covers achievable rates with imperfect CSI. Essential background for this chapter.
E. Björnson, J. Hoydis, L. Sanguinetti, Massive MIMO Networks: Spectral, Energy, and Hardware Efficiency, Now Publishers, 2017
Comprehensive treatment of massive MIMO system design. Section 2.3 analyzes pilot and feedback overhead. The use-and-then-forget (UatF) bound framework provides closed-form rate expressions with imperfect CSI.
A. Adhikary, J. Nam, J.-Y. Ahn, G. Caire, Joint Spatial Division and Multiplexing—The Large-Scale Array Regime, 2013
The JSDM paper — the CommIT contribution central to Section 5. Introduces two-stage precoding, group-based dimensionality reduction, and proves that JSDM achieves the full multiplexing gain while reducing FDD overhead from $N_t$ to $r_g$ dimensions. Contains the overhead analysis and rate expressions used throughout this chapter.
D. J. Love, R. W. Heath Jr., V. K. N. Lau, D. Gesbert, B. D. Rao, M. Andrews, An Overview of Limited Feedback in Wireless Communication Systems, 2008
The definitive survey on limited feedback for MIMO. Covers codebook design (Grassmannian, DFT), quantization error analysis, rate loss bounds, and multiuser feedback. Essential reading for the codebook material in Section 3.
N. Jindal, MIMO Broadcast Channels with Finite-Rate Feedback, 2006
Establishes the rate loss scaling law $\Delta R \propto 2^{-B/(N_t-1)}$ for MIMO broadcast channels with codebook feedback. The key result used in Theorem 3 (rate loss from codebook quantization).
A. M. Sayeed, Deconstructing Multiantenna Fading Channels, 2002
Introduces the virtual (angular-domain) channel representation that is the foundation for compressed CSI feedback. The DFT-based angular decomposition and the concept of channel sparsity in the beamspace are developed here.
X. Rao, V. K. N. Lau, Distributed Compressive CSIT Estimation and Feedback for FDD Multi-User Massive MIMO Systems, 2014
Develops the compressed sensing framework for CSI feedback in massive MIMO. Uses random projections at the UE and sparse recovery at the BS. The recovery guarantees via RIP (Theorem 2 in Section 2) are based on this work.
C.-K. Wen, W.-T. Shih, S. Jin, Deep Learning for Massive MIMO CSI Feedback, 2018
The CsiNet paper — the first deep learning autoencoder for CSI feedback. Demonstrates 5–10 dB NMSE improvement over compressed sensing at low compression ratios. The architecture (FC encoder, RefineNet decoder) and training methodology described in Section 4 are from this paper.
J. Guo, C.-K. Wen, S. Jin, G. Y. Li, Overview of Deep Learning-Based CSI Feedback in Massive MIMO Systems, 2022
Comprehensive survey of deep learning CSI feedback methods post-CsiNet. Covers CRNet, TransNet, quantization-aware training, and the rate-distortion interpretation. Also discusses the generalization challenge across environments.
3GPP, NR; Physical layer procedures for data, 2023. [Link]
The 5G NR specification for CSI reporting. Section 5.2 defines the CSI-RS framework, Type I and Type II codebooks, PMI/RI/CQI reporting, and codebook restriction. The primary reference for the 5G NR material in Section 3.
3GPP, Study on Artificial Intelligence (AI)/Machine Learning (ML) for NR air interface, 2023. [Link]
The 3GPP Release 18 study item on AI/ML for the NR air interface. Includes evaluation of AI-based CSI feedback (use case 2), with training-deployment mismatch as a key challenge. Referenced in the historical note on AI for physical layer.