Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FDD CSI Feedback under Finite Downlink Training: A Rate-Distortion Perspective

Shuao Chen, Junyuan Gao, Yuxuan Shi, Yongpeng Wu, Giuseppe Caire, H. Vincent Poor, Wenjun Zhang

TL;DR

This work tackles the fundamental limits of CSI feedback in FDD-OFDM systems under finite-length downlink training by formulating a unified rate-distortion function (RDF) for the end-to-end feedback pipeline. By modeling the UE’s MMSE channel estimate as an intermediate source and using remote source coding principles, the authors derive a closed-form RDF $R_{\mathbf{H},\mathbf{S}|\mathbb{P}}(d)$ that decomposes into a direct RDF plus two pilot-induced terms, $\Delta R_{\mathbb{P},\mathbf{S}}$ and $\Delta R_{\mathbb{P},d}$, which capture the benefits and costs of estimation and a stricter distortion constraint. They establish non-asymptotic bounds and asymptotic scaling with training length $n_t$, showing that the overall RDF approaches the direct RDF at a rate proportional to $1/n_t$, even when the downlink SNR is fixed. Simulations corroborate the bounds’ tightness and reveal a meaningful RDF gap under very limited training, highlighting the importance of training-design in practical FDD systems.

Abstract

This paper establishes the theoretical limits of channel state information (CSI) feedback in frequency-division duplexing (FDD) multi-antenna orthogonal frequency-division multiplexing (OFDM) systems under finite-length training with Gaussian pilots. The user employs minimum mean-squared error (MMSE) channel estimation followed by asymptotically optimal uplink feedback. Specifically, we derive a general rate-distortion function (RDF) of the overall CSI feedback system. We then provide both non-asymptotic bounds and asymptotic scaling for the RDF under arbitrary downlink signal-to-noise ratio (SNR) when the number of training symbols exceeds the antenna dimension. A key observation is that, with sufficient training, the overall RDF converges to the direct RDF corresponding to the case where the user has full access to the downlink CSI. More importantly, we demonstrate that even at a fixed downlink SNR, the convergence rate is inversely proportional to the training length. The simulation results show that our bounds are tight, and under very limited training, the deviation between the overall RDF and the direct RDF is substantial.

FDD CSI Feedback under Finite Downlink Training: A Rate-Distortion Perspective

TL;DR

This work tackles the fundamental limits of CSI feedback in FDD-OFDM systems under finite-length downlink training by formulating a unified rate-distortion function (RDF) for the end-to-end feedback pipeline. By modeling the UE’s MMSE channel estimate as an intermediate source and using remote source coding principles, the authors derive a closed-form RDF that decomposes into a direct RDF plus two pilot-induced terms, and , which capture the benefits and costs of estimation and a stricter distortion constraint. They establish non-asymptotic bounds and asymptotic scaling with training length , showing that the overall RDF approaches the direct RDF at a rate proportional to , even when the downlink SNR is fixed. Simulations corroborate the bounds’ tightness and reveal a meaningful RDF gap under very limited training, highlighting the importance of training-design in practical FDD systems.

Abstract

This paper establishes the theoretical limits of channel state information (CSI) feedback in frequency-division duplexing (FDD) multi-antenna orthogonal frequency-division multiplexing (OFDM) systems under finite-length training with Gaussian pilots. The user employs minimum mean-squared error (MMSE) channel estimation followed by asymptotically optimal uplink feedback. Specifically, we derive a general rate-distortion function (RDF) of the overall CSI feedback system. We then provide both non-asymptotic bounds and asymptotic scaling for the RDF under arbitrary downlink signal-to-noise ratio (SNR) when the number of training symbols exceeds the antenna dimension. A key observation is that, with sufficient training, the overall RDF converges to the direct RDF corresponding to the case where the user has full access to the downlink CSI. More importantly, we demonstrate that even at a fixed downlink SNR, the convergence rate is inversely proportional to the training length. The simulation results show that our bounds are tight, and under very limited training, the deviation between the overall RDF and the direct RDF is substantial.
Paper Structure (13 sections, 5 theorems, 45 equations, 2 figures)

This paper contains 13 sections, 5 theorems, 45 equations, 2 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Downlink Channel Training
  4. Channel Estimation and Feedback
  5. Main Results
  6. Derivation of RDF
  7. RDF Decrease from Compressing the MMSE Estimate
  8. RDF Increase from a Stricter Distortion Constraint
  9. Impact of Finite Training on the Overall Performance
  10. Conclusion
  11. Proof of Theorem \ref{['thm:RDF_main']}
  12. Proof of Theorem \ref{['thm:RDF_decrease']}
  13. Proof of Theorem \ref{['thm:RDF_increase']}

Key Result

Theorem 1

Consider a downlink channel vector $\mathbf{H} \sim \mathcal{CN}(\mathbf{0}, \mathsf{C}_{\mathbf{H}})$ with arbitrary spatial correlation among $m_{t}$ antennas and frequency correlation across $n_{c}$ subcarriers. Using the Gaussian ensemble pilot matrix $\mathbb{P}$ over $n_{t}$ training transmiss where and Here we let $\mathbb{A} = \mathsf{C}_{\mathbf{H}}^{1/2}\mathbb{P}$, $r_{H} = \mathsf{r}

Figures (2)

  • Figure 1: System model for CSI feedback.
  • Figure 2: Relationship between downlink training length and the overall RDF of the CSI feedback system.

Theorems & Definitions (7)

  • Remark 1: Availability of channel statistics
  • Theorem 1
  • Remark 2: Range of the imposed distortion constraint
  • Proposition 1
  • Theorem 2
  • Theorem 3
  • Theorem 4