Superimposed CSI Feedback Assisted by Inactive Sensing Information

Mintao Zhang, Haowen Jiang, Zilong Wang, Linsi He, Yuqiao Yang, Mian Ye, Chaojin QingSensors 2025 · DOI

TL;DR

This paper introduces a novel method for improving the accuracy of channel state information (CSI) feedback in massive MIMO systems by leveraging previously unused "inactive" sensing data (like location and speed) from devices. This approach effectively suppresses interference introduced by superimposed CSI feedback, leading to better recovery of both downlink CSI and uplink data without requiring extra hardware.

What Problem Does This Paper Address?

Massive MIMO systems rely heavily on accurate channel state information (CSI) for efficient communication. In frequency division duplex (FDD) systems, this CSI needs to be fed back from the user equipment (UE) to the base station (BS). To reduce the significant bandwidth overhead of this feedback, a technique called "superimposed CSI feedback" is used, where CSI is embedded within uplink data. However, this superposition introduces interference that degrades the accuracy of both the recovered CSI and the uplink data. While machine learning (ML) methods have shown promise in mitigating this interference, they often require complex interactions and significant system resources, contradicting the goal of reducing overhead. This paper addresses the need for a simpler, more resource-efficient method to improve superimposed CSI feedback accuracy by utilizing readily available, but previously ignored, sensing information.

Methods

The core innovation of this paper is the use of "inactive sensing information" to assist superimposed CSI feedback. This information, such as UE location, speed, and path, is typically collected by devices like GNSS chips but not actively used for communication. The proposed method extracts this sensing data at the BS and uses it to derive "prior information" about the communication channel.

Specifically, the method leverages:

  • Doppler Shift Prior: UE speed, derived from GNSS location data over time, is used to estimate the maximum Doppler shift. This information helps define a boundary in the delay-Doppler (DD) domain, aiding in denoising.
  • Path Delay Prior: Uplink CSI is used to estimate the delay domain characteristics of the channel. Due to path-delay reciprocity, this uplink information is then used as a prior for the downlink channel's delay support set.

With these priors, the paper proposes a denoising process applied to the downlink CSI in the delay-Doppler domain. This denoising aims to suppress interference and improve the accuracy of the recovered CSI and subsequently the detected uplink data. The method is presented as an enhancement that can be integrated with existing CSI recovery techniques, including ML-based ones, without requiring additional hardware at the BS.

Datasets

Not discussed in the paper. The paper relies on simulated data generated using the extended vehicular channel model-A (EVA) and a classic Jakes spectrum for Doppler shifts.

Key Insights and Ideas

  • Unlocking "Inactive" Data: The paper demonstrates that readily available sensing data, often overlooked in communication systems, can be a powerful resource for improving communication performance. This opens up new avenues for integrated sensing and communication.
  • Prior Information for Denoising: Using sensing data to create prior information in the delay-Doppler domain is a key enabler for effective interference suppression. This prior information acts as a guide for denoising, leading to more accurate CSI recovery.
  • Hardware-Agnostic Enhancement: The proposed method doesn't require new hardware at the BS, making it a practical and cost-effective upgrade for existing systems.
  • Synergy with ML: The method is designed to be compatible with ML-based approaches, suggesting a path towards even more sophisticated and efficient CSI feedback systems.

Key Results

The paper presents simulation results showing significant performance improvements:

  • Reduced NMSE: The proposed method ("Prop") consistently achieves lower Normalized Mean Squared Error (NMSE) for downlink CSI recovery compared to existing methods like "Xucla," "ELM," and "LoSSen," across various Signal-to-Noise Ratio (SNR) levels.
  • Improved BER: The Bit Error Rate (BER) for uplink data detection is also significantly reduced by the proposed method, indicating better overall system performance.
  • Effectiveness of Priors: Comparing "Prop" with variants that use only delay or only Doppler prior information ("Prop (without delay)" and "Prop (without Doppler)"), the paper shows that fusing both types of prior information yields the best results.
  • Robustness: The method demonstrates robustness against variations in key parameters like the power proportional coefficient ($p$) and UE speed ($v$), maintaining its performance advantage. For instance, even with parameter variations, "Prop" consistently outperforms baseline methods.

Limitations and Future Work

  • On-Grid Assumption: The current work assumes on-grid channel parameters. In real-world scenarios, off-grid effects can occur, which are not addressed here.
  • Sensing Data Errors/Delays: The paper does not consider scenarios where sensing data might contain errors or delays.
  • Future Work: The authors propose to investigate sensing-assisted CSI feedback for off-grid conditions and to develop methods that incorporate GNSS error models to enhance CSI reconstruction accuracy. This would make the approach more practical for real-world applications.

Summary updated

Abstract

In massive multiple-input and multiple-output (mMIMO) systems, superimposed channel state information (CSI) feedback is developed to improve the occupation of uplink bandwidth resources. Nevertheless, the interference from this superimposed mode degrades the recovery performance of both downlink CSI and uplink data sequences. Although machine learning (ML)-based methods effectively mitigate superimposed interference by leveraging the multi-domain features of downlink CSI, the complex interactions among network model parameters cause a significant burden on system resources. To address these issues, inspired by sensing-assisted communication, we propose a novel superimposed CSI feedback method assisted by inactive sensing information that previously existed but was not utilized at the base station (BS). To the best of our knowledge, this is the first time that inactive sensing information is utilized to enhance superimposed CSI feedback. In this method, a new type of modal data, different from communication data, is developed to aid in interference suppression without requiring additional hardware at the BS. Specifically, the proposed method utilizes location, speed, and path information extracted from sensing devices to derive prior information. Then, based on the derived prior information, denoising processing is applied to both the delay and Doppler dimensions of downlink CSI in the delay-Doppler (DD) domain, significantly enhancing the recovery accuracy. Simulation results demonstrate the performance improvement of downlink CSI and uplink data sequences when compared to both classic and novel superimposed CSI feedback methods. Moreover, against parameter variations, simulation results also validate the robustness of the proposed method.
BibTeX
@inproceedings{zhang2025superimposed,
  author = {Mintao Zhang and Haowen Jiang and Zilong Wang and Linsi He and Yuqiao Yang and Mian Ye and Chaojin Qing},
  title = {{Superimposed CSI Feedback Assisted by Inactive Sensing Information}},
  booktitle = {Sensors},
  year = {2025},
  doi = {10.3390/s25196156},
}
Figure