Asymptotic Performance Analysis of Large-Scale Active IRS-Aided Wireless Network

Yan Wang, Feng Shu, Zhihong Zhuang, Rongen Dong, Qi Zhang, Di Wu, Xuehui Wang, Liang Yang, Jiangzhou WangIEEE Open J. Commun. Soc 2023 · DOI

TL;DR

This paper analyzes the performance of large-scale active Intelligent Reflecting Surface (IRS) aided wireless networks, focusing on how the number of IRS elements and their reflective power impact signal-to-noise ratio (SNR). It introduces a new metric, average SNR at the IRS, and derives optimal power allocation strategies, demonstrating significant rate gains and quantifying performance losses due to finite phase shifters.

What Problem Does This Paper Address?

Traditional wireless communication systems are limited by their static environments. Intelligent Reflecting Surfaces (IRS) offer a promising solution by allowing wireless environments to be reconfigured. However, analyzing the performance of IRS-aided systems, especially with a large number of IRS elements (large-scale) and active IRS (which can amplify signals and introduce noise), presents significant challenges. Existing research often assumes ideal continuous phase shifts, neglecting the practical limitations of finite phase shifters and the associated quantization errors. This paper aims to bridge this gap by providing a comprehensive asymptotic performance analysis of large-scale active IRS-aided networks, considering practical aspects like finite phase shifters and their impact on SNR, achievable rate (AR), and bit error rate (BER).

Methods

The authors employ the weak law of large numbers to derive asymptotic expressions for the system's performance metrics as the number of IRS elements ($N$) becomes large. A key contribution is the definition of average SNR at the active IRS ($\gamma_0$), which is shown to be a dominant factor affecting the user's received SNR. They analyze the impact of IRS reflective power ($P_I$) and base station transmit power ($P_S$) on the received SNR, deriving optimal power allocation strategies under a total power constraint. To account for practical IRS implementations, the paper derives closed-form expressions for performance loss (PL) and approximate performance loss (APL) due to phase quantization errors introduced by finite phase shifters. These expressions are then analyzed using Taylor series expansions.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and simulation-based validation.

Key Insights and Ideas

  • Average IRS SNR ($\gamma_0$) is Crucial: The paper identifies $\gamma_0$ as a critical factor for system performance, directly influencing the user's received SNR. This metric captures the IRS's ability to reflect signals effectively while considering its own noise.
  • Active IRS Amplifies Both Signal and Noise: Unlike passive IRS, active IRS can amplify signals, but this also amplifies any noise present at the IRS. This leads to a trade-off: increasing IRS reflective power can improve SNR up to a point, after which it degrades performance due to amplified noise.
  • Optimal Reflective Power Exists: For a fixed base station transmit power, there's an optimal level of IRS reflective power that maximizes the received SNR. Exceeding this optimum yields diminishing returns and can even degrade performance.
  • Power Allocation Matters: Under a total power budget for the BS and IRS, a specific power allocation strategy can yield significant rate gains (up to 0.83 bits) compared to an equal power distribution.
  • Finite Phase Shifters Introduce Quantization Errors: Practical IRS implementations use finite phase shifters, leading to quantization errors. The paper quantifies the performance loss associated with these errors for SNR, AR, and BER.
  • 3-Bit Phase Shifters are Sufficient: The simulations suggest that using just 3-bit discrete phase shifters is enough to achieve a negligible performance loss in large-scale active IRS networks, making practical deployment more feasible.

Key Results

  • The user's received SNR is shown to be a linear increasing function of the product of the number of IRS elements ($N$) and the average IRS SNR ($\gamma_0$) for large-scale IRS.
  • An optimal IRS reflective power ($P_I^{opt}$) is derived, which maximizes the received SNR for a fixed BS transmit power ($P_S$).
  • An optimal power allocation strategy under a sum constraint ($P_I + P_S = P_T$) achieves a rate gain of up to 0.83 bit over equal power allocation.
  • Performance loss analysis shows that for active IRS, 3-bit phase shifters result in SNR and AR losses less than 0.22 dB and 0.08 bits/s/Hz, respectively, with negligible BER performance loss.

Limitations and Future Work

  • Rayleigh Fading Channel Assumption: The analysis is primarily based on Rayleigh fading channels. Future work could extend this to other channel models, such as Rician fading, which might be more representative of certain scenarios.
  • Single-Antenna BS and User: The system model assumes single antennas for the BS and user. Analyzing systems with multiple antennas would be a valuable extension.
  • Ideal Channel State Information (CSI) for BS-IRS and IRS-User: While the paper discusses CSI estimation challenges in the introduction, the core analysis assumes perfect CSI for the BS-IRS and IRS-user links. Investigating the impact of imperfect CSI would be a practical next step.
  • Focus on Asymptotic Performance: The paper emphasizes asymptotic analysis for large $N$. While this provides valuable insights into the behavior of large-scale systems, analyzing the performance for smaller $N$ might also be relevant for certain applications.

Summary updated

BibTeX
@article{wang2023asymptotic,
  author = {Yan Wang and Feng Shu and Zhihong Zhuang and Rongen Dong and Qi Zhang and Di Wu and Xuehui Wang and Liang Yang and Jiangzhou Wang},
  title = {{Asymptotic Performance Analysis of Large-Scale Active IRS-Aided Wireless Network}},
  journal = {IEEE Open J. Commun. Soc.},
  year = {2023},
  doi = {10.1109/ojcoms.2023.3324064},
}
Figure