[Paper Review] Distributed Beam Training for Intelligent Reflecting Surface Enabled Multi-Hop Routing

Weidong Mei, Rui Zhang|arXiv (Cornell University)|Jun 22, 2021

Advanced Wireless Communication Technologies11 references4 citations

TL;DR

This paper proposes a distributed beam training scheme for intelligent reflecting surface (IRS)-assisted multi-hop wireless networks, enabling joint active/passive beamforming and optimal path selection without exhaustive channel estimation. By leveraging time-invariant channel properties and cooperative training among controllers, the method achieves end-to-end channel gains close to sequential beam search but with significantly reduced training overhead and complexity.

ABSTRACT

Intelligent reflecting surface (IRS) is an emerging technology to enhance the spectral and energy efficiency of wireless communications cost-effectively. This letter considers a new multi-IRS aided wireless network where a cascaded line-of-sight (LoS) link is established between the base station (BS) and a remote user by leveraging the multi-hop signal reflection of selected IRSs. As compared to the conventional single-/double-hop IRS system, multi-hop IRS system provides more pronounced path diversity and cooperative passive beamforming gains, especially in the environment with dense obstacles. However, a more challenging joint active/passive beamforming and multi-hop beam routing problem also arises for maximizing the end-to-end channel gain. Furthermore, the number of IRS-associated channel coefficients increases drastically with the number of IRS hops. To tackle the above issues, in this letter we propose a new and efficient beam training based solution by considering the use of practical codebook-based BS/IRS active/passive beamforming without the need of explicit channel estimation. Instead of exhaustively or sequentially searching over all combinations of active and passive beam patterns for each beam route, a distributed beam training scheme is proposed to reduce the complexity, by exploiting the (nearly) time-invariant BS-IRS and inter-IRS channels and the cooperative training among the BS and IRSs' controllers. Simulation results show that our proposed design achieves the end-to-end channel gain close to that of the sequential beam search, but at a much lower training overhead and complexity.

Motivation & Objective

• To address the high complexity of joint active/passive beamforming and multi-hop beam routing in multi-IRS systems.
• To eliminate the need for exhaustive or sequential beam searches by avoiding explicit channel estimation.
• To reduce training overhead in multi-hop IRS networks with large numbers of IRSs and beam patterns.
• To enable practical deployment through codebook-based beamforming and cooperative controller training.
• To achieve near-optimal end-to-end channel gain with low computational and time complexity.

Proposed method

• The system uses predefined active/passive beamforming codebooks at the BS and each IRS, limiting beam patterns to a finite set.
• A distributed beam training protocol is designed, where BS and IRS controllers exchange local beam routing tables (BRTs) to build a global BRT.
• The global BRT enables efficient multi-hop path selection and joint beamforming design without full CSI acquisition.
• The method exploits (nearly) time-invariant BS-IRS and inter-IRS channels to reduce training overhead.
• Beam training is performed cooperatively among controllers, minimizing redundant or exhaustive searches.
• The approach uses DFT-based codebooks for beam pattern generation at the BS and IRSs, with 16 and 32 patterns respectively.

Experimental results

Research questions

• RQ1 Can a distributed beam training scheme achieve near-optimal end-to-end channel gain in multi-IRS multi-hop systems without exhaustive beam search?
• RQ2 How does the performance of the proposed distributed beam training compare to sequential beam search in terms of channel gain and training overhead?
• RQ3 What is the impact of IRS element count and Rician factor on the performance gap between the proposed method and sequential beam search?
• RQ4 How does path length (number of hops) affect the approximation error and performance of the distributed training scheme?
• RQ5 To what extent do scattered multipath components degrade system performance under the optimized beamforming and routing?

Key findings

• The proposed distributed beam training achieves end-to-end channel power gain within a small gap (less than 1.5 dB) of the sequential beam search across all tested configurations.
• The performance gap increases slightly when the number of IRS elements per dimension exceeds 22, due to higher hop counts in the optimal path.
• As the Rician factor increases to 15 dB or more, the performance gap between the proposed method and sequential beam search becomes negligible, indicating improved approximation accuracy.
• For user location 2, which is closer to the BS, the performance gap is smaller due to fewer hops in the optimal reflection path, reducing accumulated training error.
• The cascaded LoS channel power gain is comparable to the overall channel gain, indicating that scattered multipath components contribute negligibly to the total channel power.
• The proposed method reduces training overhead and complexity significantly compared to exhaustive or sequential beam search, while maintaining near-optimal performance.