[Paper Review] Cascaded Channel Estimation for Large Intelligent Metasurface Assisted Massive MIMO

Zhen-Qing He, Xiaojun Yuan|arXiv (Cornell University)|May 20, 2019

Advanced Wireless Communication Technologies25 references36 citations

TL;DR

The paper proposes a two-stage JBF-MC algorithm to estimate the cascaded BS-LIM and LIM-user channels in LIM-assisted massive MIMO by combining sparse matrix factorization (BiG-AMP) and matrix completion (Riemannian gradient).

ABSTRACT

In this letter, we consider the problem of channel estimation for large intelligent metasurface (LIM) assisted massive multiple-input multiple-output (MIMO) systems. The main challenge of this problem is that the LIM integrated with a large number of low-cost metamaterial antennas can only passively reflect the incident signal by a certain phase shift, and does not have any signal processing capability. To deal with this, we introduce a general framework for the estimation of the transmitter-LIM and LIM-receiver cascaded channel, and propose a two-stage algorithm that includes a sparse matrix factorization stage and a matrix completion stage. Simulation results illustrate that the proposed method can achieve accurate channel estimation for LIM-assisted massive MIMO systems.

Motivation & Objective

• Motivate and enable accurate CSI acquisition in LIM-assisted massive MIMO where the LIM elements are passive and cannot process signals.
• Formulate the cascaded BS-LIM and LIM-user channels as a bilinear factorization with sparsity and rank-deficiency.
• Develop a two-stage algorithm to recover G and H through sparse matrix factorization and matrix completion.
• Evaluate performance gains over baseline methods through simulations.

Proposed method

• Formulate Y = G diag(s) H X + W and then Y = G Z + W with Z = S odot (H X) to enable bilinear factorization.
• Design training signals with a random Bernoulli S for sparsity in Z and a full-rank X to aid factorization.
• Apply BiG-AMP for sparse matrix factorization to estimate G and Z from Y.
• Apply Riemannian gradient-based matrix completion to recover H from Z, leveraging the rank-deficient property of H.
• Compute H as ind(rom the completed A: ormat H = rom A times X^†.
• Provide computational complexity notes: BiG-AMP dominates with O(L N T) per iteration; RGrad costs O(r N T) per iteration.

Experimental results

Research questions

• RQ1 How can the cascaded BS-LIM and LIM-user channels be accurately estimated when the LIM is fully passive?
• RQ2 Can a two-stage approach combining sparse matrix factorization and matrix completion recover the cascaded channels effectively?
• RQ3 What training signal design (S and X) enables reliable factorization and completion for LIM-assisted MIMO?
• RQ4 What are the performance gains of the proposed JBF-MC algorithm versus baseline methods in NMSE under varying SNR and pilot counts?

Key findings

• The JBF-MC algorithm achieves accurate cascaded channel estimation for LIM-assisted massive MIMO systems.
• BiG-AMP-based sparse factorization effectively estimates G and Z from Y, outperforming K-SVD and SPAMS baselines in simulations.
• Matrix completion via RGrad exploits the rank-deficient H to recover H from Z, improving estimation of the LIM-user channel.
• NMSE gains are significant, especially for the G estimation, across SNR and pilot configurations.
• There is a trade-off in sparsity level: too small hinders matrix completion and too large hinders factorization (phase transitions shown).