[Paper Review] Outage Capacity of Rayleigh Product Channels: a Free Probability Approach

Zhong Zheng, Lu Wei+4 more|arXiv (Cornell University)|Feb 19, 2015

Advanced MIMO Systems Optimization41 references3 citations

TL;DR

This paper analyzes the outage capacity of Rayleigh product MIMO channels using free probability theory, deriving the asymptotic variance of channel capacity and proving its Gaussian distribution in the large-dimensional limit. The key contribution is a closed-form expression for the finite-SNR diversity-multiplexing tradeoff, validated numerically and shown to closely approximate practical systems with moderate antenna and scatterer counts.

ABSTRACT

The Rayleigh product channel model is useful in capturing the performance degradation due to rank deficiency of MIMO channels. In this paper, such a performance degradation is investigated via the channel outage probability assuming slowly varying channel with delay-constrained decoding. Using techniques of free probability theory, the asymptotic variance of channel capacity is derived when the dimensions of the channel matrices approach infinity. In this asymptotic regime, the channel capacity is rigorously proven to be Gaussian distributed. Using the obtained results, a fundamental tradeoff between multiplexing gain and diversity gain of Rayleigh product channels can be characterized by closed-form expression at any finite signal-to-noise ratio. Numerical results are provided to compare the relative outage performance between Rayleigh product channels and conventional Rayleigh MIMO channels.

Motivation & Objective

• To characterize the outage capacity of Rayleigh product channels under delay-constrained decoding, where ergodic capacity is not a relevant metric.
• To address the lack of analytical results for outage capacity in double-scattering (rank-deficient) MIMO channels, especially in the finite-SNR regime.
• To establish a rigorous asymptotic framework using free probability theory for large-dimensional random matrices.
• To derive a closed-form expression for the finite-SNR diversity-multiplexing tradeoff (DMT) in Rayleigh product channels.
• To validate the asymptotic results through numerical simulations showing good approximation even for moderate system dimensions.

Proposed method

• The authors model the channel as a product of two independent complex Gaussian matrices, representing a simplified double-scattering MIMO model.
• They apply second-order free probability tools, specifically the second-order Cauchy transform and R-transform, to derive the asymptotic variance of the channel capacity.
• A Central Limit Theorem (CLT) is proven for Linear Spectral Statistics (LSS) of the Rayleigh product ensemble, showing that capacity becomes asymptotically Gaussian distributed.
• The derivation leverages the convergence of the empirical spectral distribution (ESD) to the Marčenko-Pastur law and uses contour integration and resolvent techniques.
• The method conditions on the channel matrix structure and separates the randomness into two independent components: one from the product matrix and one from the input matrix.
• The resulting covariance structure of the capacity fluctuations is derived using complex analysis and resolvent identities, leading to a closed-form expression for the variance and the CLT.

Experimental results

Research questions

• RQ1 How does the outage capacity of Rayleigh product channels behave in the finite-SNR regime, especially under delay-constrained decoding?
• RQ2 What is the asymptotic distribution of the channel capacity when the dimensions of the channel matrices grow large?
• RQ3 Can a closed-form expression for the finite-SNR diversity-multiplexing tradeoff (DMT) be derived for Rayleigh product channels?
• RQ4 How do the second-order eigenvalue fluctuations of the Rayleigh product ensemble affect the capacity variance?
• RQ5 To what extent do the asymptotic results approximate practical MIMO systems with finite dimensions?

Key findings

• The channel capacity of Rayleigh product channels is rigorously proven to be asymptotically Gaussian distributed in the large-dimensional limit.
• The asymptotic variance of the capacity is derived in closed form using free probability theory and second-order Cauchy transform techniques.
• The capacity fluctuations follow a Central Limit Theorem (CLT) for Linear Spectral Statistics (LSS), generalizing prior results to generic analytic functions.
• The finite-SNR diversity-multiplexing tradeoff (DMT) is characterized by a closed-form expression, enabling precise performance tradeoff analysis.
• Numerical results confirm that the asymptotic results provide accurate approximations even for moderate system dimensions, such as those found in practical WLAN systems.
• The derived capacity variance and distribution are independent of the specific realization of the input matrix, relying only on the limiting spectral distribution of the product matrix.