[Paper Review] Sphere-Packing Bound for Symmetric Classical-Quantum Channels

Hao–Chung Cheng, Min-Hsiu Hsieh+1 more|arXiv (Cornell University)|Jan 11, 2017

Wireless Communication Security Techniques26 references3 citations

TL;DR

This paper establishes a refined sphere-packing lower bound for symmetric classical-quantum channels, achieving a polynomial pre-factor in the finite-blocklength regime by leveraging a sharp concentration inequality and properties of the error-exponent function. The result matches the best-known random coding upper bound in the classical case, yielding exact asymptotics for the sphere-packing bound in symmetric c-q channels.

ABSTRACT

We provide a sphere-packing lower bound for the optimal error probability in finite blocklengths when coding over a symmetric classical-quantum channel. Our result shows that the pre-factor can be significantly improved from the order of the subexponential to the polynomial. The established pre-factor is essentially optimal because it matches the best known random coding upper bound in the classical case. Our approaches rely on a sharp concentration inequality in strong large deviation theory and crucial properties of the error-exponent function.

Motivation & Objective

• To derive a finite-blocklength sphere-packing lower bound for symmetric classical-quantum channels with improved pre-factor scaling.
• To close the gap between sphere-packing lower bounds and random coding upper bounds in the quantum setting by achieving a polynomial pre-factor.
• To extend classical finite-blocklength results—particularly those of Altuğ and Wagner—to the quantum regime using strong large deviation techniques.
• To establish exact asymptotics for the error probability in symmetric c-q channels by matching the pre-factor to the best-known classical upper bound.
• To provide a framework applicable to the medium error probability regime in quantum communication.

Proposed method

• Utilizes a sharp concentration inequality from Bahadur and Ranga Rao (1960) to control tail probabilities in the strong large deviation regime.
• Applies the Fenchel-Legendre transform to relate the cumulant generating function of the log-likelihood ratio to the error exponent function.
• Employs the sphere-packing exponent function $ E_{ ext{sp}}(R) $ as the central object, leveraging its strict concavity and differentiability properties.
• Introduces a dual optimization framework involving the Legendre transform of the cumulant generating function $ ilde{ heta}_{P_{oldsymbol{x}^n}}(t) $, which characterizes the optimal trade-off between rate and error probability.
• Uses the symmetry of the channel to ensure that the optimal dual variable $ t^* eq 0,1 $, enabling strict concavity and uniqueness of the solution.
• Derives the pre-factor in the bound by analyzing the second derivative $ ilde{ heta}''_{P_{oldsymbol{x}^n}}(t) > 0 $, ensuring the existence of a unique maximizer in the optimization.

Experimental results

Research questions

• RQ1 Can the pre-factor in the sphere-packing bound for symmetric classical-quantum channels be improved from subexponential to polynomial order?
• RQ2 Does the sphere-packing lower bound in the quantum regime match the best-known random coding upper bound in the classical case?
• RQ3 What is the exact asymptotic behavior of the optimal error probability in symmetric c-q channels at finite blocklength?
• RQ4 How do strong large deviation techniques and error-exponent function properties enable tighter finite-blocklength bounds in quantum channels?
• RQ5 Can the framework developed for symmetric c-q channels be extended to general classical-quantum channels?

Key findings

• The pre-factor in the sphere-packing bound is improved from subexponential $ ext{subexp}(-O( ilde{n}^{1/2})) $ to polynomial order, significantly tightening the bound at rates near capacity.
• The derived lower bound matches the best-known random coding upper bound in the classical case, implying that the pre-factor is essentially optimal.
• The error exponent function $ E_{ ext{sp}}(R) $ is strictly positive and differentiable, ensuring the existence of a unique optimizer in the dual formulation.
• The second derivative $ ilde{ heta}''_{P_{oldsymbol{x}^n}}(t) > 0 $ for all $ t o [0,1] $, which guarantees strict concavity and uniqueness of the solution in the Fenchel-Legendre transform.
• The bound recovers the classical result of Altuğ and Wagner for symmetric classical channels, including the binary symmetric and binary erasure channels.
• The exact asymptotics of the optimal error probability are established for symmetric c-q channels, with the pre-factor matching the classical upper bound.