IEEE Wireless Communications Letters
RV-QUBO: Reduced-Variable QUBO for Decoding Short Systematic Polar Codes
Huayi Zhou, Yixiao Liu, Warren J. Gross, Chuan Zhang
Abstract
Quantum annealer decoders have been proposed to unlock new potential by addressing channel decoding problems in quadratic unconstrained binary optimization (QUBO) form. However, the variables required by the existing node constraints in polar-QUBO decoding increase rapidly with the code length, resulting in prohibitively high computational complexity. In this letter, we propose reduced-variable QUBO (RV-QUBO) for short systematic polar codes, based on three key steps:1) applying equivalent generator matrix constraintsto replace the node constraints;2) introducing a common subexpression elimination approachto identify and replace recurringxorsubexpressions with new variables;3) imposing an extra length restriction on eachxorexpressionto balance the error-correction performance and the variable count. Numerical results indicate that the proposed scheme achieves a 73.60% reduction in variable count for$\mathcal {P}(32,28)$, while reducing the required iterations for convergence by 66.7%.