Derandomized sampling algorithm for lattice decoding

Abstract

The sampling decoding algorithm randomly samples lattice points and selects the closest one from the candidate list. Although it achieves a remarkable performance gain with polynomial complexity, there are two inherent issues due to random sampling, namely, repetition and missing of certain lattice points. To address these issues, a derandomized algorithm of sampling decoding is proposed with further performance improvement and complexity reduction. Given the sample size K, candidates are deterministically sampled if their probabilities P satisfy the threshold PK ≥ 1/2. By varying K, the decoder with low complexity enjoys a flexible performance between successive interference cancelation (SIC) and maximum-likelihood (ML) decoding.