Improving the Performance of Nested Lattice Codes Using Concatenation

TL;DR

This paper proposes a concatenated coding scheme combining nested lattice codes with high-rate linear codes to achieve AWGN channel capacity efficiently, with reduced encoding and decoding complexity.

Contribution

It introduces a novel concatenation approach using lattice and linear codes to attain capacity with polynomial complexity and low error probability.

Findings

Capacity-achieving codes with $O(N^2)$ complexity

Further reduction to $O(N\log^2N)$ complexity using expander codes

Applicable to wiretap and compute-and-forward schemes

Abstract

A fundamental problem in coding theory is the design of an efficient coding scheme that achieves the capacity of the additive white Gaussian (AWGN) channel. The main objective of this short note is to point out that by concatenating a capacity-achieving nested lattice code with a suitable high-rate linear code over an appropriate finite field, we can achieve the capacity of the AWGN channel with polynomial encoding and decoding complexity. Specifically, we show that using inner Construction-A lattice codes and outer Reed-Solomon codes, we can obtain capacity-achieving codes whose encoding and decoding complexities grow as $O(N^2)$, while the probability of error decays exponentially in $N$, where $N$ denotes the blocklength. Replacing the outer Reed-Solomon code by an expander code helps us further reduce the decoding complexity to $O(N\log^2N)$. This also gives us a recipe for converting a high-complexity nested lattice code for a Gaussian channel to a low-complexity concatenated code without any loss in the asymptotic rate. As examples, we describe polynomial-time coding schemes for the wiretap channel, and the compute-and-forward scheme for computing integer linear combinations of messages.