Code-Aware Power Allocation for Irregular LDPC Codes

Abstract

In this paper, we investigate a code-dependent unequal power allocation method for Gaussian channels using irregular LDPC codes. This method allocates the power for each set of coded bits depending on the degree of their equivalent variable nodes. We propose a new algorithm to optimize the power allocation vector using density evolution algorithm under the Gaussian approximation. We show that unequal power allocation can bring noticeable gains on the threshold of some irregular LDPC codes with respect to the classical equal power allocation method depending on the code and the maximum number of decoding iterations.

This work was carried out within the framework of Celtic-Plus SHARING project.