Spherical codes generated by binary partitions of symmetric pointsets

Recently, Ericson and Zinoviev presented a clever, new construction for spherical codes for the Gaussian channel using ideas of code concatenation and set partitioning. This family of new spherical codes is generated from sets of binary codes using equally spaced symmetric pointsets on the real line. The family contains some of the best known spherical codes in terms of minimum distance. However, no efficient decoding algorithm is known for this new construction. In this paper, we present a new decoding algorithm for this family of spherical codes which is more efficient than maximum likelihood decoding.