Progressive Rate-Filling: A Framework for Agile Construction of Multilevel Polar-Coded Modulation

Abstract

We propose the progressive rate-filling method as a framework to study agile construction of multilevel polar-coded modulation. We show that the bit indices within each component polar code can follow a fixed, precomputed ranking sequence (e.g., the Polar sequence in the 5G standard) while their allocated rates (i.e., the number of information bits of each component polar code) can be fast computed by exploiting the target sum-rate approximation and proper rate-filling methods. In particular, we propose two rate-filling methods based on the capacity and the rate considering the finite blocklength effect. The proposed construction methods can be performed independently of the actual channel condition with ${O(m)}$ ( ${m}$ denotes the modulation order) complexity and robust to diverse modulation and coding schemes (MCSs) in the 5G standard, which is a desired feature for practical systems.