Polar Coding Strategies for the Interference Channel with Partially-Joint Decoding

Abstract: Existing polar coding schemes for the two-user interference channel follow the original idea of Han and Kobayashi, in which component messages are encoded independently and then mapped by some deterministic functions (i.e., homogeneous superposition coding). In this paper, we propose a new polar coding scheme for the interference channel based on the heterogeneous superposition coding approach of Chong, Motani and Garg. We prove that fully-joint decoding (the receivers simultaneously decode both senders' common messages and the intended sender's private message) in the Han-Kobayashi strategy can be simplified to two types of partially-joint decoding, which are friendly to polar coding with practical decoding algorithms. The proposed coding scheme requires less auxiliary random variables and no deterministic functions, and can be efficiently constructed. Further, we extend this result to interference networks and show that the proposed partially-joint decoding scheme is a general method for designing heterogeneous superposition polar coding schemes in interference networks.