Polar alignment for interference networks

Polar coding has originally been introduced as a capacity achieving low complexity code for binary input symmetric channels. Polar codes can be understood as transformations that replace a probabilistic channel with parallel deterministic counterparts. This paper builds on this interpretation of polar codes, using it to perform alignment over the resulting deterministic channels to obtain gains for interference networks. It is important to note here that polar codes are not chosen with encoding and decoding complexity in mind, which is just a fortuitous side-benefit, but with the aim of transforming the original channels into a class of deterministic parallel channels over which interference-alignment is well-understood. A degraded one-sided interference network is chosen as the illustrative example. Polar alignment is shown to increase the achievable sum rate over known random coding schemes. The paper concludes with a brief discussion of possible extensions.