Interference alignment via random codes and he capacity of a class of deterministic interference channels

We study the capacity of a class of deterministic discrete memoryless interference channels. Recent studies show that, in general, interference alignment is required to achieve capacity in interference channels. While interference alignment in general needs structured coding, we identify two scenarios where random coding achieves capacity and leads to single-letter capacity characterizations in deterministic interference channels. The first scenario is a noisy interference regime, where we show that interference can be aligned with random codes, obviating the need of structured codes. The second scenario is over the many-to-one interference channels where alignment is precluded because the multiple interferers are resolvable at the receiver which faces interference.