Retrospective Interference Alignment: Degrees of Freedom Scaling With Distributed Transmitters

In this paper, we consider the K-user single-input single-output interference channel with delayed channel state information at the transmitters (CSIT). Our main contribution is to show that even with delayed CSIT and distributed transmitters, the achievable degrees of freedom (DoFs) still scale with the number of users. More specifically, we propose a method that achieves K/(2 root K - 1) >= root K/2 DoF. The main idea behind the proposed method is to use interference alignment (IA) at the receiver side in conjunction with repetition coding at the transmitters. Repetition coding repeats the data R times while IA makes the interference generated by a given transmitter identical: 1) at all non-intended receivers and 2) all transmission repetitions. Therefore, one retransmission per transmitter is sufficient to cancel all the interference. This reduces significantly the overhead required for interference removal since the number of interference terms is reduced from RK(K - 1) to K.