Blind Interference Alignment for MapReduce: Jointly Exploiting Side-information and Reconfigurable Antennas

In order to explore how blind interference alignment (BIA) schemes may take advantage of side-information in computation tasks, we study the degrees of freedom (DoF) of a
$K$
user full-duplex wireless MapReduce setting where receivers are equipped with reconfigurable antennas, and no channel state information at the transmitters (CSIT) is assumed. The presence of side-information changes the nature of DoF optimal alignments, requiring both inter-message and intra-message alignments. The inter-message alignments are made transparent by a relabeling of messages which recasts the MapReduce problem into an alternative but equivalent problem, called Unicast-with-Side-Information (USI) in this work. New constructions of BIA schemes are presented that jointly exploit both reconfigurable antennas and side-information.