Instantly decodable network codes for cooperative index coding problem over general topologies

Abstract

We consider a group of n wireless clients and a set of k messages. Each client initially holds a subset of messages and is interested in an arbitrary subset of messages. Each client cooperates with other clients to obtain the set of messages it wants by exchanging instantly decodable network coded (IDNC) packets. This problem setting is known as the cooperative index coding problem. Clients are assumed to be connected through an arbitrary topology. In the absence of any known algorithm to complete the exchange of packets for general network topologies, we propose a greedy algorithm to satisfy the demands of all the clients with the aim of reducing the mean completion time. Our algorithm, in a completely distributed fashion, decides which subset of clients should transmit at each round of transmission and which messages should be coded together by each transmitting client to generate an IDNC packet. The algorithm encourages transmissions which are decodable for a larger number of clients and attempts to avoid collisions. We evaluate the performance of our algorithm via numerical experiments.