A Generalized Model for Cost and Fairness Analysis in Coded Cooperative Data Exchange

Abstract

We consider the issues of cost and fairness in the problem of cooperative data exchange among a group of wireless clients. In this problem, each client initially holds a subset of packets and needs to obtain the full set of packets through cooperation with other clients via a shared broadcast channel. To find minimum cost transmission schemes, we propose a general model for the problem which is based on network information flow with side information available to the sinks. As a special case of minimum cost solutions, the minimum number of required transmissions is studied in detail. We show that packet splitting is a natural consequence of solving the linear programming associated with the mentioned network flow problem. Our main observation is that splitting the packets not only minimizes the number of transmissions, but also it results in considerably more fairness compared to the case where splitting is not allowed. Hence, incentive-based long-term cooperation among users can be sustained.