Coding for parallel links to maximize expected decodable-message value

Future communication scenarios for NASA spacecraft may involve multiple communication links and relay nodes, so that there is essentially a network in which there may be multiple paths from a sender to a destination. The availability of individual links may be uncertain. In this paper, scenarios are considered in which the goal is to maximize a payoff that assigns weight based on the worth of data and the probability of successful transmission. Ideally, the choice of what information to send over the various links will provide protection of high value data when many links are unavailable, yet result in communication of significant additional data when most links are available. Here the focus is on the simple network of multiple parallel links, where the links have known capacities and outage probabilities. Given a set of simple inter-link codes, linear programming can be used to find the optimal timesharing strategy among these codes. Some observations are made about the problem of determining all potentially useful codes, and techniques to assist in such determination are presented.