Distributed utility-based rate adaptation protocols for prioritized, quasi-elastic flows

This paper considers rate adaptation for streaming multimedia data in a wireless sensor network (WSN) consisting of multiple missions, where each mission subscribes to data streams from multiple sensors, and each sensor's data is utilized by multiple heterogenous missions. We specifically consider the application of the distributed network utility maximization (NUM) framework to a previously unconsidered scenario where the different missions have different priorities, as well as minimum utility demands. When all the utility demands are feasible, we first show that the addition of a penalty for failing to reach the minimum utility demand to the base NUM protocol leads to maximization of the global utility. The paper, however, principally focuses on those situations where the minimum demands cannot all be satisfied due to resource constraints. To address such practical scenarios, we present and evaluate a) a family of modified NUM-based protocols to determine the optimal satisfied set, when the missions have unique priority order, and b) heuristics for applying NUM, when multiple missions have the same priority.