Distributed Optimization and Duality in QoS Control for Wireless Best-Effort Traffic

This work addresses the problem of node-by-node QoS control in wireless ad-hoc networks carrying best-effort (elastic) traffic. We characterize a duality behavior of the original and so-called adjoint network in the sense of optimization theory. Relying on the adjoint network concept we propose a decentralized algorithm yielding the optimal power allocation and analyze it in the framework of stochastic approximation