Quality of Information Maximization for Wireless Networks via a Fully Separable Quadratic Policy

摘要

An information collection problem in a wireless network with random events isconsidered. Wireless devices report on each event using one of multiplereporting formats. Each format has a different quality and uses different datalengths. Delivering all data in the highest quality format can overload systemresources. The goal is to make intelligent format selection and routingdecisions to maximize time-averaged information quality subject to networkstability. Lyapunov optimization theory can be used to solve such a problem byrepeatedly minimizing the linear terms of a quadratic drift-plus-penaltyexpression. To reduce delays, this paper proposes a novel extension of thistechnique that preserves the quadratic nature of the drift minimization whilemaintaining a fully separable structure. In addition, to avoid high queuingdelay, paths are restricted to at most two hops. The resulting algorithm canpush average information quality arbitrarily close to optimum, with a trade-offin queue backlog. The algorithm compares favorably to the basicdrift-plus-penalty scheme in terms of backlog and delay. Furthermore, thetechnique is generalized to solve linear programs and yields smoother resultsthan the standard drift-plus-penalty scheme.