QoS optimization in ad hoc wireless networks through adaptive control of marginal utility

Applications consisting of messaging, voice, and video are used to provide situational awareness to decision makers and emergency responders in high criticality crisis scenarios such as disaster management. Here, ad hoc wireless networks are often quickly provisioned to provide the necessary connectivity to support these applications. Applications ill prepared to deal with the constant fluctuation of available bandwidth will stall or fail and contribute to mission failure. Our algorithm, D-Q-RAM (Distributed Quality of Service (QoS) Resource Allocation Model) allows applications to satisfy their specific QoS expectations in dynamically fluctuating networked environments by incorporating a distributed optimization heuristic that results in near optimal adaptation without the need to know, estimate, or predict available bandwidth at any moment in time. This paper describes our approach for managing that optimization heuristic in a manner that is decentralized, that is network routers are unaware of the semantics of the applications, and the applications can arbitrate among competing signals from numerous network routers and select an appropriate QoS level which results in an improved overall global utility of available network bandwidth.