Resource Allocation in Dynamic Environments

This technical report examines two challenges related to resource allocation that can negatively affect system operation in a dynamic environment, where warfighter needs for resources, resource availability, environmental effects, and mission conditions can change from moment to moment. The first challenge occurs when warfighters overstate their individual needs of a shared resource, leading to inefficient allocation. Overstatement may bring local optimization; however, it can cause global inefficiencies that result in a detriment to overall mission success. This challenge is addressed by using computational mechanism design, more specifically, the dynamic Vickrey-Clark-Groves allocation mechanism. The second challenge involves resource availability that may change frequently. Such is the case in a wireless mesh network where routes and bandwidth may vary over even small intervals of time. In such a case, an adaptive quality of service (AQoS) approach is used, and the available resource is allocated using the Dynamic QoS-based Resource Allocation Model (D-Q-RAM). Computational mechanism design is used to allocate sensors, and the AQoS approach allocates the available network bandwidth in a way consistent with the sensor allocation, providing an approach for dealing with resource allocation and adaptation in a dynamic environment. Initial experimental results of applying the approach are reported.