Distributed Task Allocation for Visual Sensor Networks: A Market-Based Approach

This work in progress presents a novel distributed task allocation method for visual sensor networks based on a computational market. Our proposed method automatically adapts the QoS levels of the individual tasks, depending on the resource requirements and the user-defined interest level of the service. Therefore, we define virtual commodity markets, where decentralized producer agents sell resource shares of nodes and communication links. Producer agents adapt their unit prices for resources depending on the demand for the individual resources. In this paper we discuss two pricing models: adjusted linear pricing and rate adaptive pricing. To allocate resources required for executing a task, a task allocation agent requests a number of offers from different producer agents. Task allocation agents use the received interest levels, which correspond to the virtual money, to lease resource shares for a specific time.