Self-Adaptive resource allocation for event monitoring with uncertainty in Sensor Networks

Event monitoring is an important application of sensor networks. Multiple parties, with different surveillance targets, can share the same network, with limited sensing resources, to monitor their events of interest simultaneously. Such a system achieves profit by allocating sensing resources to missions to collect event related information (e.g., videos, photos, electromagnetic signals). We address the problem of dynamically assigning resources to missions so as to achieve maximum profit with uncertainty in event occurrence. We consider timevarying resource demands and profits, and multiple concurrent surveillance missions. We model each mission as a sequence of monitoring attempts, each being allocated with a certain amount of resources, on a specific set of events that occurs as a Markov process. We propose a Self-Adaptive Resource Allocation algorithm (SARA) to adaptively and efficiently allocate resources according to the results of previous observations. By means of simulations we compare SARA to previous solutions and show SARA’s potential in finding higher profit in both static and dynamic scenarios.