Optimal Sensor allocation and scheduling

Mesut Yavuz, Leader

UAVs can be used to collect data from some sites of interest in a more economical way than alternative technologies. For example, we might fly a UAV equipped with right sensors over a production facility and measure heat, pollution, humidity, etc. characteristics. Collected data ensure that the area is under control. A similar application arises in using the satellites to watch areas of interest. In most practical situations, a sensor's time can be allocated to several sites, creating a scheduling problem. The sensor scheduling problem aims to maximize the information about the sites and/or minimize the cost of operation. The problem can be solved by either a probabilistic or a deterministic approach. Probabilistic approaches yield schedules that are unpredictable to the opponent/enemy. The deterministic approaches, on the other hand, assure that each site will be visited in a finite period of time, and, hence, the information about a site will never be lost. We have been studying both probabilistic and deterministic approaches and have obtained significant results. The deterministic version is also related to real-time periodic scheduling problems widely studied in computer science literature, and our solution methods make significant contributions to "pinwheel scheduling" in particular.