Event-driven probabilistic topology management in sparse wireless sensor network

In a sparse wireless sensor network (WSN), event parameters such as event location and affected area cannot be directly estimated from the sensor observations, as the average distortion of sensor observations, because of the sparsity of network, is greater than in the dense ones. Moreover, observations of all the nodes which detect an event do not have the same significance in the estimation of event parameters. The authors propose a sensor collaboration scheme, named probabilistic event monitoring in sparse network (PEMS), which assists, in a distributed manner, the sensor nodes with significant observations to participate in the event monitoring process. The proposed collaboration scheme, PEMS, is based on the principle of Monte Carlo methods. Further, the event monitoring nodes restructure the network topology with the help of facility location theory for efficient energy management. Extensive simulations show that the event parameters with nominal distortion can be estimated from the observations transmitted by the PEMS-assisted event-monitoring nodes and those nodes can enhance their energy efficiency at the time of event monitoring.