A critical observation collection method for Sensor Networks inspired by behavioral ecology

In this paper, inspirations from behavioral ecology are applied for mobile agent assisted data collection in a Wireless Sensor Networks (WSNs). With the help of the marginal value theorem based strategy (MVTS), each observation (Λ), which is gathered by a given sensor node, is considered as a marginal information source with a relative entropy H(Λ). The mobile agent exploits the correlation and chooses the next sensor node to be visited, which is deemed that the information contribution of the contained observation is not smaller than a predefined threshold (TH). Therefore, understanding the correlation models can benefit the WSNs system from two aspects. On one hand, the resource consumption could be reduced during the acquisition process of the observation; on the other hand, the accuracy of the reconstructed field data is least compromised, due to relatively critical observations being collected by the mobile agent over a dynamically changing environmental. The resource consumption such as energy and bandwidth, is proportional to the number of visited sensors. With MVTS, the resource consumption is optimized through bypassing the sensors with relatively unimportant observations. Illustrated analytical and simulation results confirm the above achievements.