Skip to Main Content
In this paper, we propose an adaptive data-harvesting approach for mobile-agent-assisted data collection in wireless sensor networks (WSNs) inspired by Behavioral Ecology. By using the marginal value theorem, we divide the entire sensor field into small patches and gather the correlated data from each patch. Each observation X gathered by a given sensor node to be considered to be a marginal information source with a relative standard deviation σ(x|Y, I), where Y is a set of previously collected observations by the mobile agent, and I is the background knowledge learned from the sensor field. The mobile agent estimates the correlation based on the available knowledge gathered from the current patch and the previous patches and then chooses the next visiting sensor node. The next node should have the maximum information gain obtained until σ(x|Y, I) is smaller than a predefined threshold (TH). Since, in a dynamically changing environment, the correlation varies among different patches, an efficient way to understand the correlation model is the key to efficient data harvesting. The proposed estimation technique of the marginal value theorem, which is called estimation technique based on the marginal value theorem (EMVT), is used to maintain the fidelity of the interested data with relatively fewer collected sensor observations.