A self-organization algorithm in wireless sensor networks

Facing the requirement of effective information acquisition and low energy cost from the wireless sensor networks, we construct an object function by combining information gain index and energy cost index for the self-organization optimization. The information index is constructed from the mutual information, so that information increment can be represented as an alternative sensor node to the sensor cluster. The energy cost is derived from the node communication hop number from the sensor node to the sensor cluster leader. By optimizing the performance index online, a dynamic collaborative self-organization (DCS) algorithm is proposed, through which the clustering leader can be automatically elected and its member nodes can be selected sequentially. Compared with IDSQ in our simulation, DCS can relieve the problem of 'too frequent leader switches' and achieve almost the same tracking accuracy at the cost of low energy consumption.