ACE: An Emergent Algorithm for Highly Uniform Cluster Formation

The efficient subdivision of a sensor network into uniform, mostly non-overlapping clusters of physically close nodes is an important building block in the design of efficient upper layer network functions such as routing, broadcast, data aggregation, and query processing. We present ACE, an algorithm that results in highly uniform cluster formation that can achieve a packing efficiency close to hexagonal close-packing. By using the self-organizing properties of three rounds of feedback between nodes, the algorithm induces the emergent formation of clusters that are an efficient cover of the network, with significantly less overlap than the clusters formed by existing algorithms. The algorithm is scale-independent — it completes in time proportional to the deployment density of the nodes regardless of the overall number of nodes in the network. ACE requires no knowledge of geographic location and requires only a small constant amount of communications overhead.