Cooperative Density Estimation in Random Wireless Ad Hoc Networks

Density estimation is crucial for wireless ad hoc networks for adequate capacity planning. Protocols have to adapt their operation to the density since the throughput in an ad hoc network approaches asymptotically to zero as the density increases. A wireless node can estimate the global density by using local information such as the received power from neighbors. In this paper, we propose a cross layer protocol to compute the density estimate. The accuracy of the estimate can be enhanced and its variance can be reduced through cooperation among the nodes. Nodes share the received power measurements with each other. Based on the collected observations, the maximum likelihood estimate is computed. It is shown that cooperative density estimation has better accuracy with less variance than the individual estimation. When nodes share received power measurements from further away neighbors, the variance of the estimate is further reduced.