Energy Efficient Geographic Anycast in Wireless Sensor Networks

One of the basic operations in a wireless sensor network is the enquiry and transmission of sensed data from sensor devices in some specified regions to the data center for further processing. A key challenge in data enquiring and transmitting is to minimize the total energy cost occurring at all sensor nodes involved. In this paper, we introduce and study the energy efficient data enquiring problem under the geographic anycast model: given the location of a data sink and some sensor nodes in a set of regions, construct a tree rooted at the sink such that at least one sensor in each given region is in the tree. The objective is to minimize the total energy cost of the transmitting nodes in the tree. We first prove that this problem is NP-hard and unlikely has an approximation algorithm with a performance ratio in logarithmic of the number of sensors in the network. We then propose some tree based approximation algorithms. We make both theoretical and simulation analysis for the performances of proposed algorithms. Our study shows that anycast approach is more energy efficient than broadcast or multicast