Fault-tolerant EM algorithm for GMM in sensor networks

This paper presents a novel distributed scheme named fault-tolerant expectation maximization (FEM) algorithm for estimating the parameters of gaussian mixture model (GMM) in sensor network scenario where fault-tolerance and communication conservation are considered. The sensor network data are assumed to arise from a GMM. But several dimensions of some samples may not be available. Performing estimations in a series of subspaces, FEM is able to handle the incomplete data from sensor networks which have some malfunction sensors. Furthermore, FEM efficiently approximates the underlying distribution of sensor data by transferring limited parameters in each communication, which leads to the economization of power consumption. Both the experiments on synthetic dataset and real dataset demonstrate the performance of FEM in sensor network applications.