Abstract: Due to the inherent stringent bandwidth and energy constraints，a link loss rate inference method applied to wireless sensor network is proposed.According to the sequence of data loss/receipt collected in the sink，the problem is formulated as a Maximum-Likelihood Estimation problem and is solved using the Expectation-Maximization algorithm.Simulation results demonstrate that the proposed methodology can infer the loss rate of packet accurately without any burden on network，and scale to networks size.