A self-diagnosis design for wireless sensor node is a big challenge for designers. Particularly, when sensor nodes are deployed in harsh environment, it's very difficult for human to intervene in case of hardware failure of node components. In this paper, we present our novel self-diagnosis for the discrete event systems (DES) like sensor node, which includes a complete strategy of self-diagnosis based on both functional and nonfunctional tests. Our approach helps sensor node to detect automatically its component failure, and then to take a corrective solution. And then, the implementation of our approach in the real material, which is based on the results of power measurement of node component, is presented. Finally, we also indicate how to optimize our self-diagnosis to make it more energy-efficient.