In wireless sensor networks（ WSNs）

soft fault nodes may generate and transmit false data

which will not only causes wrong decision but also will consumes more energy

because of this

present an energy-efficiency fault diagnosis（ EFD） algorithm to save energy. The algorithm makes use of spatial similarity of node sense datas to determine state of detected node by comparison of sensing data sensed by neighboring nodes. Considering transient fault

the algorithm uses total fault-free（ TF） model idea to avoid unnecessary data comparison

and reduce detection times of time redundancy. Simulation results show that the EFD algorithm can improve precision of network diagnosis and reduce energy consumption in diagnosis process.