The invention discloses a method for tracking an unknown moving node in real time in a sensor network, is applied to the field of sensor networks, and aims at solving the problem that high-precision real-time tracking of sensor nodes is completed under the constraint of limited energy. The method comprises: establishing a sparse matrix model based on extended Kalman filtering, using a Fast-ADMM-with Restart algorithm to solve the model , the number of tracking nodes is reduced as much as possible within the allowable range of the tracking precision, and therefore the average service life of the nodes in the whole sensor network is prolonged. In addition, the method is based on loop iteration of a server segment algorithm, and the calculation consumption of the sensor node can be effectively reduced.