Sensor networks comprised of multiple nodes with sensing, processing and communication capabilities are ubiquitous in tracking systems. However, when a tracking system is required to track a large number of targets, the computation and communication loads arise. One possible solution is to use a distributed scheme. In this paper we propose a distributed multiple target tracking algorithm, which takes into account restrictions on the sensor network functioning. Our method is based on the modification of Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm with a perturbation on the input. For the proposed algorithm we get an upper bound of residual between estimates and show the simulation results.