Wireless surveillance sensor networks are often required to track multiple targets accurately, especially when the targets come close to each other at so-called mixing regions. This research presents eTrack protocol to address the following two problems for a number of targets and their mixing region: (i) which sensor nodes should be scheduled to sense a particular event such that targets are tracked accurately, and (ii) which pair of tracks passing through the mixing region belong to a particular target. To tackle the scheduling problem, eTrack introduces a first-of-its-kind closed-loop coverage control technique that allocates resources of sensor nodes for each target in proportion with its required tracking accuracy. To handle the identity management problem, eTrack exploits the fact that targets prefer certain paths through a mixing region and takes advantage of the spatio-temporal correlation of events detected by different sensor nodes within the mixing region for the same target. Simulation results that eTrack significantly reduces the computation and communication overhead with mixing regions while achieving better tracking accuracy than existing multiple target tracking techniques.