This paper addresses the problem of trajectory tracking that deals with gathering coherent information on the past behavior of a mobile target. When a target enters and moves within a region covered by a sensor network, information about this target is generated by any active sensor that detects the target in its monitoring area. These gathered data have to be received by some registered nodes that are in charged of identifying the trajectory of the target to perform latter complex computations at the application level (e.g., trajectory forecasting, pursuer/evader; optimization of the management of natural disasters, etc.). Our frst contribution consists in formally specifying the problem: the proposed formal specification is the frst one to the best of our knowledge. We also propose an original architecture combining three distinct abstractions which allows to describe various solutions. Some algorithmic solutions that are necessary for the self-stabilizing implementation of the trajectory tracking specification are outlined. The overall solution blends together; in the context of tracking applications, three research areas: temporal correlated data, causal correlated (content related) data, and self-stabilizing overlays.