Estimation of target trajectories based on distributed channel energy measurements

Source localization using channel energy measurements of distributed sensors has been well studied, but assumes the target not to move significantly during both the communication between the sensors and the calculation. We want to estimate the trajectory of a moving target, which passes through a sensor field. The sensors are simple in the sense that they can only measure the received signal strength (RSS), but no angular information; nor do they know the target source level. The target moves significantly between measurements, and communication between sensors only takes place at long intervals relative to the target motion. To reduce communication load, we suggest each sensor first processes its measurements locally and transmit only some aggregate value(s). Based on a free-space propagation model, we find each sensor should locally estimate three parameters to fully characterize its observations, namely the maximum measured amplitude, the corresponding time index and the speed with which the target signal fades, e.g, the time difference between the maximum and its 6 dB point. Based on two sensors, this allows us to reconstruct the target trajectory with only two possible mirror ambiguities, which could either be handled by a higher layer tracking algorithm as false alarms or resolved via a third sensor.