Sorting methods for estimating the number of emitters and their parameters

The current literature contains many algorithms for determination of the number of emitters and their parameters when the emitter signals are all present over the observation period of the receiver. In the problem considered, emissions are of short duration, and interference between emissions at the receiver is a rare event. The receiver estimates the angle, frequency, amplitude, and time of arrival for each emission assuming that no other emissions are present. These estimates are then collected for some time interval and passed to a sorting method that estimates the number of emitters and the parameters associated with each. Two methods are presented. These methods are ad-hoc, although the 3D method resembles the Parzen with a normal kernel for estimating a probability density function and the maximum a posteriori method for estimating the parameters. The 2D method is a modification to the 3D that trades off performance for speed of execution. Both methods were evaluated using over 100 data sets. The data varies from sparse, containing 50 or less hits over the collection interval, to dense, with over 1000 hits. In sparse environments both methods produce about the same emitter reports. In dense environments, the 2D method sometimes misses emitters that the 3D method detects