High Resolution Harmonic Retrieval Using Jdef (joint Detection/estimation Filter)

This paper addresses the problem of high resolution parameter estimation of sinusoids corrupted with additive noise. The approach involves maximum a posteriori (MAP) detection and nonlinear estimation using the Extended Kalman Filter. The problem is set up as a multiple hypothesis testing problem with uniform a priori probabilities. Each hypothesis is associated with a particular set of initial conditions that pertain to a given range of values of the signal parameters. The joint detection/estimation filter (JDEF) is then applied to detect the presence of the sinusoid and estimate its parameters. Experimental evaluation of this technique demonstrates accurate performance' in Gaussian noise.