Multiple signal detection under a specified false alarm constraint

This paper presents a sequential detection scheme for sinusoidal signals observed in spatially colored noise which controls the probability of overestimating the number of signals. The scheme is based on the Sequentially Rejective Bonferroni Test together with the nonlinear weighted least squares approach for estimation of the signal parameters under the hypotheses and alternatives respectively. The power of the method is compared via computer simulations to the commonly used minimum description length method, MDL. The proposed scheme performs well in comparison to the MDL method and gives at the same time control of the false alarm probability.