Performance of two simple recursive multiple frequency estimators in additive, multiplicative and colored noise

In this paper, two algorithms that can be used efficiently to obtain very accurate multiple frequency estimates are discussed. It is demonstrated that the two methods, based on the Discrete Fourier Transform (DFT) interpolation and the use of signal correlations, are most advantages for computationally simple hardware implementations. The two algorithms use a single recursion for accurate multiple frequency estimation. The paper discusses the performance of the algorithms in additive white Gaussian noise as well as the performance using a more complete signal model having multiplicative and colored noise. The bias due to the presence of multiple sinusoids on frequency estimation has also been investigated in detail.