Spectral estimation of nonuniformly sampled data using an interpolation technique

A standard technique for estimating the spectrum of a nonuniformly sampled signal is to first estimate a set of uniformly sampled correlation coefficients and then calculate the power spectrum of the signal as the discrete Fourier transform (DFT) of the correlation sequence. In the popular correlation slotting technique, the average of all the samples in a chosen slot is used to compute the correlation value. In the proposed technique, the value of the signal at a desired location is obtained using a linear interpolation technique. Bias in the spectral estimate due to interpolation is determined using theoretical analysis, and variance is estimated using simulation studies. The new method requires fewer computations than the slotting technique to yield a specified accuracy