Polynomial order reducing property of lattice filters in detection of linear FM signals

Order-reduction techniques have already been successfully applied to the detection of linear FM signals in white noise. In this paper, reflection coefficients of adaptive lattice filters are used, in a new aspect, to reduce the order of input polynomial phase FM signals. This concept is then used to detect chirp signals in white noise. To define an appropriate hypothesis problem for the transformed signals in the first adaptive reflection coefficient, its statistical behaviour is experimentally investigated. It is shown that for small confidence intervals, the relevant probability density function for the transformed signal can be assumed Gaussian.