Time - Frequency Analysis in Subbands Using Adaptive Multichannel Sequential Lattice Prediction Filtering Method

Abstract

The objective of this presentation is to propose a time-frequency analysis technique suitable for cognitive radio spectrum sensing, and radar, sonar imaging applications. The input data is windowed and converted to two dimensional (2-D) format, and modeled as time varying autoregressive moving average (ARMA). It is then decimated and filtered in coloumn wise fashion so as to apply the adaptive multichannel sequential lattice prediction filtering method for ARMA spectrum estimation in subbands after which conversion to process parameters is accomplished using a Levinson-Durbin type conversion algorithm. The spectrum estimation in each subband is accordingly consummated for each coloumn of input data as part of a time-frequency representation. The performances of the proposed technique in fullband and two subbands cases are visually presented, and compared against the performance of time-varying Periodogram for two cisoids embedded in white noise with time-varying frequencies case.