Multi-path signal decomposition with white-noise and correlation analysis

Abstract

Decomposition of composite multi-path signals is a complex problem. The windowing approach in the time domain cannot be used for overlapping signals. While the filtering approach in the frequency domain cannot separate the overlapping signal spectrum. One of the key solutions to this problem is to estimate the time-delay for each signal component. This study discusses techniques for separating multi-path signal components through time-delay estimation by analysing residual signal and its correlation with original signal. The residual signal is the error between the reference and the received signal. Overlapping multi-path signal components are detected in two different approaches. First, when the residual signal is random, the whiteness analysis is applied to detect the signal component. Second, when the whiteness test failed, which means the residual signal has a correlation with the reference signal, the correlation test can then be applied. The simulation results show that this proposed method successfully detected the signal components.