Enhancement of angle of arrival estimation by sparse deconvolution algorithm

Ray tracing is a technique often used to model indoor and outdoor environments. In order to improve the parameters used in ray tracing, measurements from channel sounding can be employed to test algorithms and build more accurate models of the channel. However, channel sounding can be limited as the antenna radiation pattern may change dramatically over a wide frequency span. As a consequence, discrepancies may arise between measurement by channel sounding and simulation by ray-tracing model. To overcome this limitation, in this paper sparse deconvolution will be adopted as a method to test and improve data measurements from real environments. Initial measurements are performed in a controlled (anechoic) environment where the sources of radiation can be specifically defined and show the effect of reduction in bandwidth on accuracy of the signal estimation. Comparisons are made between the Inverse Discrete Fourier Transform technique for processing the measured data and that using the Cauchy-Gaussian model. Good results are obtained for two different scenarios and it is shown that the Cauchy-Gaussian model requires less than half the bandwidth of the impulse response method using only IDFT.