The SAGE algorithm with a three-dimensional antenna array
Jo Verhaevert, Emmanuel H. Van Lil, Antoine Van de Capelle · Ghent University Academic Bibliography (Ghent University) · 2002
This study investigates the behaviour of the Space Alternating Generalised Expectation maximisation (SAGE) algorithm in three dimensions and in a multi-user environment. The SAGE optimisation algorithm is used to replace the high dimensional optimisation procedure necessary to compute the joint maximum likelihood estimate of the parameters by several separate maximisation processes which can be performed sequentially. The calculated parameters in this paper are the relative delay, the azimuth angle, the elevation angle and the complex amplitude of the waves. Convergence and performance analysis are based on numerical simulations. We calculate the resolution of this algorithm, the convergence speed, and the Bit Error Rate (BER) as function of the Signal to Noise ratio (SNR). Based on simulations with a uniform spherical distributed antenna array, we can conclude that the SAGE algorithm is a powerful tool that can be succesfully applied for parameter extraction with high resolution.