Performance analysis of the SAGE algorithm: Changing the update procedure
Musayanah, Puji Handayani, Gamantyo Hendrantoro, Prasetiyono Hari Mukti · 2016
The extraction of multipath component signals from measured radiowave propagation channel can be carried out using some algorithms. One of these algorithms is Space Alternating Generalized Expectation-Maximization (SAGE). SAGE is based on two procedures, i.e., Expectation Step (E-Step) and Maximization Step (M-Step). In the E-Step, we decompose signal and noise while in the M-Step we update the parameters estimate, which are delay, DOA (direction of arrival), DOD (direction of departure), doppler shift, and complex amplitude consecutively. However, if the measurement systems used to sounding the channel have low delay resolution, we will have difficulty in separating multipath component signals (MPCs). In this paper we propose to change the update procedure to overcome this difficulty. The simulation results show that the performance of the proposed procedure does not change significantly compare to the original SAGE.