Performance comparison of blind source separation algorithms using spatial time-frequency distributions
Christian Euler, L.A. Cirillo, Abdelhak M. Zoubir · 2007
Three blind source separation techniques using time-frequency distributions are examined in terms of achievable performance and computational complexity. The schemes considered involve joint diagonalization and off-diagonalization of a set of spatial time-frequency distribution matrices. These matrices must be constructed at particular time-frequency points for separation to be successful. Robustness to correct choice of the time-frequency points is also examined in this work.