On signal decomposition performance of Null Space Pursuit

Null Space Pursuit (NSP) is a new operator-based algorithm of adaptive signal decomposition without need for a priori bases, and signal is modeled as a superposition of local narrow band signals. Little attention has been paid to NSP decomposition performance at this stage. Here with a simplest model of two-tone signal, NSP decomposition performance was investigated. To measure the quality of decomposition, the cross correlation coefficient was adapted, and it was found proper initial values of parameters had great influence on NSP decomposition performance. The results showed that NSP archived best performance at certain λ
1
⁰
value, and varied while the sampling frequency or frequency of subcomponent changed.