Owe Under The Conditions Of The Formulation And Single-channel Mixed-signal Given Order

In non-cooperative communication environment, since only the observation signals and limited prior information could be obtained form the sensor, it became one of the most difficult problems in communication reconnaissance to process communication signals and radar signals rapidly and accurately under this sort of circumstance, which, though, would be handled by blind source processing algorithms, a recently thriving technology. As for blind signals processing, the sources number ought to be estimated through certain algorithms based on given information, known as mixture order determination, to assure more premises or advantages could be utilized for succeeding steps like source separation and parameter analysis. Currently, most of the cases in source number estimation research is in determined or overdetermined condition, instead of underdetermined condition, where number of sensor or mixing channel is less than source number, which is actually the common case in actual application, thus order determination of blind mixing signals in underdetermined and single channel condition is one of the most valuable and practical research projects.In this dissertation, it is concentrated on order determination of blind mixing communication and radar signals in underdetermined condition. The key work and innovations mainly include:1. Based on the introduction of related research status home and abroad, several classic and new decision criteria are analyzed and compared with each other.2. Signal processing utilizing sparsity is researched and developed. Order determination algorithm based on signal sparsity is proposed for source number estimation of signals blindly mixed in three channels.3. Algorithm based on probability spectral density function ratio is deduced and adopted to deal with signals in communication reconnaissance, with improvement proposed according to signals characteristics for better adaptability.4. Several kinds of channel extension methods are researched, including methods based on source generation model, methods utilizing sensor spacial virtual tap and methods based on high-order cumulant, and developed for channel extension of observation signal in single or multiple channel. Channel extension method being incorporated with certain order decision criteria, order determination technology is proposed to estimate the source number in underdetermined situation.5. A series of simulations and experiments are conducted to verify the algorithms validity and analyze its estimation performance.