Signal separation and classification algorithm for cognitive radio networks

In the context of spectrum sharing, many approaches were developed and many algorithms were proposed in order to model and regulate the use of spectral resources. Despite the proposed solutions and spectrum access policies, there is still a big issue in cognitive radio networks with users who may intend (or not) to violate these communication rules and force their radios to access the spectrum bands when some other users are already communicating. These users become hostile terminals in the network and the fusion center has to eliminate their interfering signals. In this context we1 propose a mixed signals separation and classification algorithm that helps eliminating hostile devices. The first step consists in locating the frequency band over which the hostile terminal is communicating and then, by some mixed signals separation technique, isolate and then eliminate its interfering signal by analyzing the obtained signals from the mixture. For the simulations, we introduced some metric for the probability of detecting and classifying the hostile terminal as such.