Channel Estimation For Cognitive Radio In Nongaussian Noise

In recent years,with the rapid development of communication technology,information age has arrived.As many types of communications services coming today,the limited spectrum resources can not meet the demand any more.Cognitive Radio as the main candidate for the next generation,is capable to maximize the efficiency of utilizing spectrum resources to meet people's requirements for communication services.Channel estimation is one of the basic cognitive radio technology.Its results will directly determine the performance of cognitice radio system.Conventional channel estimation techniques generally assumed Gaussian background noise which made the channel was modeled as a linear equation.Although the system model becomes easy to analyze,but taking intuo account the actual work environment for cognitive radio systems,which is often subject to highly impulsive noise such as atmospheric thunderstorms,channel estimation based on Gaussian noise model is difficult to play the original performance,so that cognitive radio system can not work properly.Therefore,it is necessary to find a new channel estimation method for cognitive radio system.In this paper,considering the actural operating environment for cognitive radio systems,we analyze the non-gaussian noise signal,and use alpha stable distribution to establish the noise model.At the same time,according to the dynamic state space method,we make a model for cognitive radio transmission channel.After studying the existing channel estimation methods,a novel algorithm is presented to estimate and track the channel state by using the particle filter framework.Then,after demonstrated the advantages and disadvantages of the particle filter framework in detail,we improve the method in importance sampling and resampling phase:first,use Kalman filter as a proposal distribution generator,it can be highly approximate to the real posterior probability density function.Then in resampling phase,the genetic algorithm optimized by cloud model is used to guide the mutation operator for umproving the diversity of the particle set.Simulation results shows that the new algorithm can accurately estimate the channel state in the non-gaussian noise,it is more adaptive than the traditional method.