Research On The Methods Of Spectrum Sensing In Cognitive Radio

With the continuous extension of communication service,the demand on wireless communication of people has been increasing continuously.Under current fixed spectrum allocation strategy,the available spectrum resource has been scarcer and scarcer.The wireless spectrum as a scare resource can not satisfied the increasing demand of wireless spectrum band for people.Cognitive radio adopts dynamic spectrum allocation strategy instead of fixed spectrum allocation strategy,and allows secondary users(SU)to exploit the licensed spectrum without harmful interference with primary users(PU).It fundamentally solves the problem of spectrum insufficient producing by unreasonable spectrum allocation strategy,and is the inevitable developing trend of future communication.Spectrum sensing is the key and premise of cognitive radio.To solve some problems in current methods of spectrum sensing,this paper researches spectrum sensing for single node and single channel,parallel cooperative spectrum sensing for multi-node and multi-channel,and secure spectrum sensing,respectively.For single node sensing,this paper analyzes the fractal dimension of signal and noise in both time domain and frequency domain,and researches the blind spectrum sensing method which can sense the spectrum fast,accurately and effectively in poor situation.For multi-channel and multi-node spectrum sensing,the parallel cooperative spectrum sensing method maximizing the system profit and cross-layer parallel cooperative spectrum sensing method for heterogeneous multi-channel are researched.They optimize the system proformence by optimally selecting the sensing nodes,assigning the sensing task and determining the sensing time.For the secure problem in spectrum sensing,this paper researches secure and effective spectrum sensing methods to against the signal noise ratio(SNR)attack and primary user emulation(PUE)attack.This paper mainly includes:Firstly,the fractal dimensions of noise and signals with different modulation types are both analyzed in time domain and frequency domain.According to different characters of the fractal dimension between signals and noise,several fast blind spectrum sensing methods based on fractal dimension are propose,and their performance are detailedly analyzed.Among them,the spectrum sensing methods base on fractal dimension in frequency domain,including spectrum sensing method based on PFDF,KFDF and SFDF,respectively,can overcome the limitation of box dimension based method which can not detect the signals with some modulation types and some modulation parameters,and have some advantages such as less influenced by modulation parameters,insensitive to noise uncertain,high accuracy in low SNR situations,and less calculation time.So it can be used to fast blind spectrum sensing in situations of low SNR,noise uncertain and without any priori knowledge of PU.Secondly,current cooperative spectrum sensing methods has some problems,such as only considering sensing accuracy instead of the sensing overhead and the system profit,neglecting the different sensing performances between different SUs and channels,and requiring more cooperative SUs.To solve those problems,three parallel cooperative spectrum sensing methods are propose,including greedy based parallel cooperative spectrum sensing method,parallel cooperative spectrum sensing method based on genetic algorithm(GA),and parallel cooperative spectrum sensing method based on iterative Kuhn-Munkres(KM)algorithm.They obtain the sensing task assignments which indicate when and which SU should sense which channel by greedy algorithm,GA and iterative KM algorithm,respectively.And their performances,complexities and application scenes are detailed analyzed.These three parallel cooperative spectrum sensing methods fully consider the sensing overhead,system profit,and the difference of sensing performance between different SUs and channels.They can overcome the shortcomings of current spectrum sensing methods which only consider the sensing accuracy and require more cooperative SUs,and optimal determine the sensing strategy to improve the system profit and reduce the number of cooperative SUs effectively.Thirdly,for parallel cooperative spectrum sensing in heterogeneous multi-channel situation,this paper fully considers the difference of sensing performance between different SUs and heterogeneous channels,the valid system throughput and the sensing duration.And based on energy detection,a model of jointing sensing task assignment and sensing duration selection is built.After formulating it as a multi-objective nonlinear combination optimization problem,two parallel cooperative spectrum sensing methods based on heuristic algorithms,which are parallel cooperative spectrum sensing for multiple heterogeneous channels based on heuristic algorithm with sensing time exhaustion(HATE)and parallel cooperative spectrum sensing for multiple heterogeneous channels based on heuristic algorithm with sensing time heuristic(HATH),are proposed.They can maximize the valid system throughput by optimal assigning sensing task and optimal selecting sensing duration under the constraint of detection probability and false alarm probability.Finally,to some potential secure problems in spectrum sensing,two secure spectrum sensing methods which can against malicious attack are proposed.This paper first proposes a spectrum sensing method based on SNR weighted.In order to defend against the potential SNR attack of SNR weighted method,a spectrum sensing method based on the geographical location information against SNR attack is proposed.It calculates the suspicious degree of each node according to its geographic location information and reported SNR value to identify the malicious nodes and eliminate the impact caused by them.Finally,to defense against the PUE attack,two spectrum sensing methods against PUE attack based on fractal dimension are proposed,which are spectrum sensing method against PUE attack based on NSFDF and spectrum sensing method against PUE attack based on double dimension fractal dimension.They identify the modulation type of signal by SVM classifier according to the fractal dimension of received signal to identify the malicious attack.