Research On Cooperative Spectrum Sensing Based On Cognitive Radio

In recent years, with the popularity of wireless communication network and therapid development of wireless communication system, the demand on the performanceand capacity of communication system increases faster and faster. At the same time, thesurge of wireless communication business type and quantity also makes the demand ofspectrum increase faster. This increase inevitably leads to the shortage of spectrumresources. At this point, the traditional way of fixed frequency spectrum managementwill no longer be applicable, then cognitive radio arises at the historic moment.Compared with the traditional way of spectrum management, cognitive radio systemhas the ability of the perception of the environment, the ability to adapt to the changeof environment, the ability of reconfiguration and can work autonomously. Thus, incognitive radio systems, the unlicensed users can utilize the licensed frequencies whilethe primary user is absent. Due to the huge potential of cognitive radio and theflexibility of spectrum utilization, this technology has become a promising techniqueto solve the conflicts and enable dynamic spectrum sharing.Sensing the spectrum efficiently in a reliable manner is of vital importance. It is acritical part of the cognitive radio system, because it is the first step of the wholecognitive radio technology. This is just the focus of this thesis. This thesis studied themain theory of spectrum sensing, including the basic theory, algorithm design,performance evaluation and the parameter optimization problems. The main work andcontributions of this thesis are given as follows:(1) A collaborative spectrum sensing algorithm based on double threshold energydetection has been proposed in this thesis. In the traditional single thresholdbased energy detection, the threshold setting is usually selected according tothe background noise environment, and it is often difficult to accuratelydetermine the noise power with the presence of uncertainty, so the threshold isdifficult to set accurately. In addition, even if the threshold can be determined accurately, due to the random signal, the credibility of the decision near thethreshold is not so high, so it is easy to show a false result. In this case twothreshold can be used to filter out the middle section of unreliable information.The system gives the occupied decision when the energy increase to a certainvalue, otherwise the system will give the absent decision. While a doublethreshold method can reduce misjudgement, the detection failure happens.Aiming at the insufficiency, the thesis proposed to carry out the transfer of themiddle part of the energy to solve the problem of failed test problem. At thesame time, on this basis, in order to avoid data overhead increase problems inthe process of sensing information transmission, we put forward to quantifythe middle part of the energy value, and only transmit when the test failureoccurrs. Results show that the proposed method achieves a good trade-offbetween sensing performance and communication overhead.(2) An improved energy detector based cooperative spectrum sensing method withmultiple antennas is proposed in thesis. Through the adjustment of thecalculation structure of the energy detector, we can obtain greater sensingresolution so as to improve the accuracy and the detecting ability of sensingsystem. At the same time, we use multiple antenna to gather sensinginformation. Through multi-antenna diversity gain, a stronger signal will beobtained to be processed so as to improve the accuracy of sensing system.However, due to the random signal, false alarm and miss detection willinevitably happen in the process of detection. Due to the correlation of the falsealarm probability and the missed detection probability, in previous studies,they often fix a quantity and reduce the other. In this thesis, the performanceis also analysed to find the minimum sum of the probability of missed detectionand the probability of false alarm. The optimal values of the improved energydetector and the detection threshold in each cognitive radio user are obtained.We also get the optimized voting rule to implement cooperative spectrum sensing and find the least number of secondary users needed to achieve a targetdetection performance. It is also shown that, by optimizing the parametersjointly, the detection performance can be promoted greatly.(3) A low-cost screening mechanism of malicious nodes is proposed to solve theproblem of malicious attack in collaborative spectrum sensing networks.Compared with the single node detection, although collaborative detection canimprove the detection performance of the sensing system, it is also veryvulnerable to the interference of malicious users. Malicious users maypropagate false sensing information to the fusion center to mislead the decisionmaking and severely degrade the detection performance of the sensing system.In this thesis, to solve the spectrum sensing data falsification attack problem,we first define the sensing efficiency to evaluate the sensing performance, andanalyse the sensing efficiency of the system with the presence of malicioususers. A symmetric cryptographic mechanism is investigated to mitigate theimpact of the malicious on sensing efficiency. Moreover, to cope with the extraoverhead in the filtering of false information, we propose to optimize thenumber of reported sensing bits to save the energy consumption. It is shownthat, the sensing system will achieve a good trade-off between detectionperformance and sensing overhead.(4) A detection user selection criterion is proposed for cooperative spectrumsensing system in cognitive radio networks. Considering the problem that thesingle user based sensing information may be uncertain, inaccurate orincomplete, various cooperative spectrum sensing approaches based on anumber of cognitive users have been proposed. However, power consumptionand communication overhead with the cooperation among large number ofusers tend to become very large. In this thesis, we study the effect ofcooperation for spectrum sensing in cognitive radio networks. The resultsdemonstrated that increasing the number of cognitive relay users will improve the detection capability dramatically. We also provide the optimal guidelinesto select the cooperative users to guarantee detecting performance in a practicalradio environment. It is found that when the square of the sum of the gains ateach cognitive user increases to a certain degree, more users and larger gainsmay not lead to higher detection probability.