Research On Key Techniques Of Cognitive Radio Systems

With the rapid development of wireless communication technologies and the increasingbroadband communication demands, the shortage of spectrum is becoming an outstandingproblem. On one hand, the conflicion between spectrum supply and demand is increasing. On theother hand, according to the survey of related organizations the utilization of licensed spectrumsis very low under the present static spectrum management mode. Under the above background,the cognitve radio technology is proposed, which aims at enhancing the utilization of licensedspectrum with intelligent using the unused spectrum of licensed systems.The key techniques of cognitive radio can be summarized in three aspects: SpectrumSensing, Spectrum Sharing and Spectrum Management. Spectrum sensing aquires theinformation of existence and spectrum holes in the surrounding wireless environment, which canbe used for adjusting communication parameters to achieve the maximum spectrum efficiencywithout causing any harmful interference to primary user.Spectrum sharing is to scheduling thefrequency band utilization to make several wireless communication systems be working in thesame spectrum band. The goal of cognitive radio is to provide the best available spectrum tocognitive user, with which the QoS requirements can be meet. Spectrum management plays therole of deciding the spectrum with the best performance.Effectiveness and easy implementation always are the continuous pursuit of communicationengineering, in this dissertation, we take the cognitive radion system as the study backgroundand the engineering practical applicability as the starting point, on the basis of analysing andmining a number of problems in key technologies of cognitive radio, a serious research of energydetection, peak-to-average power ratio and out-of-band power reduction in NC-OFDM systems,and cognitive engine design is performed. Precisely, the work can be divided into the followingfour parts:1. The performance of energy detection in double-Rayleigh fading environment is studiedand analized for the first time. The probability density function of signal-to-noise ofdouble-Rayleigh is derived. By using the alternative representations of MarcumQ function, theclosed form solution of average detection probability for energy detection in double-Rayleighfading environment is derived. The performance increasement of selection combining and switchcombining are also studied.Simulations and numerical results give the provement of ourconclusiong, which provide a valuable reference and guidance to the appication of energydetection in double-Rayleigh environment.2、The statistical properties of PAPR in NC-OFDM systems is analyzed, and the importance of PAPR reduction is pointed out. Then a PAPR reduction method based on subcarrier weightingis proposed, which reduces the PAPR by weighting the subcarriers data with optimumcoefficiences. The optimum weighting coefficiences is derived by using genetic algorithm, and adeep analysis is conducted in PAPR performance after weighting.The simulation results verifythe effectiveness.3. In order to suppress the out-of-band power in NC-OFDM system, an improved subcarrierweighting algorithm is proposed. It deduces the OBP by weighting both the real and image partof the subcarrier data. Theory analysis and simulation results show that at least a3dBsuppression gain can be obtained with the improved algorithm and no further SER performancedegration can be taken. There is a severe conflict between PAPR and OBP reduction, to solvethis problem, due to the fact that subcarrier weighting can reduce both of them and its simpleness,a joint reduction method is proposed based on subcarrier weighting. Simulation results verifythat the joint reduction method can effectively mitigate the conflict between PAPR and OBPreduction, and compared to the separate reduction method, an obvious better suppressionperformance can be obtained. Also, according to the real wireless environment, it can povide theflexibility to satisfy the communication requirements by adjusting the trade-off parameter.4. Taking high effectiveness and low implementation complexity, an improved frameworkof cognitive engine is designed. It can schedule optimization engine and learning&reasoningengine more efficiently, reduces the implementation difficulty of cognitive radio, and enhancesthe practicality. A multi-class SVM decision tree based on genetic algorithm called GABDTSVMis poposed, which generates the optimum decision tree structure by calculating the optimumgrouping method at each node with GA. Then it is applied to the learning engine design, and thefeasibility and effectiveness are verified combined with a specific wireless scenario. Also, thesuperiority and advantage of GABDTSVM to other multi-class method is proved with thesimulaiton results.