Research Of Spectrum Sensing Algorithm And Primary Users' Spectrum Occupancy Behavior Modeling

In recent years, with cognitive radio as the representative of the dynamic spectrum access technology provides an effective solution to address the current problem of low utilization of spectrum resources, pointing out a convenient and effective way to address the limited spectrum resources in contradiction to the growing demand of spectrum. In cognitive radio networks, the spectrum consumers are divided into two categories:one is called the primary user which has precedence to access to the band, and the other one is secondary user with less accessing priority. The basic principle of cognitive radio is that the secondary users dynamically access the spectrum without interfering with the conventional communication of primary users, thereby enhancing the spectrum efficiency, a prerequisite to employ this technology is the accuracy of spectrum sensing both in time and space domain by secondary users.This dissertation mainly focus on temporal spectrum sensing technology in cognitive network. Firstly we reviews the signal detection technologies which is the theoretical foundation of spectrum sensing, and then proposes a cooperative blind spectrum sensing technology based on double threshold energy detector, while detailed analysising the approximation errors introduced by Gaussian distribution instead of chi-square distribution based on Cental Limit Theorem. Secondly the primary users' spectrum occupancy behavior model is established based on the actual operating data collected form the mobile communication networks, using the theory of time series analysis. With the help of this model, we obtain the prediction of the primary users' behavior of spectrum occupancy in the further. According to the former research results, we design an intermittent energy-saving spectrum sensing strategy as our last contribution. Finally, we conclude the dissertation and prospect the future work.The contributions of this dissertation include:(1) We improve the existing cooperative spectrum sensing technology to introduce a double threshold energy detection method. According to this method, there are two thresholds in the energy detector of the secondary users, and they transmit their local decisions or energy values according to the region the received energy values located. This method takes advantage of both the two kinds of information, the final decision is made by a fusion center through information fusion, to decide whether the primary users are presence or not. The results show that when the received energy values fall into the delay decision region with high probabilities, this method can significantly improve the cooperative spectrum sensing capability.(2) We detailed analysis approximation errors introduced by using the Gaussian distribution instead of chi-square distribution to study the performances of the energy detector based on Central Limit theorem. We investigate the overall and the specific error characteristics through two error functions:normalized mean square error function (NEF) and absolute error function (AEF), respectively. Numerical simulations show that the reasonable lower bound of the degree of freedom under both hypothesizes have the same value of three, or four in stricter constraints. Furthermore, we list the exact relationship between the minimum absolute errors and the corresponding chi-square distribution probabilities in several tables to facilitate looking up the most suitable zones which researchers might be interested in.(3) In order to analyze the spectrum occupancy of primary users, and improving the performance of spectrum sensing, this dissertation discusses the occupancy duration of carrier level in the existing GSM network in detail. We establishes the mathematical model of primary users'spectrum occupancy behaviors based on duration data of carrier level collected from actual network, and discussed two kinds of methods:the direct and indirect method, to modeling this operating indicator. The simulation results show that the mathematical model could provide accurate prediction of the primary users'spectrum occupancy in one hour.(4) We proposes an intermittent energy-saving spectrum sensing strategy based on the previous research results in this dissertation:double threshold energy detection and primary users'spectrum occupancy behavior model. In this strategy, secondary users set their silent periods according to the prediction of the primary users' occupancy duration, to reduce the energy consumption. The simulations of actual data show that the proposed strategy could save about 6% energy consumption of secondary users, with little impact on spectrum efficiency.In this dissertation, we study a blind spectrum sensing technology without any priori information of primary users, as well as an intermittent spectrum sensing strategy while knowing the primary users' spectrum occupancy model. We focus on enhancing sensing capability of secondary users, reducing energy consumption, thereby utilizing the limited spectrum resources more effectively, and improving the spectrum efficiency. Our work is benefit to the design of more effective spectrum sensing methods and strategies, and provides feasible research approaches to modeling the primary users' behavior.