Research On Spectrum Detection Technology In Cognitive Radio System

As the service of wireless communication is increasing persistently, the wireless communication system requires more and more wireless spectrum resource. Thus, the suitable wireless spectrum resource of the wireless communication service appears rarity. In fact, the suitable wireless communication spectrum resource is scarce in one aspect and in the other aspect is the wireless spectrum resource is wasted very much. That is because the static spectrum programming system and the dynamic spectrum programming system are not matched. So the wireless spectrum programming is inefficient and the use efficiency of the spectrum is low. How to enhance the use efficiency of the whole spectrum from the system-level becomes the main path solving the problems presented above. In this case, one of the future wireless communication system's strategic developing directions is to break through the key technologies of sharable spectrum wireless communication system, exploit spectrum resource shared wireless communication system.In the cognitive radio system, the spectrum could be shared and use efficiency could be enhanced. As the key technologies of the shareable spectrum, effective algorithm of spectrum detection supplies shared spectrum and high spectrum use efficiency. The algorithm of spectrum detection in the future cognitive radio system can not only enhance the ability of the spectrum holes detection, but also reduce or eliminate the interference to the primary users. And this is the basic condition of the dynamic spectrum access. The spectrum detection in the cognitive radio system is presented as a masterstroke in this paper. Several algorithms of spectrum detection are researched in the paper, such as cyclic spectral characteristics detection, energy detection, cooperation and data fusion detection.(1). Cyclic spectral characteristics detection scheme:Many of the communication signals in use today may be modeled as cyclostationary signals due to the presence of one or more underlying periodicities. Therefore, signals have one or more cyclostationary characteristics in nature. And the different modulation signals have different cyclostationarity. However, Gaussian-white-noise is not exhibit the cyclostationary property. According to the cyclostationary characteristics, useful signal could be detected and abstracted from the spectrum pooling system.The 2nd order cyclic spectral characteristics of primary users and cognitive users are gained to detect ideal spectrum. According to the cyclostationary characteristics of signal which is received by cognitive users, whether the primary users exist or not could be forecasted. In this way, the distributive state information of the channel could be estimated.Another cyclostationarity spectrum detection scheme is put forward. Through this way, the performance of spectrum detection is improved, the robust of against noise average power fluctuation is enhanced, and the robustly of detecting weak signal is polished up. And the requirement of the spectrum detection technology under a lower SNR (signal noise power ratio) environment is satisfied. The obvious shortcoming of this algorithm is the complexity, so the common algorithm used nowadays is the energy detection.(2). Energy detection scheme:the algorithm of energy detection is researched in most literatures, because the algorithm is simple. In this algorithm, the characteristic of the signal is not needed. A power threshold is set to judge the signal. But the algorithm is sensitive to the noise. Especially when the algorithm is used in a low SNR cognitive radio system, the problem becomes even more obviously.The relationship among the SNR, the noise average power fluctuation in short time, and the detection time is researched firstly. And a conclusion is achieved that a tiny change of the short time noise average power arouses a sharp decline of the detective performance. As a result, a novel energy detection algorithm with dynamic threshold is bought forward. This algorithm boosts up the robust ability in rivalry to the noise average power fluctuation in short time, and the detection sensitivity of the energy detection is increased.At the same time, another improved spectrum detection algorithm is presented. In this new algorithm, the signal sent by primary users could be classified into two parts. They are pilot signal and the signal whose structure is unknown. Both matched filter and energy detection are used in this algorithm. The known pilot signal is detected by the matched filter, and then the result would be detected by the energy detection. This algorithm is also suitable to increase the performance of the energy detection and the detection sensitivity.(3). Cooperative and data fusion detection:Because the interaction information between the cognitive users and the primary users is absent. And the spectrum detection of the cognitive users is often carried out in a low SNR condition. In most of the detection condition, there are often sight-line which contains multi-path and large-scale decline and hidden station. In order to get more exact detection information, multi-users cooperative detection and data fusion scheme could be adopted.Firstly, a multi cognitive user cooperative spectrum detection algorithm based on maximum ratio combining is represented. On the low SNR occasion, even the short time noise average power fluctuation is evident, when the number of the cooperative detection users is increased, the decrease of the detection performance, which is brought by the noise average power fluctuation could be eliminated. In this way, more accurate detection performance is gained. The theory is analyzed, and the simulation is put forward in the paper.In addition, the data fusion spectrum detection algorithm is also researched. The expression of log-likelihood ratio function about received data in fusion center and the number of secondary users confirmed that primary user being was studied firstly. The secondary users'number was referred ahead including tow parts, one is correct decision and the other is false-alarm decision. Then according to the statistical character between them, the detection performance is gained in the fusion center. At the same time, the simulation result proves the theory above. And the spectrum detection performance of the new algorithm is compared to the traditional "Logic-And" and "Logic-Or", the performance of proposing scheme improved evidently.