| With the rapid growth of wireless communications and the existing of heterogeneous networks, the available spectrum is becoming overcrowded. At the meantime, the end-to-end performance of heterogeneous networks is degraded. In order to improve the spectrum efficiency and end-to-end performance, the cognitive wireless networks (CWNs) is proposed. Cognition technology is a fundamental to achieve CWNs, among which the spectrum sensing is regarded as the essential task. Therefore, spectrum sensing has became a hotspot for the wireless researchers.Recently, a numerous researches have focused on spectrum sensing. There been many significant developments on this technology. Some typical spectrum sensing algorithms are proposed. Even though these algorithms could detect the spectrum hole, they still face great challenges. In order to promote the development of CWNs and its application, we should put great efforts and in-depth studies to address these issues. In this thesis, the spectrum sensing technologies of CWNs are studied in detail. First, the background of CWNs is introduced, including the fundamental theory, the function and the key technologies of CWNs. Subsequently, the existing spectrum sensing technologies are introduced. In order to overcome the afore-mentioned problems, we propose a cyclostationary detection with a high-accuracy and low-computational complexity on sensing performance. The sensing performance of the proposed detection is then fully analyzed in different situations. The main contributions of our works are summarized below:1) High-accuracy and Low-computational Complexity Cyclostationary Detection:The existing cyclostationary detection has a high accuracy on sensing performance, but it requires extensive computation to provide sufficiently low error probability, causing high complexity. To solve this problem, we propose an improved cyclostationary detection aiming at reducing the computational complexity by simplifying the test statistic. The closed-form expressions of detection probability and false alarm probability are then derived. Numerical results demonstrate the high-accuracy and low-computational complexity of the proposed detection. Firstly, the analysis of computational complexity shows that the proposed detection is more efficient than existing detector via simplifying of the test statistic. Secondly, the analytical results match well with the results from the simulation, confirming the accuracy of the analysis. Even though there is0.36dB sensing performance degradation compare with the exisiting detection, it is still maintains a satisfactory accuracy. The proposed detection enables to reduce the computational complexity while maintains an enough accuracy of detection sensitivity, which is of great importance for the design of optimal spectrum sensing in CWNs.2) Sensing Performance of Proposed Cyclostationary Detection over Multi-path Fading and Shadowing Channels:Since the channel between a primary user and a secondary user is multi-path fading and shadowing, if we want to examine the exact performance of the proposed detection, the effect of multi-path fading and shadowing should be considered. Therefore, the sensing performance is investigated over different multi-path fading and shadowing channels, such as Nakagami, Rayleigh, Rician fading channels and composite Rayleigh-Lognormal shadowing fading, composite Nakagami-Lognormal shadowing fading channels. The corresponding closed-form expressions of detection probability are then derived by using probability density function (PDF) approach and moment generation function (MGF) approach. Compare with these approaches, the PDF approach is more complicate but suitable for different situations, while the MGF approach is easier but it is limited by the values of fading parameter. Thus, the choice between PDF or MGF methods depends on the limitations imposed in derivations and the complexity. Even though both of the methods have their advantages and disadvantages, they still provide powerful methods to evaluate the detection probabilities over fading channels. Moreover, these approaches are also suitable for analyzing the performance of the other spectrum sensing algorithms, for example, the energy detection.3) Multiple Antennas Spectrum Sensing based on Proposed Cyclostationary Detection:The researches of multiple-antenna based cyclostationary detection are limited. Furthermore, the existing works mainly focus on multiple independent antennas, none of work investigate the multiple correlated antennas. Since the correlation between multiple antennas decreases the spatial diversity gain, the sensing performance is then degraded. To fully investigate the achievable performance of multiple antennas, this paper analyzes the sensing performance with multiple independent and correlated antennas. For the case of multiple independent antennas, considering the complete knowledge of channel state information (CSI) is difficult to obtain in CWNs, we choose a simpler technique square-law combining (SLC) to combine the received signal of multiple antennas. Subsequently, the sensing performance is investigated in different fading channels. For the case of multiple correlated antennas, the equally correlated, exponentially correlated and a linear array arbitrarily correlated antennas are treated. Based on SLC diversity, the corresponding closed-form expression of average detection probability over Nakagami fading channel is derived by MGF approach. The simulation and analytical results show that the multiple antennas technology mitigates the impact of correlation fading effect and introduces a significant improvement in the detection performance. These results help quantify the performance gains for cyclostationary detection with multiple antennas, which can help emerging applications in practical CWNs.As the last part of the thesis, we summarize our work and analyze the prospect of the developing tendency aiming to further improve the system performance of spectrum sensing. |
PDF Full Text Request