Research On Key Technologies For Spectrum Sensing In Cognitive Radio

With the fleeting progress of wireless communication technology, spectrum scarcity has become the bottleneck of improving the existing communication technologies and developing novel communication systems. On the other hand, the licensed spectrum is rarely utilized due to the conventional inefficient fixed frequency allocations. As a result, cognitive radio (CR) arises to be a tempting solution to the spectrum scarcity. In CR system, the secondary user (SU) exploit the unused part of the spectrum by spectrum sensing, and according to the dynamic radio environment, the SU adjusts its operating parameters through the underlying Soft Defined Radio (SDR) to opportunistically access the idle spectrum. Being the focus of this paper, spectrum sensing is one of the most critical functions for the CR to enable Dynamic Spectrum Access (DSA). The main achievements and results of this dissertation are listed at follows.1. We analyze the robustness and induce the“SNR wall”of the energy detector over multipath fading channels in the CR system. Analysis and numerical results show that the multipath fading results in a higher“SNR wall”of energy detector. When the received signal-to-noise ratio (SNR) on SU is lower than the“SNR wall”, the robust detection can not be achieved no matter how much time we prolong for the detection. Therefore, it is more likely that the energy detector will fail to achieve the target sensing performance over multipath fading in the CR system.2. In order to overcome the shortcomings of the energy detector and to improve the sensing efficiency, a cooperative spectrum sensing algorithm based on sequential detection is proposed firstly. Moreover, the average sensing time of the new algorithm is optimized and the achievable minimum sensing time is obtained as well as the accordingly parameter settings. Secondly, we propose an efficient sequential probability ratio test (SPRT) based cooperative spectrum sensing method for large scale CR networks. A subset of the SUs with the largest received SNRs is selected for cooperative spectrum sensing. It is shown by theoretical analysis and simulation that the average number of SUs required for cooperative spectrum sensing is substantially decreased. Thus the local sensing information forwarded to the fusion center (FC), as well as the energy consumption, can be significantly reduced compared with the existing works. Finally, in CR networks the SU's system may have a strict time delay requirement and the spectrum sensing must be completed in the assigned sensing slot. However, the detection time required to reach a decision in sequential detection is a random variable. Therefore, in order to improve the average speed of spectrum sensing as well as to avoid some too-long-sensing-time tests in CR networks, a truncated sequential detection algorithm is proposed. Simulation results show that the proposed algorithm, under the constraint of limited sensing time, can satisfy the performance requirement while reducing the average sensing time.3. Besides the design of spectrum sensing algorithm in CR networks, allocation of the sensing duration is another problem of critical importance for efficient spectrum sensing. Given a fixed total sensing time, we focus on optimal allocation of the sensing duration among multiple primary channels in chapter 4. Under the constraint that the primary users are sufficiently protected, the problem is formulated as a convex optimization, and the convexity of the problem is verified by proving that the Hessian matrix of the objective function is negative definite. Then, we propose a searching algorithm combining Penalty method and Newton's method to obtain the optimal solution.4. It is inevitably that malicious users exist in cognitive radio networks. To resist attacks against the spectrum sensing algorithms. We figure out the performance loss of conventional cooperative spectrum sensing methods under malicious attacks based on the established threat model, and propose a weighted cooperative spectrum sensing algorithm. In our method, each SU is allocated a reputation value firstly, and the reputation values are updated according to the behavior history of CRs. Then, the weighted factors are derived from reputation values, and the final result is made by weighted fusion at last. Simulation results show that our method outperforms the conventional cooperative spectrum sensing algorithms under threat scenario.