Research On Network Architeture And Key Technologies Of Cognitive Radio Networks

With the rapidly development of wireless networks, the users’ requirements are fully satisfied. Meanwhile, it also introduces many challenges. On one hand, in order to meet the user’s increasing service requirements, various wireless networks with different operation modes have emerged. Hence, the interconnection between heterogeneous networks becomes very important. On the other hand, spectrum resources are very precious and the traditional spectrum management methods are static. Furthermore, according to the research reports of authoritative organizations, e.g. FCC, the spectrum resources shortage problem is not just because the physical limit of spectrum resources, but the poor spectrum management schemes.Under this background, the CR (Cognitive Radio) technique is born. With the characteristics of flexibility, autonomy and adaptation, CR can solve the heterogeneous networks interconnection problem and improve the network spectrum efficiency, so study on CR has become one of the popular topics in wireless communications. CRN (Cognitive Radio Networks), which based on CR, have received a lot of attention recently. In order to solve the challenges of CRN, we focus on the research of novel network architecture, spectrum sharing schemes and energy-efficient transmission schemes. We also obtain some theory and technique innovation achievements. By studying the contents of this thesis, we can provide solution to heterogeneity interconnection and improve the spectrum and energy efficiency. In all, the key contributions of this thesis include three parts:Firstly, we proposed a novel network architecture for CRN, which enjoys systematization, full functions, heterogeneous interconnection and intellectuality. From two aspects of heterogeneous interconnection and intellectuality, the proposed network architecture improves the network capability by introducing some new functional module, e.g. cognitive information management module, smart management module, network reconfiguration module and so on. The cognitive information management module can provide complete functions of cognitive information acquisition, representation, transmission and storage, so the network’s cognitive capability is improved. The smart management module can provide learning and reasoning functions to other modules, which will improve network’s autonomic capability. The network reconfiguration module can change the operation parameters according to the network states and improve the network’s adaptive capability. Then it can provide seamless access and handover among different network modes to users, which means heterogeneous network interconnection. In addition, the functions of each module and the interactions between modules in the proposed network architecture are described in detail.Secondly, spectrum sharing scheme in cognitive OFDM networks, which includes channel allocation and power allocation, is investigated by using graph theory. The key point of using graph theory to solve spectrum sharing problem is graph construction. Different from the traditional scheme, the secondary links is modeled as the vertex of graph and the edge represent whether the two links is interfere with each other. Furthermore, we adopt weight to describe the interference strength. To avoid the accurate computation of interference, we adopt the relative geographical positions between different links to describe the interference level. Then the weighted interference graph is constructed. Based on the constructed graph, we proposed a low-complexity two-stage heuristic algorithm and analyzed the optimality and complexity of the proposed algorithm. The analysis result indicates that the proposed algorithm can obtain approximate performance with the optimal scheme, but the computational complexity decreased toO(N2/2+N12+M), where N represents the number of secondary link, M represents the number of channel. In addition, we further extend the scheme by considering primary users’location when setting the weight of weighted interference graph. From the simulation results, the proposed scheme can improve the network spectrum efficiency considerably and the weight selection has an enormous effect on the network performance. The primary users’ locations are considered when setting weight can further improve the network performance.Thirdly, the energy-efficient transmission under hybrid spectrum sharing method in CRN is investigated. Here we adopt the bits/joule to describe energy-efficiency, which represents the transmitted bits per joule consumed. Then the optimization model of energy-efficient transmission under hybrid spectrum sharing is given. Since the original optimization problem is too complex, we analyze the existence and uniqueness of optimal scheme from two aspects independently. One for power allocation, the other is sensing duration selection. Then a low-complexity iterative algorithm is proposed to approximate the optimal scheme. The complexity of proposed algorithm is O(Ns*Nt), where Ns and Nt are the number of iterations for find the optimal sensing duration and transmission power respectively. In addition, we also analyze the effect of detection threshold selection on energy-efficient transmission, the theoretical analysis results indicate that the complexity of optimization problem can be reduced by proper selecting the detection threshold. Simulation results verify that energy-efficient transmission can improve the energy efficiency greatly, and the performance gains of hybrid spectrum sharing and low-complexity iterative algorithm are given.