Resource Allocation Research Based On Spectrum Aggregation Constraint In Cognitive Radio Network

The rapid development of the mobile Internet, which is featured by the soaring growth of many new wireless services, brings about several key technical problems, such as how to improve the spectrum efficiency, how to interact effectively with the surrounding environment to learn and obtain information and make decisions, and how to rapidly adapt to achieve intelligent radio.Cognitive Radio System provides an effective way to solve spectrum resource shortage problem, which can optimize resource allocation processing to improve spectrum utilization efficiently through active sensing, intelligent learning, dynamic adjustment and decision. While it still faces some major technology challenges, such as spectrum sensing, dynamic spectrum access and management, power-control, adaptive transmission and so on. It is these technologies that make it differentiate from the traditional radio technology. Therefore, this thesis focuses on studying spectrum allocation problem in the cognitive radio network.In this paper, we adopt graph-coloring model which usually is used to solve spectrum allocation problem in Cognitive Radio Networking. Moreover, we introduce spectrum aggregation technology to further reduce the high complexity of centralized resource allocation policy. Using spectrum aggregation technology will result in loss of spectrum utilization, therefore we adopt rapidly water filling power allocation algorithm for aggregated subcarriers power-optimization. On this basis, we propose the C-Sum algorithm which has been proved that the joint frequency and power resource allocation based on aggregation can take into account of the computational complexity and spectrum utility by simulation.Three algorithms have been introduced in the study of distributed resource allocation. However, these three algorithms didn’t consider the difference reward between nodes. To resolve the allocation problems in this scenario, two improved algorithms, M-Greedy algorithm and M-Fair algorithm, are proposed. The two improved algorithms allocate frequency based on the tag value of each node and the tag value is defined to consider collaboration between each nodes. Simulation results show that the two improved can achieve higher spectrum efficiency compared to the original algorithm. In particular, the M-Fair algorithm can better take into account of the spectral efficiency and the fairness between nodes.Finally, this thesis takes advantage of Simulink to design the non-continuous orthogonal frequency division multiplexing link. It aims to demonstrate decision-making process of cognitive users according to the varying state of the licensed user subcarrier occupancy, and provide a reference to further studying of NC-OFDM.