Research On The Key Technologies In Cognitive Radio Networks

With the emerging of new wireless technologies and applications, the communication systems require more radio spectrum resource. But the lack of spectrum resource is the severe obstacle of these communication systems. Actually, it is reported that there are vast temporal and spatial variations in the usage of allocated spectrum, and the fixed spectrum allocation policy is censured as the major reason. Cognitive radio (CR), which can opportunistically access the radio spectrum in temporal and spatial dimensions, is considered as a promising technology to achieve the dynamic spectrum access and solve the problem of lack spectrum. CR network is the networking of cognitive radio technologies, where the external wireless environment information can be obtained by the cognition function that will be then analyzed and learned in order to provide intelligent decision. The wireless environment can be dynamically adapted by reconfiguration. In this thesis, the following three hot problems are investigated:channel access, cross-layer design, and cooperative communication. Also, several novel schemes are proposed to provide better performance. Detailed theoretical analysis and simulation are provided. The main contributions of the thesis are as follows.1. The distributed channel access scheme in multi-channel sensing framework is proposed.Firstly, the sensing problem in CR networks is analyzed. Due to the limitation of hardware, one secondary user can only sense one (or several) channel at a time, not the multi-channel sensing. Therefore, a multi-user cooperative sensing scheme is proposed to solve this problem, where the channel sensing is assisted by surrounding secondary users. Secondly, under the multi-channel sensing framework, a TCP throughput based distributed channel access scheme is proposed, where the channel access is at the goal of maximizing the TCP throughput.Thirdly, the channel access process is formulated as a restless bandit system. The index rule is capable of reducing the complexity.Finally, the simulation results show that the TCP throughput is significantly improved by using this scheme.2. A joint channel sensing and access policy as well as parameter design at the wireless link layer based TCP performance improvement is proposed.Firstly, how the parameters in the wireless link, the sensing policy, and channel access to affect TCP performance is discussed. Based on the detailed theoretical analysis and simulation, it is implied that the AMC scheme at the physical layer, frame size at the data link layer, sensing, sensing operating parameter, channel access all affect the TCP performance. Hence, jointly considering the above mentioned parameters and policies, the TCP performance in CR networks can be improved.Secondly, the wireless channel is modeled as a finite state Markov channel (FSMC). Markov chain is used to model the state change of wireless channel and spectrum occupancy by primary users.Thirdly, the CR network is formulated as a partially observable Markov decision process (POMDP) where the spectru state information at the next time slot is inferred by the information state that incorporates the history of the observation and action. Consequently, an optimal policy can be derived to maximize the TCP throughput.Finally, simulation results show that the design parameters in CR networks have a significant impact on the TCP performance, and the TCP throughput can be substantially improved by using cross-layer design.3. The relay scheduling and power control scheme for supporting the cooperative communication is proposed.Firstly, the cooperative transmission in underlay paradigm based CR network is discussed, the power constraint at the source node and relay node is analyzed, and the system capacity is developed.Secondly, a reward function that jointly considers the achievable rate and the energy consumption is developed. Moreover, the relay node scheduling and power control is achieved to balance the achievable rate and network lifetime under the objective of maximizing the obtained reward.Thirdly, the relay node selection and power control in CR networks is formulated as a Restless Bandit system where the relay node is characterized by spectrum state, link states and residual energy state. All the states transition according to the Markov chain.Finally, simulation results shows that the proposed scheme can improve the total system reword, and the network lifetime can be prolonged by sacrificing some achievable rate.