Design And Analysis For Cross Layer Design In Cognitive Radio Network

With the sharp increase of the wide application of the communication technologies, the current unlicensed spectrum is too scarce to satisfy with this trend. However, according to the statistics, most of the licensed bands exists the problem of underutilization. Cognitive radio (CR) is considered as a promising advanced technologies to solve the problem of both spectrum scarcity and inefficiency. The main idea of this technology is the unlicensed users(or called secondary user (SU)) can opportunistically access to the licensed channel under the certain interference limit to the licensed users(or called primary user (PU)). This concept, in simple terms, defines a communication device scan the spectrums on the purpose of searching spectrum holes, and accorded with the channel side information (CSI) of the primary users, transmit the data on this vacant bands carefully to avoid interfering these primary users.The work of this thesis is as follows:first cognitive cooperative sensing platform was based on 802.22 draft. After that, a cross-layer framework between transport layer and lower sensing layer is proposed in the cognitive radio network. Furthermore, a new algorithm deployed in the sensing module and a theoretical performance bound in the flow control at the transport layer are discussed in details:in the sensing module, multi-antenna sensing scheme is designed; in the flow control, performance bound with retort to the network calculus is evaluated and conducted.As a key technique of Cognitive Radio (CR), cooperative sensing has become a hotspot worthy for research. In this thesis, we introduce combining schemes of cooperative sensing, and analyze the detailed characteristics of each one. In accord with 802.22 draft, a cooperative sensing simulation platform is proposed for fair comparison of different schemes. Also, the simulation results are well analyzed and proved.Cognitive radio networks have an innovative infrastructure with dynamic spectrum access, which are quite different from the conventional networks. Thus, more adaptive network designs are required stringently. In this thesis, we contrive the cross-layer network structure between the transport layer and MAC&PHY layer in the uplink underlay CR networks. The spectrum and power allocations are aimed to achieve the maximum throughput rationally, concerned not only with the constraint of interference temperature of primary user, but with the service request conditions. Specifically, at the MAC&PHY layer, scheduling schemes try to keep balance between the QoS and the service access capacity; at the transport layer, taken into consideration of real resource supply, flow control schemes are modeled with Markov chains to maintain the stability and efficiency for the service. Simulation results reveal the significant improvement on flow control compared with conventional isolated design.Next, this thesis proposes a novel structure of multi-antenna spectrum sensing to solve the problem of the SNR Wall under harsh channel condition like deep fading. Our objective is to exploit the spatial diversity largely with resort to singular value decomposition (SVD) of the auto-correlation of the parallel signal flows. To evaluate its performance, the ROC of the proposed scheme is deduced in close form. Furthermore, normalized constellation distance is utilized as a criterion for measuring signal spacc under the same noise variance in comparing our proposed scheme with conventional energy detection. Simulation results reveal the significant performance enhancement especially in low SNR. This method can be applied widely, because it can be in the acylic primary user signal.Finally, In this paper, we use stochastic network calculus to analyze the performance bounds for a cognitive radio network with two classes of input traffic. First, stochastic service curves for primary users and secondary users are obtained based on the system model and stochastic network calculus. Then, we derive the general expressions of backlog and delay bounds for both primary and secondary users under two methods, i.e., min-plus convolution and independent case analysis. Finally, numerical results as well as simulation results are compared and discussed.