Research On Radio Resource Scheduling In Cognitive Radio Networks

As a promising wireless communications technology, cognitive radio aims to solve the under-utilization problem of the occupied radio spectrum currently. In 1999, based on software defined radio (SDR), Mitola proposed the concept of cognitive radio (CR). As a new intelligent SDR-based wireless communication technology, it enables spectrum sensing in the time, frequency and spatial domain on a wide-band spectrum, to obtain the activity of authorized users on the licensed spectrum, and adjust their own parameters according to the wireless environment. The main effect of cognitive radio technology is to find the spectrum opportunities for communication between unlicensed users, without interfering the communication between licensed users. As an advanced technology accessing to the wireless spectrum dynamically in an unlicensed manner, cognitive radio improves the spectrum efficiency effectively, enables the dynamic wireless spectrum sharing, and greatly improves the spectrum utilization of communication systems.For orthogonal frequency division multiple access (OFDMA) based systems, from the perspective of resource scheduling, this paper studies cognitive radio technology for different network architectures. It aims to optimize the wireless resource allocation of unlicensed users to maximize the sum utility of licensed and unlicensed systems, or to maximize the utility of unlicensed system under specific interference constraint to protect the licensed system. This paper introduces the technical background of cognitive radio and resource scheduling first, and then describes the research work on optimal resource scheduling schemes in centralized CR network, optimality of resource scheduling schemes and convergence of power optimization algorithm in distributed CR network, sub-optimal resource scheduling algorithm design in distributed CR network, cross-layer design considering route selection in a multi-hop CR network. Main contributions of this paper are as follows:Firstly, for resource scheduling in centralized CR network, an optimal resource scheduling scheme is proposed based on a general form of utility function. There exists a lot of study on subcarrier and power allocation scheme in OFDMA system, but in CR networks, in order to protect the licensed users, resource scheduling for unlicensed users is subject to some constraints. In this paper, the sum of interference caused by all unlicensed users at each licensed receiver must be below a predefined threshold. Since the existing study did not consider this constraint, the existing subcarrier and power allocation schemes for throughput maximization of multi-access channel (MAC) are no longer applicable. Based on the optimization theory, this paper models the problem as a convex optimization program. Using the dual decomposition method, the optimal subcarrier and power allocation scheme is derived. Based on it, the practical algorithm is proposed, which applies the Lagrangian dual method to obtain the optimal Lagrangian multipliers iteratively. Finally, the simulation result is presented with corresponding analysis.Secondly, for distributed CR network, the network architecture can be modeled as a Gaussian interference channel (GIC) model. The channel capacity of GIC and convergence of distributed power optimization algorithm are still research hot points, and they are also the theoretical basis of resource scheduling design in distributed CR network. In this paper, the innovative points on the two aspects include:For the channel capacity problem, under the model of GIC, the paper studies and compares the impact on system throughput of two different spectrum sharing approaches, i.e., the Overlay manner and the Underlay manner, namely, frequency division multiplexing (FDM) and spectrum reusing. In the objective of maximizing total system throughput, the sufficient condition is derived to choose FDM rather than Frequency Reusing for spectrum sharing between licensed and unlicensed users. Besides, for the convergence problem, under a general form of utility function, this paper studies the convergence of non-cooperative distributed power optimization scheme. For this problem, the existing research is limited to the Shannon capacity utility function. This article extends it to the general form of utility function, and derives the sufficient conditions to guarantee the algorithm convergence.Thirdly, based on the theoretical work for distributed CR network above, algorithm design is considered for practical network scenario. Specifically, the network scenario is the distributed CR network, which reuses the cellular uplink resources for data transmission, to improve the spectrum efficiency of the whole network. We propose an interference coordination scheme to achieve efficient resource sharing between the CR sub-system and cellular sub-system. In the proposed scheme, taking into account of the fast dynamic scheduling feature of most existing cellular network, CR links employ a cellular UE scheduling based transmission power control mechanism, to adapt to the time-varying interference environment caused by cellular communication. In addition, in order to reduce the channel uncertainty caused by the fast fading, we use the stochastic optimization method to track the long-term channel state. Finally, based on convex analysis to the optimization problem, a practical scheme is developed, which can be integrated into the existing cellular network protocols. At last, the convergence is proved in a mean-square-error sense, and the link and system performance of the interference coordination scheme are investigated by simulation results.Fourth, in the multi-hop distributed CR networks, combining with route selection, the cross-layer design problem is considered. In the objective of maximizing different types of Peer-to-Peer (P2P) performance, this paper aims to solve the joint optimization problem of route selection and resource scheduling. Firstly, modeling the interference relationship between the links in a way of conflict graph, it is proved that the optimization problem can be modeled as a convex optimization problem, and the expression of the optimal solution is derived. Based on this conclusion, we propose an iterative algorithm for distributed implementation, which applies Lagrange duality theory and Frank-Wolf method to achieve the optimal solution. The simulation results study the relationship between P2P bottleneck throughput, hops and interference threshold.