Research On User Distribution And Resource Allocation In Heterogeneous Wireless Networks

In the past few years, with the rise and development of mobile Internet, mobile traffic data amount has grown rapidly. According to related statistics, mobile traffic data volume has been growing more than100%per year since2008. On the one hand, with the development of smart handheld devices, more subscribers have been admitted to wireless networks. On the other hand, with the development of multimedia services, mobile users have an increasing demand for higher data transmission rate. Therefore, upgrading the capacity of the global wireless networks, to meet the increasing bandwidth requirements of users, has become an urgent need of developing wireless communication. However, due to the bandwidth scarcity, the conventional cellular network with large coverage cannot satisfy the ever increasing data rate demand. One feasible solution is to increase the cell density, by implementing micro access points with small coverage in the hotspots to offload, which increases the bandwidth reusability and enhances the global network capacity accordingly. Another potential solution is to implement wireless network access points with different radio technics in the hotspots, such as wireless LAN access points, to offload from the cellular networks. Both solutions mean that integrating heterogeneous wireless networks to enhance service rate has become a promising trend.Different types of wireless networks have different network characteristics, such as coverage, network capacity and service support capability, etc. Different user equipments (UEs) may have different characteristics, such as traffic type, multi-mode access ability, channel state and moving speed, etc. Former works have shown that combining the characteristics of heterogeneous wireless networks and UEs to implement resource management can improve the global network utilization and service quality. Therefore, UEs distribution and resource allocation based on characteristics combination in these heterogeneous wireless networks is a problem worth studying. Focus on this core problem, this dissertation gives some solutions from three aspects:1) For the distribution problem of UEs with different received signal qualities in heterogeneous wireless networks, this dissertation studies physical link rate-differentiated restricted admission control scheme and dual thresholds based load scheduling scheme. Consider a scenario of cellular network and WLAN (wireless local network) interworking, where different UEs may have different channel conditions. So these UEs may have different physical link rates in different networks when taking rate adaption technic into consideration. Thus, on the one hand, to distribute these new arrived UEs, we configure different admission regions for UEs with different physical transmission rates. On the other hand, when load imbalance occurs, we use the dual thresholds to trigger the scheduling, and configure different scheduling priorities for UEs with different physical link rates. The admission control problem and load scheduling problem are formulated to maximize the global network throughput with service quality guarantee constraints, respectively. We consider the user mobility characteristics, traffic arrival and network service process, and employ the queue theory and moment function analysis method to obtain the optimal admission regions and thresholds. Results show that the proposed two schemes can significantly improve the global network throughput when the traffic load is light in the area outside hotspots.2) For the admission problem of UEs with different moving speeds in the heterogeneous cellular networks sharing same spectra, we study the moving speed-differentiated admission control, with combination of bandwidth reservation and bandwidth allocation. Consider a scenario of heterogeneous cellular networks with frequency sharing, where different types of base stations have different transmitting power and cell-size levels. So high-mobility UEs have small residence time in the micro cells. Distributing all the high-mobility UEs in the micro cells to these micro base stations, would bring a great amount of handovers, which leads to much communication overhead and service quality degrading. But distributing all the high-mobility terminals in the micro cells to these macro base stations means large bandwidth reservation, leads to the decrement of bandwidth reusability. Therefore, firstly we consider joint admission control and bandwidth reservation, configure different admission probabilities and reserved bandwidths for UEs with different moving speeds, and formulate it as a problem of maximizing the compromise of network capacity and handover rate. Secondly, we further consider joint admission control and bandwidth allocation, configure different admission probabilities and bandwidth for UEs with different moving speeds and traffic types, and formulate it as a problem of minimizing the sum weighted handover rates. By employing stochastic geometry modeling, the handover rate expressions of different types of UEs are derived, and through adopting convex optimization analytical method, the method for optimal solutions of admission probabilities, reserved bandwidths and allocated bandwidths are obtained. The numerical results show that the proposed schemes can effectively reduce handover rate, and the optimal way to handle the high-mobility UE admission is:distributing some high-mobility UEs to macro base stations for larger residence time, and distributing the rest high-mobility UEs to micro base stations by providing larger bandwidth through bandwidth reservation or bandwidth allocation.3) To achieve optimal distribution of UEs or their traffic loads in heterogeneous wireless networks from the time dimension, this dissertation studies UEs or their traffic loads distribution based on the future network load states. Indeed distributing UEs only based on the current network status or utility cannot achieve the optimal allocation in the time dimension. From the perspective of time dimension, to distribute UEs or their traffic loads, we need to consider the current network status, the future network status and the effect of the current decision on future network state and future decisions. So, firstly, we propose a load scheduling scheme based on the future network load state prediction, which aims to minimize the sum of network capacity weighted network idle durations, and give a two-stage iteration method to predict the idle duration between two schedules. Through convex optimization analysis, we give a gradient descent method to obtain the optimal scheduling loads in the overlapped regions. Secondly, we propose a scheme of joint vertical handover and resource allocation based on the long term global network utility. We give a Markov decision process based method to solve the long term global network utility optimization problem. And by employing state decomposition, we give a distributed decision-making solution that the network decides the resource allocation and the handover UE decides the network selection, which can greatly deduce the state space dimension and calculation complexity. Results show that the performance of the proposed schemes with consideration of future network status is significantly better than the performance of schemes only based on the instantaneous network state or utility.