In order to meet the rapidly growing demand for wireless services, future communications systems need to be more flexible and robust. Software Defined Networking (SDN) is a recent networking architecture with promising properties relative to the management of increasingly complex network structures, thus providing diversified services. Specifically, resource allocation is one of the key technologies to optimize network resources and ensure Quality of Service (QoS). This dissertation focuses on resource allocation algorithms in Software Defined Wireless Networks (SDWN), including a high precision localization algorithm based on semi-definite programming in SDWN, a wireless resource allocation algorithm for energy consumption minimization in Software Defined Wireless Sensor Networks (SDWSN), an energy-efficient resource allocation algorithm in Software Defined Cellular Networks (SDCN), and a price-based resource allocation algorithm in Ad Hoc Mobile Cloud (AHMC).
The main contributions of this dissertation can be summarized as follows:
1. To obtain the precise location estimation of mobile targets, a centralized localization algorithm is proposed in SDWN. Firstly, the mobile targets obtain the ranging information and inertial information, and then upload the information to the controller based on OpenFlow forwarding scheme. Then, both the ranging measurement errors and step size are modeled using Gaussian mixtures, respectively, and the locations of the mobile nodes are determined by the maximum likelihood estimator. Additionally, by using Jenson’s inequality and semidefinite relaxation, the initial NP-hard problem is transformed into a convex optimization problem, which globally optimal solution could be attained using semidefinite programming. Simulation and experimental results show that the proposed algorithm can achieve higher positioning accuracy than the traditional algorithms. Especially, when the system noise and observation noise in the state space model are no more following Gaussian distributions, the propose algorithm could improve the localization performance with the approximation of the noise distribution, thus, laying the foundation for upper-level services in SDWN.
2. To solve the energy limitation of sensor nodes and extend the lifetime of SDWSN, a wireless resource allocation algorithm is proposed to minimize the energy consumption. Considering the minimum signal to interference plus noise ratio required for the transmission, an optimization problem is formulated to minimize total energy consumed by the sensor nodes in SDWSN. Then, the original non-convex problem is transformed into a convex optimization problem via convex relaxation, and a centralized algorithm is proposed to realize adaptive bandwidth and power allocation. Besides, to analyze the performance of the proposed algorithm, two special cases are elaborated for adaptive bandwidth allocation and adaptive power allocation, respectively. Moreover, in order to fully acquire and utilize the global information of the network, a centralized resource allocation scheme is designed based on the OpenFlow communication protocol. In contrast, a distributed resource allocation scheme is also given to serve as a performance benchmark. The simulation results show that the proposed centralized algorithm could balance the utilization of power and bandwidth resources. Meanwhile, by taking advantage of global optimization, the proposed centralized algorithm could reduce the impact of network size and node heterogeneity on the overall network performance.
3. To improve the energy efficiency (EE) of the users accessed through Femtocell, an energy efficient resource allocation algorithm is proposed in SDCN architecture. Considering the user QoS and cross-tier interference tolerance, a mixed integer programming (MIP) problem is developed to maximize the network EE. Then, to reduce the complexity of solving the original nonlinear and nonconvex problem, a centralized resource allocation algorithm is proposed via convex relaxation. Additionally, a distributed non-cooperative game is derived using Cauchy inequality to act as the performance baseline of the proposed centralized algorithm. Furthermore, an SDN-based centralized resource allocation scheme is designed with elaborate forwarding rules. The simulation results show that the proposed centralized algorithm could be much closer to the upper bound of EE than the distributed algorithm at the cost of computation complexity, and improve the overall network throughput as well.
4. To study the efficient utilization of wireless and computing resources, a price-based resource allocation algorithm is proposed in AHMC. First, by taking into consideration of communication and computational costs, a buyer-seller game is formulated to maximize the individual utility of mobile users and base stations, which is handled by the SDN controller. Then, the Stackelberg equilibrium of the game is derived in a quasi-static scenario with non-uniform pricing and uniform pricing, respectively. Additionally, an OpenFlow-based forwarding rules placement is designed to make the game reach the equilibrium more efficiently. Furthermore, the proposed algorithm is extended to the dynamic scenario with guaranteed convergence. Simulation results show that through workload offloading, the buyer can make full use of the cloud computing resource in the AHMC, and in the non-uniform pricing scheme the base stations and other users could be motivated to provide more wireless and cloud computing resources.