多點協調無線通訊網路中之擁塞控制與路由演算法

Abstract

多點協調(Coordination Multi-Point, CoMP)技術在異質網路中扮演著重要的角色，該技術使得各基地台及行動裝置(mobile device, MD)間得以相互協調與合作。行動裝置的普及與資料流量的增加，導致網路服務提供商面臨資源嚴重不足的巨大挑戰，為了提供更好的服務，網路服務提供商勢必要更有效益的利用有限的網路資源；本論文將利用流量控制機制來解決資源嚴重不足的問題。另一方面，每種應用程式的資源需求都不相同，有鑒於此，本論文提出之路由演算法，將基於使用者需求來配置網路資源。網路服務提供商針對此異質無線網路進行集中式管理，透過後端控制中心實現基於應用程式特性之擁塞控制與路由管理，並使得利潤最大化。 本論文建構一個最佳化的數學模型，並使用拉格蘭日鬆弛法(Lagrangean Relaxation, LR)進行求解程序，利用 LR 將問題拆解成多個獨立的子問題，再以次梯度法(Subgradient)反覆迭代求得近似最佳解。

Coordinated Multi-Point (CoMP) technology plays an important role in the heterogeneous wireless networks. It provides coordination and cooperation mutually between the base stations and mobile devices (MDs). The operators try to achieve high efficient and effective resource management to face the increasing number of MDs and the amount of data requirements under the limited resource constraints. A flow control mechanism is designed to solve resource insufficient problem. Furthermore, various types of application have diverse requirements of wireless network resources. Hence, this thesis tries to design a flow control with multi-session routing algorithm to handle resource allocation. This thesis considers a centralized management scheme in the heterogeneous wireless networks with the 4G base station, the fixed public AP and a back-end control center. The operator supports the flow control and routing selection, such as routing assignment through 4G network and Wi-Fi. The objective function is constructed as a mathematical model to obtain high revenue minus the penalty costs. Lagrangean Relaxation (LR) based method is proposed to optimize the objective value. The primal problem will be decomposed into several independent sub-problems by LR, Subgradient method, and the routing algorithm are proposed will be used jointly and repeatedly to get the approximate optimal solution quickly. The solution can be applied for operators in next generation wireless communication networks.