研究軟體定義網路在機器對機器通訊網路和無線網路的效益

Abstract

本論文中，我們研究軟體定義網路在機器對機器通訊網路與無線網路的效能強化。在機器對機器通訊網路中，我們著重在無數的裝置產生的巨量資料導致資料中心的問題。在無線網路中，我們著重在由於無線裝置成長快速所導致的能源利用的問題。 在機器對機器通訊網路中，為了解決巨量資料導致的伺服器高負載、同時滿足服務品質，負載平衡常常被提出。最近幾年，軟體定義網路在強化負載平衡的技術中有很大的潛力，因此我們提出以軟體定義網路下的流量相關的負載平衡機制，所提出的機制結合了軟體定義網路的網路監控的能力，藉由流量辨識以及動態的網路繞送，滿足不同服務的服務品質要求。實驗結果顯示出，在現實的軟體定義網路中，所提出的機制可以顯著的減少服務延遲，同時減緩軟體定義網路中控制層的高負載。 在無線網路中，網路存取點被很快速的布建在各個角落來滿足大量的網路要求，特別是在公司以及校園，然而許多文獻中顯示，網路存取點中有很大的比例是閒置狀態，因此浪費了大量的能源，為了減少浪費的能源，同時不影響服務品質，我們提出以軟體定義網路的服務品質相關的無線存取點省電機制，所提出的機制結合了軟體定義網路的動態監控網路的能力並且估算各個裝置的流量要求，透過控制網路繞送來提供低延遲的換手。實驗結果顯示出所提出的機制可以有效的減少網路存取點的能源消耗，同時滿足服務品質要求。 在本論文中，我們以軟體定義網路來強化機器對機器通訊網路的負載平衡以及無線網路的省電機制，我們發現相較於現有的負載平衡系統，軟體定義網路的負載平衡系統能有較好的效能，特別是在機器對機器通訊網路環境下。同時軟體定義網路確實有潛力來提高無線網路的能源效率。由於軟體定義網路擁有高彈性的特性，本論文中提出的兩個系統皆有潛力能再被優化。

In this thesis, we exploit the benefits of Software-Defined Network (SDN) specially for Machine-to-Machine (M2M) and wireless networks. For M2M networks, we focus on the load balancing issues of the M2M cloud center due to the traffic from enormous M2M devices. For wireless networks, we focus on the energy issue of wireless networks while the number of mobile devices grows rapidly.

For M2M, to relieve the heavy loads caused by burst M2M traffic subject to the various quality of service (QoS) requirements, load balancing are often applied to M2M networks. In recent years, SDN has be shown to improve the performance of load balancing techniques. We further propose a traffic-aware SDN-based load balancing mechanism for M2M networks. The proposed mechanism can satisfy different the QoS requirements of M2M traffic to dynamically monitor and control the entire network, and thus reroute different traffic type dynamically. Based on the experimental results of our SDN testbed, we demonstrate that the proposed scheme can significantly reduce the service response time and provide quality of service while relieving the heavy workload of SDN controller.

For the wireless local area networks (WLAN), access points (APs) are being rapidly deployed to satisfy QoS. However, many empirical studies showed that many idle APs result in the significant energy losses. To overcome this issue, we propose a SDN-assisted QoS-aware AP energy saving mechanism. We leverage the capability of the SDN controller to dynamically monitor the network conditions to estimate the network bandwidth requirement of the devices and then manage the network forwarding for seamless handover. Our experimental results show that the proposed scheme can effectively reduce the energy consumption of WLAN APs, while still satisfying the QoS requirements.

In this thesis, we consider two scenarios : 1) SDN-based load balancer for M2M networks, and 2) SDN-based WLAN for energy efficiency improvement. We find that SDN-enabled load balancer can improve the performance of M2M networks. Furthermore SDN can improve the energy efficiency of WLAN with different QoS requirement. Because of the high flexibility provided by SDN, both M2M and WLAN we demonstrate that can be enhanced significantly with different QoS requirement.