Load-Aware Traffic Control in Software-Defined Enterprise Wireless Local Area Networks

With the growing popularity of Bring Your Own Device (BYOD), modern enterprise Wireless Local Area Networks (WLANs) deployments always consist of multiple Access Points (APs) to meet the fast-increasing demand for wireless access. In order to avoid network congestion which leads to issues such as suboptimal Quality of Service (QoS) and degraded user Quality of Experience (QoE), intelligent network traffic control is needed. Software Defined Networking (SDN) is an emerging architecture and intensively discussed as one of the most promising technologies to simplify network management and service development. In the SDN architecture, network management is directly programmable because it is decoupled from forwarding layer. Leveraging SDN to the existing enterprise WLANs framework, network services can be flexibly implemented to support intelligent network traffic control. This thesis studies the architecture of software-defined enterprise WLANs and how to improve network traffic control from a client-side and an AP-side perspective. By extending an existing software-defined enterprise WLANs framework, two adaptive algorithms are proposed to provide client-based mobility management and load balancing. Custom protocol messages and AP load metric are introduced to enable the proposed adaptive algorithms. Moreover, a software-defined enterprise WLAN system is designed and implemented on a testbed. A load-aware automatic channel switching algorithm and a QoS-aware bandwidth control algorithm are proposed to achieve AP-based network traffic control. Experimental results from the testbed show that the designed system and algorithms significantly improve the performance of traffic control in enterprise WLANs in terms of network throughput, packet loss rate, transmission delay and jitter.