Design And Implementation Of An Adaptive QoS Routing System Based On Overlay Network

With the development of the Internet, many Internet applications which have high demand on quality of service (QoS) have appeared, such as the video conference, the video on demand (VOD) and the distance learning. These real-time applications require ISPs to provide end-to-end QoS service.Because of the initial network design principles and network architecture, it is difficult to provide QoS routing services to the multimedia applications. Presently,the Internet supports routing based on minimal hops or minimal cost. It is difficult to extend the network layer to support QoS routing, multi-path routing, or other new routing mechanisms. However, it is a practical approach to implement these new routing mechanisms with overlay network technology. Under this background, we proposed a overlay network, which provides end-to-end QoS service at the overlay network layer, solving the QoS routing problem partially that the Internet faces currently.In this article, the characteristics and limitations of the traditional IP network and the overlay network is analyzed. Based on the analysis, a QoS-adaptive routing overlay network system is proposed. The system structure, module function, the communication mechanism and the communication interfaces are introduced.Based on the system architecture, we propose a shortest-delay adaptive overlay routing algorithm, which can dynamically establish multiple paths between end systems. This process does not need to establish the fixed paths in advance. It only needs overlay nodes to poll the end-to-end delay. The QoS routing mechanism guarantees that the chosen path satisfies the QoS requirements. The design of the algorithm, the selection strategy of data forwarding overlay nodes, the data transmission strategy and the operating procedures are described in detail.Finally, with the help of the link emulation platform NIST Net, the data transmission test is carried on the QoS-adaptive routing overlay network system. The results indicate that, compared to the traditional data transmission method, the proposed system has a shorter transmission delay and high reliability, hence it can support end-to-end QoS effectively.