Dynamic Protocol Selection for Communication Systems

Abstract

Advances in networking are leading towards intelligent communication protocols, particularly with respect to Quality of Service. Traditional communication protocols are static which results in lower performance in the network. Such protocols concentrate on reliability related Quality of Service aspects and ignore aspects such as synchronization, security, and negotiation of Quality of Service. In this thesis we propose dynamically changing communication protocols based on the user requirements and best available Quality of Service provided by the underlying network. The dynamically changing protocol system is built with differing transport characteristics. These include connection oriented, connectionless, error-free, limited error reduction, delivery of messages in the order in which they were sent or messages with no guarantee about the order of delivery, security, no security compared to the traditional fixed end-system protocols. This will meet fast changing user requirements such as request-response, bulk transfer, security, and real-time data transfers. The dynamically changing communication protocol proposed here begins the communication with the best available protocol and then changes if needed to the best protocol based on measured network Quality of Service parameters such as bandwidth, jitter, and delay. Hence optimal performance is achieved in data communication when loss, errors or congestion is encountered in the network. The Analytical Hierarchy Process, a powerful and flexible decision making process is used for selecting the appropriate communication protocol from the protocol stack based on Quality of Service parameters. Results demonstrate that the proposed dynamically changing protocol achieves better throughput and less packets loss compared to traditional systems under changing network conditions.