Reducing Client-Server Communication for Efficient Real-Time Web Applications: The Use of Adaptive Polling as A Case Study for Multi-User Web Applications

A key challenge of current multi-user web applications is to provide users with interesting events and information in real-time. This research reviews the most common real-time web techniques to identify drawbacks while exploring solutions to improve simplicity, efficiency, and compatibility within a client-server environment. Two solutions are proposed for enhancing the efficiency of real-time web techniques by reducing client-server communication. First, a model of browser monitoring control observes the browser activity and decides if to postpone client-server communication in the case of inactive tabs. This model was implemented and tested with results demonstrating that a significant number of client-server connections can be avoided in the browser background. These results suggest the solution can be optimised for any real-time technique as it benefits from being a developer side technique that works consistently on all browsers. Second, ‘Adaptive Polling’ is a pull-based real-time web technique to overcome bandwidth issues of the reverse AJAX method of ‘Polling’ by controlling the frequency of requesting updates from the server based on the last server response. This approach is implemented and tested with results showing how a significant number of redundant connections can be avoided while the server does not return updates. This solution is a good alternative to other real-time web techniques as it features low latency, the simplicity of implementation, and compatibility with all browsers and servers.

URI

http://hdl.handle.net/10454/17341

Type

Thesis

Qualification name

PhD