Researches On Key Technologies Of Mobile Web Browsing Systems

Web tends to become one of the mostimportant application platforms. On the other hand,mobile devices have become increasingly powerful in both hardware and software thereforethey grow rapidly in recent years. Mobile Webapplications inherit the advantages of both the Web and mobile devices therefore they could fit the trends of cloud computing and mobile computing and are widely adopted. As the clients of mobile Web applications, mobile Web browsers are the keys of the user experience of mobile Web applications.However,current full-fledge mobile Web browsers usually provide poor performance and Webpage compatibility, and drain the batteries of mobile devices quickly. Current split mobile browsers usually cause excessive data transmissions and cannot leverage the processing power of mobile devices.Mobile Web browsers must be specifically designed to meet the hardware and software limitations of mobile devices. Current researches on mobile Web browsers mostly focus on designing new algorithms for Webpage processing to improve the performance of full-fledged mobile Web browsers while reducing the computation workloads, new architectures for split mobile Web browsers and benchmark systems for evaluating the user experience of Web browsers.This dissertationmainly focuses on the key technologies of designing new algorithms for Webpage processing and a new architecture for split mobile Web browsers.Based on the concept of eliminating redundant computations and redundant network transmissions, this dissertation proposes the optimization algorithms for Webpage processing of full-fledged mobile Web browsers and the design of a new split browser. The major contributions of this dissertation are:First, this dissertationpresentsthe adaptive memoization schemes for style formatting and layout calculation during Webpage processing. Memoization algorithms could reuse the computation results to avoid redundant computations. However, traditional memoization algorithmscould reuse the cached computation results only if the input data keep exactly the same. Unlike them, when there are changes on Webpages, the memoization schemes proposed in this dissertationcould adaptively reuse either the final or the intermediate resultsfor better cache efficiency. These memoization schemes could significantly improve the efficiency of Webpage processing while reducing the computation workloads. According to the experimental results, our adaptive memoization schemes could reduce about46%of the Webpage processing time.Second,this dissertation presents the design of a new split mobile Web browser as well as the key optimization algorithms for it.Existing split Web browsers usually treat mobile devices as dump terminals and cannot cache any data on them. Unlike them, the split browser proposed in this dissertationuses the cooperative render cache scheme. The render results of a Webpage are partitioned into small bitmaps on the proxy server, and the stable bitmaps are cached on the client. Therefore the cached data are managed cooperatively by both parties.Technologies like the Render-Cache tree and the Render Cache Protocol are employed to manage the cached data whilereducing the communication overheads. Our spilt mobile Web browser could eliminate transmissions of redundant render data, leading to reduced network transmissions and better responsiveness. According to the experimental results, our mobile browser could reduce up to80%of the data transmissions for most Webpages and improve the responsiveness from about10%to up to about60%. Additionally, mobile devices participate into the Webpage processing by managing the cached data and generating the render results of Webpages. Therefore our mobile browser could leverage the processing power of mobile devices as well.In conclusion, the technologies proposed in this dissertation could improve the responsiveness of mobile Web browsers while reducing the network transmissions and power consumptions, therefore could adapt mobile browsers to the trends of the Web, cloud computing and mobile computing. Additionally, these technologies are fully compatible with current Web standardsand technologies as well as existing optimization algorithms, and have no specific hardware requirements for mobile devices. Therefore, the technologies proposed in this dissertation are valuable in both academia and practice.