Two Proposed Algorithms To Improve Mobile Browsing Using Text Summarization And Cache Management

Abstract

Mobile devices have recently become the favorite way for providing services. Mobile Internet access suffers from many problems of which are limited battery power, limited bandwidth, high mobility, and an increasing degree of heterogeneity. The performance of mobile browsers which supports these devices has become very critical bearing in mind that with the rapid growth of the small screen audience, the need for less cumbersome methods of browsing information on the Internet is growing too. This thesis is a contribution to improve the performance of mobile browsers through two ways: (i) Reducing the size of the text data by summarization of the web page content. (ii) The extension of the mobile cache memory size by using cooperative caches from other cooperating mobile devices, and designing an efficient page replacement algorithm. The thesis evaluates and implements a method for text summarization, that selects the most important sentence to be ready for web adaptation in the next step; this method is called the hybrid method. The method for text summarization is based on sentence weight. The factors which contribute to sentence weight are key words weight, title words, sentence position, and clue words. Here the abstract sentences are used in comparing the efficiency of the Hybrid method of the text summarization. Cache mechanisms have been proposed to improve access latency, reduce battery power consumption, and reduce bandwidth usage in the wireless Internet. Since mobile cache size is too small, an efficient cache replacement algorithm is needed to serve as much as possible the web page from the local cache. In order to increase mobile cache size, the proposed algorithm used a cooperative cache through ad hoc networks. The thesis proposes and implements Cooperative Caches Combined Criteria (C4) Algorithm that maximizes the hit ratio depending on a set of criterias, taking into considerations document size and cache size. These criterias are Access frequency and last access time, Document size, Distance, Number of replicas of the document, and Coherency. The performance of the proposed algorithm has been shown to be better than that of the Least Recently Used (LRU) and Least Frequently Used (LFU) in most cases and it approaches the ideal performance where the cache size is infinite