A novel dynamic cache size tuning mechanism to support fast response on the Internet with applications to mobile and telemedicine

Caching is very important to the Internet because it alleviates network congestion and hastens WWW information retrieval. The danger of congestion is aggravated by the fact that the WWW page size has a monthly growth rate of around 15% but the Internet backbone capacity only increases by 60% yearly. The massive quantity of information requiring transfer across the network can quickly deplete the amount of sharable bandwidth. Caching can shorten the service roundtrip time (RTT) for the data retrieval and reduce the data needed for transmission across the network, thus providing more backbone bandwidth for sharing and less chance of network congestion.; The advantages from caching have motivated different areas of relevant research. However, all of the existing caching schemes work with a static cache size and they aim to yield a high cache hit ratio but do not necessarily maintain it. For this reason the cache hit ratio fluctuates with respect to the system dynamics and the current relative data object popularity profile. Maintaining a given cache hit ratio needs dynamic cache size tuning which, in contrast, works with a variable cache size.; The original MACSC (Model for Adaptive Cache Size Control) conceptual framework for dynamic cache size tuning over the Internet is proposed in this thesis. It strives to maintain the given hit ratio under all conditions by adjusting the cache size according to the relative data object popularity profile as the sole parameter. The conceptual framework is based on the Zipf-like behavior. The MACSC emphasizes supporting small, inexpensive machines. It can also work with other caching schemes such as replacement strategies and admission control.; Altogether four statistical base solutions are proposed to implement the MACSC conceptual framework, and the four solutions are, namely, MACSC(PE), MACSC((M3RT), MACSC(F-PE), and RTPD/MACSC(PE). The thesis has demonstrated how the MACSC conceptual framework can support mobile and time-critical applications such as TCM (Traditional Chinese Medicine) based telemedicine. The novel MACSC framework that maintains a given hit ratio by dynamic cache size tuning has been successfully verified with respect to the four solutions.