Smart Caching for Continuous Broadband Services in Intermittent Wireless Networks

The tremendous growth of Internet traffic and the increasing demand for anywhere-anytime broadband access are the main drivers for next generation
mobile radio networks’ development. Future radio networks have to accomplish the balancing act of ubiquity and high data rate support. As the performance of radio networks vitally depends on the characteristics of the transmission path between user terminal and Access Point the coverage of larger areas with broadband access will be hardly achievable. Although remaining gaps are covered by legacy cellular networks, mobile users will perceive network depending continuously changing conditions while roaming between the different networks. For broadband services this eventually results in intermittent network connectivity. Network heterogeneity will make it hard to support applications like, e.g., high quality video streaming.
To achieve a broad acceptance radio networks must provide data intensive services even in case of patchy broadband coverage. This thesis introduces
a new technique, called Smart Caching, able to mitigate variations in network performance so that non-real-time and non-interactive services’ quality is improved, substantially. Smart Caching supports pre-fetching from a server and data buffering at the edge of the core network. While a mobile terminal is in the service range of an Access Point, the Smart Caching concept incorporates extreme high data delivery with data buffering in the terminal for later consumption. Contrary to legacy techniques the data transfer rate is much higher than the data service rate. For evaluation of the Smart Caching potential of allowing communication services even under intermittent connectivity, queuing theory models are employed. The chosen queuing systems allow a detailed modeling of all relevant system parameters. Therefore models for the source traffic of several different user applications like VoIP or video streaming are developed and used in the arrival process of the queuing model. With large care the modeling of the service process is exercised. Due to a strong relation between service process and user velocity realistic mobility models for pedestrian and vehicular users are employed.The results of the analysis of Smart Caching allow the comparison with  legacy network setups. The general advantage in terms of increased average data transfer rate and a higher capacity is clearly outlined. Dimensioning guidelines for Smart Caching enabled networks are given depending on the scenario and user’s mobility. The thesis shows the benefit of employing Smart Caching while its contribution is twofold: First, it provides an encompassing performance evaluation and second dimensioning criteria have been developed to enable the design of future networks.