Landing on the Mobile Web: From Browsing to Long-Term Modeling

Browsing the web has become a common task performed using personal mobile devices, resulting in significant access network and bat- tery limitation challenges. Efforts to alleviate these challenges are commonly based around approaches incorporating elements of on-de- vice and network optimizations. Energy-ef cient mobile web content delivery has, in turn, attract- ed a significant body of research and practical developments. However, the efforts put forth today might not result in long-term applicable results should the underlying characteristics of the mobile content change drastically over time. As caching frequently used data locally is a com- mon initial approach employed to limit network traf c and energy expenditures while “on the go,” we evaluate the long-term suitability of approxi- mating a basic set of parameters for a cache and request behavior model using a popular large data set. We present a convenient approach that can employ a general approximation of param- eters over time. Our long-term modeling of the underlying factors results in an acceptable level of peak inaccuracies in simulations for more than a year’s time horizon. In turn, practitioners and researchers are enabled to readily employ model- ing and simulation approaches over a signi cant period of time with only slight impacts on their approaches and results.