Approximation caching algorithms for energy-efficient networks

Fueled by the increasing demands for content, Internet has become one of the leading players in energy consumption, with a worldwide share of more than 10%. Network devices typically consume close to the maximum energy even if lightly loaded. Hence, straight-forward energy saving techniques that power-off network devices during periods of low demand constitute the most promising mechanism for reducing energy expenses. In this work, we show how caching policies, that place popular content close to the requesters, can bring opportunities for powering-off network devices. We then formalize the energy-minimizing caching problem, prove that it is NP-Hard to approximate within any constant factor, and present a bicriteria approximation solution. Trace-driven numerical results indicate the superiority of our approach as compared to traditional caching schemes.