Smartphones for Computing: Can They Reduce the Carbon Footprint?

Most mobile applications typically include client-side software on a smartphone for offloading a nd rendering the output and a cloud-based server for computations. However, servers incur network delays and bandwidth costs. The energy consumed by high-end servers in small organizational-level data centers potentially contributes to high carbon footprints. Edge computing and its variants, such as fog and cloudlet computing, are paradigms for processing data close to the source. However, the computing platform still remains to be a high-end server. Smartphones are loaded with multicore processors, faster clock speeds, and large RAM. Benchmark studies show that smartphones performed equally well against an Intel i3 processor. However, apart from gaming purposes, smartphones are under-utilized. In this work, we investigated the utilization of smart-phones as a computational device. To this end, we develop a mobile application that implements the complete end-to-end mo-bile sensing pipeline of gathering data, pre-processing the data, and performing classifications. O ur pilot deployment findings show that smartphones can run machine learning algorithms and consume only 350mAh battery. We received an average rating of 4/5 for user experience since users did not experience any network-related lag in the app.