Context-Aware Handoff on Smartphones

Nowadays smartphone users often enjoy the availability of multi-networks by switching between the networks for better network performance, energy efficiency of smartphones, and more data offloading to less expensive networks. However, network switching inevitably brings about network disruptions leading to user experience degradation. In this paper, we propose an application context model that is used in conjunction with a heuristic network selection mechanism, which selects a network using three metrics (i.e., network performance, energy consumption, and cost). A Bayes classifier is used to provide a probability for the network selection given applications running during network disruptions. We construct the classifier via crowd-sourced data by considering smartphone users profile and the operating environments. We implement a prototype context-aware handoff on the Android platform and conducted an experiment in a real world scenario through one case study, switching between cellular and WiFi networks. The evaluation results suggest that context aware handoff achieves 25% energy cost, nearly one-third data offloading, and more than twice throughput with only one third of the network switchings.