Providing diagnostic network feedback to end users on smartphones

Despite the continued technology advances for smartphones, there are still few tools available to end users for understanding the causes of the network problems when encountered. The users often have to rely on the signal bars, which unfortunately is a poor indicator of real network issues that may arise even when the signal is good. In this paper, we present a network troubleshooting system for Android platform, which can automatically collect network measurements and perform on-device diagnosis of likely causes of the network problems. The diagnosis model is based on a decision tree that was constructed with domain knowledge and the model's thresholds were learned statistically from empirical performance analysis. Experimental results show that the proposed approach is effective in troubleshooting network problems on smartphones. By continuously using this app, end users can make informed decisions to upgrade service plans, change mobile devices, or switch service providers.