Flux Research Group / School of Computing

ABSENCE: Usage-based Failure Detection in Mobile Networks

Binh Nguyen, Zihui Ge, Jacobus (Kobus) Van der Merwe, He Yan, and Jennifer Yates

The 21st Annual International Conference on Mobile Computing and Networking (MobiCom) 2015.

Networking, Mobile Networking


We present our proposed ABSENCE system which detects service disruptions in mobile networks using aggregated customer usage data. ABSENCE monitors aggregated customer usage to detect when aggregated usage is lower than expected in a given geographic re- gion (e.g., zip code), across a given customer device type, or for a given service. Such a drop in expected usage is interpreted as a sign of a potential service disruption being experienced in that region / device type / service. ABSENCE effectively deals with users’ mobility and scales to detect failures in various mobile services (e.g., voice, data, SMS, MMS, etc). We perform a systematic evalu- ation of our proposed approach by introducing synthetic failures in measurements obtained from a US operator. We also compare our results with ground truth (real service disruptions) obtained from the mobile operator.