IIIT-H Staff Builds Device To Prevent UPS Failures
“Power outages are common, but affordable ways to understand how UPS systems behave during those moments are rare,” said Sannidhya Gupta, one of the researchers
Hyderabad: When power cuts disrupted servers across campus, the problem was not the outage itself but the silence that followed. Despite UPS units being installed everywhere, no one could tell what actually happened inside them when electricity failed.
However, IIIT-H staffer Prakash Nayak has come up with a low-cost scientific solution — a device based on the Internet of Things (IoT) that monitors UPS behaviour during power outages with near-second accuracy, helping identify failures that previously went undetected.
“Power outages are common, but affordable ways to understand how UPS systems behave during those moments are rare,” said Sannidhya Gupta, one of the researchers. Commercial monitoring tools exist, but cost more than Rs 20,000 per unit and often fail to record data during outages, exactly when information is most needed.
Instead of tapping into UPS internals, the team designed a non-intrusive device that clamps onto input and output lines to observe current flow before, during, and after outages. Crucially, the device runs on its own battery, allowing it to keep recording even when mains power and internet connectivity drop.
To test the system, the device was deployed across 4 UPS installations on campus over a month. During this period, it logged about 3.7 million data points and automatically detected 61 outage events. The data confirmed long-standing suspicions within the IT team. “One UPS showed repeated failures to recharge properly after outages, even though it appeared functional,” Prakash Nayak said.
The backend software automatically classifies each event into phases such as outage, stabilisation, and battery charging, without manual configuration. According to the researchers, the system recorded no missed outages and no false alarms, with timing errors usually within 3 seconds.
The findings were presented in a paper titled Low-cost IoT-based Downtime Detection for UPS and Behaviour Analysis, which received the Best Paper award at the International Conference on COMmunication System and NETworkS 2026 workshop on AI of Things.
Built using off-the-shelf components, the device costs about `2,000 and offers real-time dashboards and historical analysis for administrators. “The goal was not elegance, but clarity,” Gupta said.
“This began with an operational problem, not a research proposal. Treating IT staff as co-creators rather than end users made the research relevant and usable,” Prof Sachin Chaudhari said.