Sensor fusion merges measurements that have different strengths and failure modes. The classic example is an IMU: an accelerometer is noisy but drift-free, a gyroscope is smooth but drifts, and fusing them with a filter yields a stable orientation. Fusion also spans sensor types, such as combining a camera with radar in a vehicle, or temperature and humidity with a door contact to infer whether a cold room was left open.
In IoT, fusion can happen on the device, on a gateway or in the cloud, and moving it closer to the sensors reduces the data that needs to be sent. The cost is complexity: sensors must be time-synchronized and calibrated, and a fused output can hide which input is at fault when something goes wrong.
Part of the IoT Glossary. Spotted an error or a missing term? Tell us.