.png){kind=icon}
Return
Edge process mining for indoor positioning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.iot.2026.102072.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Indoor positioning is a key requirement for assisted living and smart home applications, where continuous monitoring of daily routines must be accurate, non-intrusive, and privacy-aware. However, most indoor positioning systems produce low-level sensor streams that are difficult to transform into structured event logs suitable for process mining. This limits their ability to support interpretable behavioural analysis beyond simple location tracking. This paper presents PositIoT, a holistic solution that integrates every component required to perform process discovery and behavioural pattern detection for indoor positioning. The edgebased framework transforms Received Signal Strength Indicator (RSSI) measurements collected from Bluetooth Low Energy (BLE) beacons and ESP32 receivers into process-aware behavioural event logs. In this way, PositIoT links low-cost indoor localisation with edge process mining, enabling daily routines to be analysed without relying on cloud-based processing of sensitive location traces. The framework was evaluated through an initial six-day feasibility study involving a single user in a real residential assisted-living environment. The experimental results show that PositIoT achieved 99.75% accuracy for room-level classification and 96.49% accuracy for position-level classification. In addition, the system detected the configured behavioural rule violations in real time and generated XES logs from which daily routine models were discovered using process mining techniques. These results show that low-cost edge infrastructure can support accurate indoor positioning and process-aware behavioural analysis in smart home environments.
Keywords:
Edge process mining
Indoor positioning
IoT
Machine learning
Smart homes
Journal
IF:
7.6
Papers:
2.0K
Citations:
6.9K
Organization
Cited Papers
No cited papers available