Extracting Usage Patterns of Home IoT Devices

Gevorg Poghosyan (1) , Ioannis Pefkianakis (2) , Pascal Le Guyadec (3) , Vassilis Christophides (4)

1 Insight Centre for Data Analytics [Dublin]
2 Hewlett Packard Labs [Palo Alto]
3 Technicolor [Cesson Sévigné]
4 MIMOVE - Middleware on the Move

Ubiquitous connectivity and smart technologies gradually transform homes into Intranet of Things, where a multitude of connected, intelligent devices allow for novel home automation services. Providing new services for home users (e.g., energy saving automations) and Internet Service Providers (e.g., network management and troubleshooting) requires an in-depth analysis of various kinds of data (connectivity, performance, usage) collected from home networks. In this paper, we explore new Machine-to-Machine data analysis techniques that go beyond binary association rule mining for traditional market basket analysis considered by previous studies, to analyze individual device logs of home gateways. We introduce a multidimensional patterns mining framework, to extract complex device co-usage patterns of 201 residential broadband users of an ISP, subscribed to a triple-play service. Our results show that our analytics engine provides valuable insights for emerging use cases such as monitoring for energy efficiency, and “things” recommendation.

Mots clés: IoT; home networks; association rules

Domaines: Apprentissage [cs.LG]; Réseaux et télécommunications [cs.NI]

Gevorg Poghosyan, Ioannis Pefkianakis, Pascal Le Guyadec, Vassilis Christophides. Extracting Usage Patterns of Home IoT Devices. ISCC 2017 - 22nd IEEE Symposium on Computers and Communications, Jul 2017, Heraklion, Crete, Greece. pp.1-7, ⟨10.1109/ISCC.2017.8024707⟩. ⟨hal-01664015⟩