Network Traffic Analysis of Smart Devices

Measuring what smart devices send and receive — especially when they are supposedly idle

Smart devices talk to the network far more than their users expect, and the traffic they generate reveals a lot: which services they contact, how often, and what they do when nobody is using them. Analyzing that traffic is one of the few ways to check a device’s actual behavior against what its documentation and privacy settings promise.

This project measures smart-device network traffic to characterize device behavior, expose privacy leakage, and give users, researchers and manufacturers a shared vocabulary for what a device should be doing in each of its modes. Its first line of work defines and studies passive modes — the states in which a device is powered on but not performing its intended function.

Passive mode — what smart devices do when they are not in use

Passive mode — what smart devices do when they are not in use

“Idle” says little about what a device actually does when it is not in use: a muted smart speaker and one listening for its wake word are both idle. This work defines passive modes formally, grounded in a categorization of smart home devices, and then measures 32 devices in those modes. 97% show near-constant network activity, many initiating LAN communication with other devices — traffic that can expose users to privacy leakage even when they believe the device is doing nothing.

Highlights

  • A formal definition of smart-device passive modes — on, but not doing the job — replacing vague “idle” labels that differ from one manufacturer to the next
  • Network traffic of 32 smart home devices measured in their passive modes: 97% stayed near-constantly active, exchanging over 3M messages in 24 hours
  • Best Artifact Award at IEEE PerCom 2025, with the study’s code and data released

Team

  • Christian BadolatoPhD Student, University of Maryland, Baltimore County
  • Georgios BouloukakisAssistant Professor, University of Patras, Greece
  • Roberto YusAssistant Professor, University of Maryland, Baltimore County
  • Kaur Kullman
  • Nikolaos Papadakis
  • Manav Bhatt
  • Don Engel
Roberto Yus
Roberto Yus
Assistant Professor

My research interests include Data Management, Knowledge Representation, the Internet of Things, and Privacy.

Related