A research preview from Kaunas University of Technology

Every digital habit leaves an electronic trace.

The Internet of Habits (IoH) is a four-layer behavioural model that explains how everyday digital activity — the way you type, the apps you open, the Wi-Fi you join, the IoT devices you talk to — composes a forensic habit-print that can prevent cybercrime before it materialises.

Try the simulator Explore the model

Behavioural biometrics
Digital forensics
NIS2 ready
GDPR by design

The four-layer IoH model

Heterogeneous artefacts are projected through the extractor Φ into a per-day, 37-dimensional habit-print vector. The model is preservable as evidence under ISO/IEC 27037 and ISO/IEC 27043 and aligns with the NIS2 reporting timeline.

Input dynamics

Keystroke timings
Pointer/touch dwell
Gait and motion

Application & files

App focus
File open / save
Browsed URLs

Communication

Email metadata
Chat / IM
VoIP / SIP records

Mobility & IoT

Wi-Fi associations
GPS / cell pings
Smart-home events

Interactive habit-print simulator

Set the parameters of a synthetic subject, optionally inject an impostor on selected days, and watch how the habit-print, the day-by-day anomaly score and the impersonation ROC respond. Everything runs in your browser — no data is uploaded.

Population Subjects (N) 20 Days observed 60 Random seed

Habit profile (focal subject) Chronotype Typing speed (ms dwell) 120 Daily artefact volume 220 App / domain diversity 6 Weekend factor 0.50

Adversary Impostor onset (day) 40 Impostor share of artefacts (%) 20 Impostor profile drift 0.60

Detector Decision threshold τ 0.65 Baseline window (days) 20

ROC-AUC—

Best F1—

Precision—

Recall—

Impostor days—

Daily anomaly score

Normal day Impostor day Threshold τ

Habit-print radar (subject vs. impostor)

Legitimate subject Impostor draws

Channel mix (last day)

ROC of impersonation detection

Share these settings (permalink & JSON)

Headline results from the paper

Validated on a synthetic 30-user, 60-day, ~343 000-artefact corpus (seed = 42).

99.1 %

Top-1 user attribution
5-fold CV, 30 classes

100 %

Top-5 user attribution
from one day's habit-print

0.71

Per-day impersonation
ROC-AUC

69 %

Impostor recall
at 18 % precision (best F1)

About the author

Dr. Šarūnas Grigaliūnas is a researcher at the Department of Computer Science, Kaunas University of Technology (KTU). His work focuses on digital forensics, cybercrime prevention, NIS2 / KSĮ implementation and post-quantum cryptography readiness for critical infrastructure operators.

Contact: sarunas.grigaliunas@ktu.lt · ktu.edu