A COMPARATIVE ANALYSIS OF BEHAVIORAL BIOMETRICS DATASETS ON MOBILE AND DESKTOP DEVICES IN TERMS OF COVERAGE, DURATION, AND USER DIVERSITY

8. Uluslararası İstanbul Bilimsel Araştırmalar Kongresi, İstanbul, Türkiye, 28 - 30 Aralık 2025, ss.1132-1142, (Tam Metin Bildiri)

Smartphones and desktop computers are now actively used in almost every field, changing expectations

for user authentication systems. Authentication methods, traditionally focused on the login moment,

cannot adequately address the security risks that may arise during the session. Therefore, using the

behavioral characteristics exhibited by the user during interaction with the device for authentication

purposes is becoming increasingly important. However, behavioral biometric datasets used on mobile

and desktop platforms differ significantly in terms of sensors used, user profiles, data collection times,

and scenario diversity.

The purpose of this study is to comparatively examine behavioral biometric datasets used on mobile and

desktop devices in terms of scope, enrollment time and user diversity. For this purpose, datasets from

widely used mobile and desktop behavioral biometric studies in the literature were evaluated using a

common methodological framework. The results show that mobile datasets better reflect real-world

usage conditions, but have a more complex structure due to high contextual variability. In contrast,

desktop datasets are collected in more controlled environments and therefore provide more stable

behavioral signals. It was also found that approaches that use multiple modalities together positively

impact validation performance on both platforms, but datasets that are long-term and cover large groups

of users are still limited. By presenting the current state of mobile and desktop behavioral biometric

datasets, the study provides a guiding assessment for the development of more context-aware and

multimodal authentication systems in the future.