PulSec: Secure Element based framework for sensors anomaly detection in Industry 4.0 - Université Gustave Eiffel Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

PulSec: Secure Element based framework for sensors anomaly detection in Industry 4.0

Résumé

In Industry 4.0, the optimization decisions are obtained by applying extensive machine learning algorithms on huge data sets collected from wide array of sensors. This data is obtained aperiodically at the cloud end using a centralized aggregator present on the industrial site. This aperiodicity of data makes monitoring of the sensors difficult as it is hard to accurately estimate when the cloud end should expect data from a particular sensor. The reliable functioning of the sensors is important in industry 4.0 context as the efficiency of the applied machine learning algorithms directly depend on it. Above this, there are some processes which are critical and hence are monitored in real-time. Their state is important for many critical decisions (like in demand response) and immutable, tamper-proof status reporting is of prime concern. In this paper, we propose a secure element based framework called PulSec which emits a periodic secure pulse (like a human heartbeat) to solve these intricate problem arising out of a need to effectively monitor remote industrial sites. We show how anomalies, tampering can be quickly, effectively and accurately determined by just analyzing the secure pulse. We demonstrate its low overhead (bandwidth = 7.73 B/s, memory = 464 B, time = 500 ms), tamperproofness, security, portability using a realized prototype based on Multos M5-P19 Secure Element.

Dates et versions

hal-04487977 , version 1 (04-03-2024)

Identifiants

Citer

Varun Deshpande, Laurent George, Hakim Badis. PulSec: Secure Element based framework for sensors anomaly detection in Industry 4.0. IFAC Manufacturing Modelling, Management and Control (IFAC MIM), Aug 2019, Berlin, Germany. pp.1204-1209, ⟨10.1016/j.ifacol.2019.11.362⟩. ⟨hal-04487977⟩
9 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More