Description
Date depot: 16 novembre 2022
Titre: Temperature-Sensitive Anomaly Detection for Crystal-Free Networks
Directeur de thèse:
Paul MUHLETHALER (Inria-Paris (ED-130))
Encadrant :
Malisa VUCINIC (Inria-Paris (ED-130))
Domaine scientifique: Sciences pour l'ingénieur
Thématique CNRS : Systèmes et réseaux
Resumé: The Smart Dust project proposed in 1997 by Prof. Kris Pisterenvisioned a 1 mm3 device with wireless communication,
processing and sensing capabilities [1] [2]. The project sparked the interest of the research community, acting as a
catalyst for the Internet of Things field [3]. The project was recently revisited at UC Berkeley with state-of-the-art
technology: the output is a prototype of the Single Chip Micro Mote (SCμM).
Thanks to its 2x3 mm size and wireless communication capability, SCμM can be added to any hardware device. An
example is a medical injector pen, stored in the fridge, which tracks the quantity of injected fluid. Another example is a
SCμM-controlled light bulb. In both of these use cases, SCμM undergoes rapid temperature changes: when the pen is taken
out of the fridge, or when the light bulb is turned on.
Due to the absence of a crystal, SCμM’s 4 internal oscillators drift at a rate up to 3 orders of magnitude higher than that of
an off-the-shelf radio transceiver. Tuning the radio to a desired frequency requires constant calibration and search
over tens of thousands of internal parameters. With the correct setting, SCμM runs a synchronized protocol stack and
communicates in a standards-compliant manner [4]. Sudden variations in temperature worsen the drift making
communication, at this point in time, infeasible.
Traditional wireless sensor network nodes trust their own
clock in order to generate the correct frequency, communicate and stay synced with the network. Due to the
drift involved, SCμM must additionally use the information provided by the network and exchange the drift settings with
its radio neighbors. The distributed nature of synchronizing clocks opens up an attack vector: An attacker can disturb the
network by heating up a subset of nodes to change their drift.
In this thesis, we will design, research and implement techniques to make SCμM resilient to changes in the
temperature. As a first step, you will design an algorithm to make SCμM work over the industrial temperature range. You
will test your algorithm by placing SCμM in the fridge and in the oven and ensuring it communicates with your
smartphone. You will then apply anomaly detection techniques to detect attacks on the network under
operations. You will use a hair dryer to heat up some of the nodes in the testbed and ensure that your detection
mechanism detects an unforeseen increase in temperature.
References
[1] K. Pister, J. M. Kahn and B. E. Boser, Smart Dust: Wireless
Networks of Millimeter-Scale Sensor Nodes, Highlight
Article in Electronics Research Laboratory - Research
Summary, 1999.
[2] J. M. Kahn, R. Katz and K. Pister, Next century challenges:
Mobile Networking for Smart Dust, MobiCom, August,
1999.
[3] M. R. Palattella, N. Accettura, X. Vilajosana, T. Watteyne,
L. A. Grieco, G. Boggia and M. Dohler, Standardized
Protocol Stack for the Internet of (Important) Things,
IEEE Communications Surveys & Tutorials, 2012.
[4] T. Chang, T. Watteyne, B. Wheeler, F. Maksimovic, O.
Khan, S. Mesri, L. Lee, I. Suciu, D. Burnett, X. Vilajosana
and K. Pister, 6TiSCH on SCμM: Running a Synchronized
Protocol Stack without Crystals, MDPI Sensors, 2020.
Doctorant.e: Faour Sara