Danish Internet Day Security of the Internet of Things Mitigating - - PowerPoint PPT Presentation

1

Farell FOLLY, Ph.D Researcher

folly.farell@unibw.de Copenhague, October 1st.

Security of the Internet of Things

Mitigating infections spread through immunisation techniques

Danish Internet Day

2

Agenda

• 1. Introduction to the IoT
• 2. Security Challenges in IoT
• 3. State of the Art of the IoT Security
• 4. Immunisation Techniques and our Approach

3

Introduction

“The Internet of Things (IoT) is a network of dedicated physical

• bjects (things) that contain embedded technology to communicate

and sense or interact with their internal states or the external environment.”

• Gartner

4

Introduction

• 1. Physical World
• People, Devices
• 2. Virtual World
• Applications, Digital Artefacts
• 3. Processes
• Actuation, sensing, etc.

5

Introduction

UML representation of IoT Domain Model

Source: « Enable Things to talk », Designing IoT solutions with the IoT Architectural Reference Model, Alexandro Bassi et al, Springer Edition, ISBN : 978-3-642-40403-0

6

Security Challenges

• Number of devices
• High diversity of things
• Variety and number of Applications
• Speed of change

… Hard to plan for a systematic Approach for Security

7

Security Challenges

• Unpredictable attacks
• Cybercriminals keep improving their techniques
• Zero-Day attacks
• Propagation of vulnerabilities towards billions
• f devices

8

Security Challenges

Packet Inspection Blocking Traffic based

• n flags, Signatures

Security built around static schemes Almost infeasible in this context (Big Data) Not adaptive / too much human intervention IoT networks are highly dynamic

Traditionally IoT Context

9

Security Challenges

• 1. 70% of the most commonly used IoT devices contain

vulnerabilities.

• 2. 56% of respondents say that it is “unlikely or highly

unlikely” that their organisation would be able to detect a sophisticated attack.

• 3. 253 billions of free Apps (2017).

Malicious apps (malware): the increase in the number

• f apps on the device increases the likelihood that

some may contain malicious code or security holes

Source: EY insights

10

State-of-the-Art

11

State-of-the-Art

• No universal framework or common approach

for IoT security

• Most devices are not primarily designed with

security and interoperability in mind

• Many manufacturers mostly rely on existing

traditional security measures

12

State-of-the-Art

• So far, no security approach tackles all security aspects associated

with the IoT

• Many projects do exist, however; that address some specific

concerns:

• 1. NEBULA: www.nebula-fia.org
• 2. uTRUSTit: www.utrustit.eu
• 3. IoT-A: www.iot-a.eu
• Many authors considered the use of the Graph Theory to cope with

the size of IoT networks and their dynamics

13

State-of-the-Art

• Auto-immunity
• Security and Identification.
• Trust
• deterministic, policy-based, reputation-based, social

network-based

14

State-of-the-Art

• Cognitive approach

(Context-Awareness)

1. Enable devices and network with the intelligence to perceive things 2. Adaptive actions based on continuous learning in a hostile environment

Tetrahedron model in the IoT context

Source: A roadmap for the Security in the Internet of Things, Arbia et al (2018).

15

Our Approach

16

Our Approach

• Mathematical underpinnings for Science-based Cybersecurity.


United States Department of Energy.

“Since we can never produce a 100% secure general system or network, we need methods to mitigate the spread of damage.”

17

Our Approach

• 1. Minimise exposure factor
• 2. Control how threats spread
• 3. Design an efficient patch or vaccines

distribution mechanism

Immunisation

18

Our Approach

NoN model with three layers

Source : Towards a Networks-of-Networks Framework for Cyber Security,

Mahantesh Halappanavar et al.

19

Our Approach

Is the system in danger?

• Risk increases
• Vulnerability reaches a

threshold

• An infection is spreading

What action to take?

• Minimise Risk ⬄ Maximise Entropy
• Reduce exposure factor
• Trigger updates / recovery processes towards

specific targets

• The most exposed (boundary nodes,

important links, giant clusters, inside dominant set, etc.)

• Use graph theory analysis to find the

most suitable metrics and influence them accurately : Immunisation algorithms

This node is having too many links !!! This cluster is having too many members, is it possible to disconnect some or move them to another cluster ?

20

Our Approach

Graph Theory Immunization

IoT Security

21

Our Approach

Graph Theory Immunization Graph-based IoT Representation Graph-based Security metrics Type of graph

IoT Security

• Clustering
• Centrality
• Betweenness
• Reachability
• Percolation
• …etc.

Graph Structure

22

Our Approach

Graph Theory Immunization Graph-based IoT Representation Graph-based Security metrics Analogy of Human Immune system Infection propagation and containment Infection propagation Infection containment Type of graph Graph Structure

• Immune system learning : Detect (trust rating,

classify as fraudulent or legal, semantic analysis)

• Activate virtual Antibodies to heal the rest
• r to efficiently stop the propagation.

IoT Security

• Clustering
• Centrality
• Betweenness
• Reachability
• Percolation
• …etc.

23

Our Approach

• Epidemic process : Susceptible-Infected-Recovered
• 1. How fast does an infection spread
• 2. What is the threat strategy?
• 3. What is the IoT network topology?
• 4. How resistant are the nodes/Clusters?

Infectiousness

Resistance Topology Strategy

24

Our Approach

• Graph challenges : clustering, groupings, and simplification

Apps Devices Original graph topology Devices interconnection Forming cliques Final compressed graph

25

Summary

“In parallel with the increasing autonomy of things to perceive and act on the environment, IoT security should move towards a greater autonomy in perceiving threats and reacting to attacks, based on a cognitive and systemic approach”

• Arbia et al.

26

Use cases discussions

• 1. A trusted device connects to a car and has been granted

permission to launch a service, What are the requirements of such an equipment?

• 2. How does a human decide to collaborate with a random

person?

Picture credit to @Gartner

27 folly.farell@unibw.de www.twitter.com/__ff__ www.linkedin.com/in/farellf

Akpé kaka !