Meetup 4/10/2016 - Het IoT platform van de stad en security
1. CONFIDENTIAL – INTERNAL USE
SERVICES & RECOMMENDATIONS ABOUT SECURITY, PRIVACY & ETHICS
THOMAS KALLSTENIUS, PROGRAM DIRECTOR DISTRIBUTED TRUST
ACKNOWLEDGEMENTS:
GUNES ACAR (IMEC– COSIC, KULEUVEN) & ROB HEYMAN (IMEC-SMIT,VUB)
04.10.2016
9. CONFIDENTIAL – INTERNAL USE
Cryptography
for IoT
• Elliptic curve
cryptography
• one point
multiplication <5µJ
• Based on optimized
HW and SW co-
design
12. RF DISTANCE BOUNDING
imec’s group COSIC has developed improved RF distance bounding protocols
With Secure localization and Key management schemes (pairing protocols)
13. CONFIDENTIAL – INTERNAL USE
Dynamic Policies for
Shared
Cyber-Physical
Infrastructures under
Attack
14. CONFIDENTIAL – INTERNAL USE
Flagship project
City of Things
• IoT reference living lab and
technology lab in Europe
• for international and local stakeholders
• to create, test and validate IoT services,
applications and technologies
• in a large scale, real life and real time
smart city environment
16. CONFIDENTIAL – INTERNAL USE
CITY OF THINGS:THREE LAYERS
Network-layer
Deploying a city-wide network
connecting multiple wireless technologies
Data layer
Providing an open data platform with a
real-time view on the city
Business layer
Living lab and analytics infrastructure
for evidence-based innovation
18. CONFIDENTIAL – INTERNAL USE
“
”
18
PRIVACY IS SECRECY FOR THE
BENEFIT OF THE INDIVIDUAL WHILE
CONFIDENTIALITY IS SECRECY FOR
THE BENEFIT OF THE ORGANIZATION
- Ross Anderson, 2008
19. CONFIDENTIAL – INTERNAL USE
PRIVACY MODEL FOR COT USERS
MARTÍNEZ-BALLESTÉ ET AL.’S 5-DIMENSIONAL PRIVACY MODEL
20. CONFIDENTIAL – INTERNAL USE
RE-IDENTIFICATION ATTACKS
“… 87% (216 million of 248 million)
of the population in the United
States had reported characteristics
that likely made them unique based
only on {5-digit ZIP, gender, date of
birth}”
- L. Sweeney, 2000.
Based the 1990 US Census summary Data
21. CONFIDENTIAL – INTERNAL USE
Every combination of quasi-identifiers should be shared by at least k respondents
Example. 3-anonymous table
1. PRIVACY-PRESERVING DATA PUBLISHING (PPDP)
K-ANONYMITY
22. CONFIDENTIAL – INTERNAL USE
HOMOGENEITY ATTACK
AGAINST K-ANONYMITY
Alice was born in September 1953 and her ZIP code is 2010
→ She has lung cancer
23. CONFIDENTIAL – INTERNAL USE
BACKGROUND KNOWLEDGE ATTACK
AGAINST K-ANONYMITY
Bob was born in July 1960, his ZIP code is 2001 & runs every day
→ He has Hepatitis A
24. CONFIDENTIAL – INTERNAL USE
1. PRIVACY-PRESERVING DATA PUBLISHING (PPDP)
L-DIVERSITY
Each equivalence class (same quasi-identifiers) must be associated with at least
L distinct values for a sensitive attribute. Example 4-anonymous, 3-diverse table
25. CONFIDENTIAL – INTERNAL USE
4. SKEWNESS ATTACK
AGAINST L-DIVERSITY
Alice was born in September 1953 and her ZIP code is 2010
→ She has HIV with 50% probability
26. CONFIDENTIAL – INTERNAL USE
Requires the distribution of a sensitive attribute in any
equivalence class to be close to the distribution of the
attribute in the overall table
1. PRIVACY-PRESERVING DATA PUBLISHING (PPDP)
T-CLOSENESS
27. CONFIDENTIAL – INTERNAL USE
LOCATION PRIVACY ATTACKS
3 months of credit
card records
1.1 million people
4 spatiotemporal
points are enough to
uniquely reidentify
90% of individuals.
- de Montjoye et al, 2015.
28. CONFIDENTIAL – INTERNAL USE
5. LOCATION PRIVACY ATTACKS
THE WEALTHY & WOMEN ARE MOST AT RISK
Reference: de Montjoye et al, 2015
29. CONFIDENTIAL – INTERNAL USE
“
”
29
IT TAKES 20 YEARS TO BUILD A
REPUTATION AND FIVE MINUTES TO RUIN IT.
IF YOU THINK ABOUT THAT, YOU'LL DO
THINGS DIFFERENTLY ..
WARREN BUFFET,American business magnate, investor and philanthropist