SCRUB: Secure Computing Research for

Users’ Benefit

David Wagner

http://scrub.cs.berkeley.edu/

1

Security is hard

2

What is SCRUB?

• SCRUB is a new center focusing on security for user’s benefit

• Model: industry funding + collaboration– 4 Intel researchers in residence at Berkeley

– $2.5M/year in funding– Open IP policy

• Schools: Berkeley, CMU, Drexel, Duke, UIUC

3

Thin intermediation layer

Mobile security

Security analytics

Data-centric security

SCRUB

Research Agenda

Security of Desktop Computing

• Problem: today’s desktops use a security architecture based upon 1970’s-era threat model.

5

Secure Desktop Computing

• A thin, low-level intermediation layer can enable secure computation• e.g., online banking – establish an island of security amidst the sea of malware

• Benefit to users: secure computing on insecure desktops

7

Hardware

Intermediation layer

OS

Web browser

Banking app

Email

Thinclient

Securing the desktop:Thin intermediation

layer

OS

Mobile Security

9

• Huge growth in third-party apps:

Mobile Security

• How do we ensure third-party apps are safe?

• New paradigm for secure computing: protect against apps, not against users

• Benefit to users: Secure smartphones, tablets

10

How do we build effective app permission systems?

How do we build effective app permission systems?

Can we make app stores more robust

and secure?

Can we make app stores more robust

and secure?

Desktop OS

• Threat model:users attacking users

• Applications run withusers’ full privileges

11

Modern Reality

• Threat model:apps attacking users

• One user per device• Users don’t trust all apps

12

Permission Systems

User approves what permissions the application receives.

13

Does this provide security benefits?

The Good News

• Permissions do limit harm from breaches

• Developers do comply• Only 30% of apps are overprivileged, and only a little

• But can users use permissions effectively?

14

Android apps get median of 4 permissions; desktop apps get 56. Only 10% of Android apps can cost users money, and only 15% get personal info.

More work needed

• Some users can use permissions effectively• 20% demonstrated awareness and some comprehension• 20% have declined to install an app because of perms

• But, at least on Android, permissions are not effective for most users• Only 17% of Android users look at permissions• Only 24% understand Android permissions (more or less)

• Our user studies partially explain why, but more work is needed to find a good solution

15

Apps as paradigm for security

• Beyond phones: app-based platforms offer a path to securing laptops and desktops.

16

PhoneDesktop

Data-centric security

• Data increasingly resides not only on end-user devices, but also on servers, cloud, …

• Can we provide consistent protection for user data as it flows through a complex distributed system, no matter where it is stored?

• Benefit to users: visibility and control over their data

17

Data-centric security• Cloud platform protects user data and ensures that apps can’t misuse it

• Each user’s data is separately encrypted. When cloud authenticates the user, it gives app access to user’s data. Cloud prevents apps from exfiltrating the data.

• Ability to audit who has access to data.

18

cloud servercloud server

App

encrypted storage

encrypted, authenticated channel

Security analytics

• How can we accurately measure security?• Current approaches lead us in the dark

• Are our efforts to secure systems making progress?

• Goal: robust security metrics and analytics

• Benefit to users: Ability to prioritize, manage, and measure security

19

Learn more

• Come visit us:SCRUB Open House, 2-4pm, 373 Soda Hall

• Come visit the TRUST center, too:TRUST Open House, 2-4pm, 337 Cory Hall

• Follow our work:http://scrub.cs.berkeley.edu/

20