Wireless Network Security Spring 2015 Patrick Tague Class #20 IoT Security & Privacy 1
Class #20 What is the IoT? the WoT? IoT Internet, WoT Web Examples of potential security and privacy problems in current and near-future IoT usage scenarios Architectural changes that may address these issues 2
The Internet of Things is? What kind of things are we interested in connecting to the internet? My computer, laptop, and phone are all things has the IoT been around for 40 years? If I put a WiFi chip in a sensor and stick the sensor on the wall, did I just create the Internet of Things? When my Nest thermostat controls my heater using data from the cloud, is that the Internet of Things? My favorite IoT quote: That's not the Internet of Things, that's the Internet with Things. 3
So, the Internet of Things is? It's complicated. Everyone has their own definition. Most are something to the effect of: Allowing embedded things to collaborate to provide some sort of service to users, apps, or other things Apps can get data from some things, process the data using other things, make decisions using other things, and affect the real world using other things Many of these things are wireless 4
Example 1: Industrial IoT 5
Maintenance / service provider Final product assembler... Org. A Org. N Org. B... 6
Maintenance / service provider Final product assembler... Org. A Org. N Org. B ALERT: Broken robot arm!... 7
Maintenance / service provider Final product assembler... Org. A Org. N Org. B ALERT: Broken robot arm! Security challenge: resource... has to manage an limited device external secure session 8
Maintenance / service provider Final product assembler... Org. A Org. N Org. B ALERT: Broken robot arm! Policy challenge: how to...which providers/orgs regulate can access data? 9
Maintenance / service provider Final product assembler Org. B challenge:... Scalability/security Org. N service orchestrator has to manage all relevant device sessions Org. A... 10
Example 2: Residential IoT 11
Weather Forecast, Calendar, Scheduled events Analytics & Control Motion Light Sound Air-flow Humidity... 12
Analytics & Control Service provider's operational domain 13
Analytics & Control Customer's physical domain 14
Analytics & Control Potential private data leakage Customer's physical domain 15
Example 3: Urban/Civil IoT 16
17
Security challenge: how do devices discover each other and verify who they discovered? 18
Security challenge: how to efficiently establish secure connections with other devices? 19
Security challenge: how to validate measurements from sources (e.g., sensors, beacons)? 20
Data-Centric Issues Who owns the data? Also, who determines who owns the sensor data? How to track where data is created, transported, analyzed, stored, used as input, etc.? What data is needed? Does your application need raw sensor data as input, or will something else suffice? What information is conveyed in the data? What can your application learn from my data? 21
When is the information more than the data? 22
Occupancy Occupancy = #people in a room A sensor aggregate that is very valuable for green HVAC Rm101 Rm103 Occ = 1 Occ = 1 Rm105 Occ = 2 Hallway, Occ = 0 Rm100 Occ = 0 Rm102 Occ = 1 Rm107 Occ = 4 Rm104 Rm106 Occ = 0 Occ = 1 It's tempting to say that occupancy is privacypreserving (in fact, many people have said it) 23
Occupancy + Context Rm101 Rm103 Occ = 1 Occ = 1 Rm105 Occ = 2 Rm107 Occ = 4 Hallway, Occ = 0 Rm100 Occ = 0 Rm102 Occ = 1 Rm104 Rm106 Occ = 0 Occ = 1 Directory: Rm100: Aaron's office Rm101: Beth's office Rm102: Carlos's office Rm103: Dennis's office Rm104: Evelyn's office Rm105: Shared lab Rm106: Kitchen Rm107: Boardroom 24
Dynamic Occupancy Rm101 Ot = 1 Rm103 Ot = 1 Rm105 Ot = 2 Ot+1 = 0 Ot+1 = 1 Ot+1 = 2 Rm107 Ot = 4 Hallway, Ot = 0, Ot+1 = 0 Rm100 Ot = 0 Rm102 Ot = 1 Rm104 Ot = 0 Rm106 Ot = 1 Ot+1 = 0 Ot+1 = 2 Ot+1 = 0 Ot+1 = 1 Ot+1 = 4 Directory: Rm100: Rm101: Rm102: Rm103: Rm104: Rm105: Rm106: Rm107: Aaron's office Beth's office Carlos's office Dennis's office Evelyn's office Shared lab Kitchen Boardroom 25
Occupancy Tracking Sufficiently fine-grained occupancy data permits location trace reconstruction of building users Context information permits labeling of location traces with user identity Occupancy Traces 1 1 0 1 2 2 0 1 2 4 1 1 1 3 Machine Learning M Labeled Location Traces u1: 202 Rooms 200 201 202 203 205 Floor Plan 2 Pantry 209 208 207 206 Toilet 3 T Office Users u1 u2 Info. Office 202 207 Timeline: 1 un: 201 202 209 202 208 un 206 26
Accuracy Privacy Risk Augsburg benchmark dataset w/ synthetic data; Estimation using FHMM + modified Viterbi algorithm 27
How can we address these issues? 28
Analytics & Control Potential private data leakage Customer's physical domain 29
Application Provider Higher-Level Analytics Shared with the Provider Local app Analytics & Control Operated by the Customer Local cloud can provide connectivity, discovery, mgmt, mediation, etc. as services 30
A Few Considerations Local cloud resources can use trustworthy computing principles to securely house 3rd-party software (just like a mobile phone) Mediating gateway can actively control information flow between internal devices and third-party resources Active migration within the local domain can help with (near-)real-time CPS requirements 31
Migration 32
How could two orgs collaborate in a constructive, efficient, privacypreserving, manner? Federation Org. A Org. B 33
Generalized IoT Domain Model Intra-domain: everything is managed locally/privately by the domain controller Org. A Inter-domain: domain controllers initiate, mediate, and manage interactions 34
Take-Away Points IoT Internet (or WoT Web) Domain federation/mediation model allows for finer-grained control of collaboration, sharing, etc. common to IoT applications Domain model comes with its own challenges, so still a lot of work to be done 35
Apr 2: Progress Presentations Apr 7: Exam Apr 9: Telecom Security & Privacy 36