A Data Intensive Reputation Management Scheme for Vehicular Ad Hoc Networks Anand Patwardhan Doctoral Candidate Department of Computer Science and Electrical Engineering University of Maryland Baltimore County Anand Patwardhan, Anupam Joshi, Tim Finin, and Yelena Yesha V2VCOM 2006 Outline • • • • • • Data management in VANETs Security perspective Trust-based security Distributed data-intensive reputation management Algorithm for screening data Simulation results GPS satellite Localized and distributed Wireless Access points Localized Info-Stream Services Hazard warnings, Detours, Inclement weather, Road conditions, Traveler info. Various forms of connectivity GSM, GPRS, EDGE, E-VDO WiMax VANET connectivity Location & directions GPS Update propagation Onboard Computer with various sensors: •GPS location •Cameras •Engine Condition •Tire pressure etc. Situation Awareness allows Adaptation Objectives • Objectives • Situation awareness for smart-vehicles • adapt to current conditions • optimal utilization of surface transport infrastructure • Provisioning context sensitive travel information locally and directly • a growing need to provide context-sensitive information to mobile handheld devices and car-computers with travel related information) • Distributed control and fault tolerance • ensure continued functioning in face of infrastructure failures arising from natural calamities or terrorist attacks • Prevalent Enabling Technologies • Smart cars with arrays of sensors (GPS, cameras, etc.) • Multimodal wireless communication (GSM, WiFi etc.) • Distributed sensor networks embedded in the transport infrastructure Background • • • • • • • Highly dynamic conditions Lack of centralized trust authority Data and security guarantees Information processing and decision making Distributed collaborative processes Softer security guarantees Trust based security Dynamic conditions • Network • Mobility of devices • Arbitrary topologies • Limited connectivity • Mobility • Time frames important (message transmission and surface velocity) • Radio ranges, interference, and obstructions • Environment • Road conditions, congestion, inclement weather, hazards etc. Trust and Risk Management • Conventional PKI, variants, or Web-of-Trust (PGP) infeasible • Limited connectivity • I&A difficult • No guarantees of intent • Security properties • Confidentiality, integrity – cryptographic methods • Availability – multiple sources, epidemic updates • Reliability of source? • Malicious entities, selfish-interest, non-cooperative nodes? VANET Security Perspective • Data • Authenticity, reliability (quality), and timeliness • Network • Reliable routes • Cooperative and trustworthy peers • Intrusion and fault resilience • Identification and Authentication • Unique persistent identifiers (e.g. SUCVs) • Decentralized reputation management Examples of collaborative processes • Routing • On demand route setup • Maintenance • Data dissemination • Relay data packets for others • Caching • Intrusion detection • Reputation management • Service discovery Stimulating collaboration • Cost of collaboration • Storage • Communication • Reputation management • Self-interest • What is the payoff? (incentives) • Higher availability (cooperation) • Improved response times • Reliability • Reciprocity (tit-for-tat) • Avenues for recourse Data dissemination model • Anchored sources (trusted) carousel information updates • Mobile devices propagate these further via epidemic updates (collaboration) • Burden of collecting relevant information and verifying it is placed on the consumer devices • Validation of data is achieved either • Trusted source (trivial case) • Agreement • Post-validation by trusted source Segment validation algorithm Simulation setup • • • • • • • • • • Glomosim v. 2.0.3 Transmission range 100m Simulated area: Dupont Circle, Washington DC Geographic area of 700m by 900m 802.11 Mobility speeds 15 to 25 m/s Pause times of 0 to 30 s 38 anchored resources (trusted) 50 to 200 mobile devices (vehicles) Simulation time: 30 mins Simulated area Autonomous and Assisted 36 36 34 34 32 32 30 30 28 26 26 26 5-6 24 24 22 4-5 20 3-4 20 18 2-3 16 1-2 16 0-1 14 16 Anchors 14 22 12 12 10 10 8 8 6 6 4 4 2 2 0 1 3 5 7 9 11 13 15 17 19 21 23 Time (mins) 25 27 29 Trusted sources only Anchors 0 1 3 5 7 9 11 13 15 16 19 21 23 25 26 29 Time (mins) Trusted sources and assisted 5-6 4-5 3-4 2-3 1-2 0-1 Validated segments 36 34 32 30 28 26 24 22 20 18 Anchors 16 14 12 10 8 6 4 2 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Time (mins) 5-6 4-5 3-4 2-3 1-2 0-1 Effect of malicious nodes 1400 1400 1400 1200 1200 1200 1000 1000 1000 800 800 50 800 100 150 600 600 400 400 400 200 200 200 0 600 200 0 TD0 VD0 ID0 0% malicious TM0 50 50 100 100 150 150 200 200 0 TD30 VD30 ID30 30% malicious TM30 TD60 VD60 ID60 TM60 60% malicious Ongoing and Future work • Distributed data-intensive reputation management • Trust relationships built using persistent identities for further trustworthy collaboration: • Basis for Distributed intrusion detection • Service discovery • Reciprocative/adaptive levels of cooperation • Contention management • Adaptive radio-ranges to increase throughput Questions?