1 / 27

SenSec: Scalable Emulation of Sensor Network Security

SenSec: Scalable Emulation of Sensor Network Security. Yi-Tao Wang, Rajive Bagrodia UCLA Computer Science Department DAWN MURI Meeting October 6, 2009. Motivation. Security is an on-going process and any secure system should undergo rigid testing

gur
Download Presentation

SenSec: Scalable Emulation of Sensor Network Security

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. SenSec: Scalable Emulation of Sensor Network Security Yi-Tao Wang, Rajive Bagrodia UCLA Computer Science Department DAWN MURI Meeting October 6, 2009

  2. Motivation • Security is an on-going process and any secure system should undergo rigid testing • Important to test sensor network security with realistic environments and attacks • Increased complexity of deployed sensor networks makes testing difficult • Current simulators are more focused on validating rudimentary behavior of sensor networks

  3. Current Tools • Simulators (TOSSIM, EmTOS, etc) • Validates basic application behavior • Lacks detailed simulation models • Restricts accuracy and expressiveness of their simulations • Cannot evaluate applications in deployment conditions • Physical testbeds • Accurate • Lacks spatial and temporal scalability • Difficult to perform simulations under complex conditions • Can’t repeat simulations

  4. SenSec Framework • Simulates sensor networks under attacks to evaluate • Vulnerability of network • Network Security Protocols (Encryption, Intrusion Detection Systems, etc.) • Effectiveness of new attacks • Runs unmodified TinyOS applications • Validate behavior • Optimize design parameters

  5. SenSec Framework (cont.) • Leverages existing physical layer models in network simulators to provide more accurate simulations • Supports a diverse set of scenarios • heterogeneous networks • mobile data mules • multiple operating systems • Models for clock drift, battery, and sensing channel

  6. SenSec Framework (cont.) • Composed of independent attack emulation and TinyOS emulation systems • Supports evaluation of attacks on emulated TinyOS applications or simulated nodes • TinyOS emulation added for higher fidelity between simulations and real-life deployment

  7. TinyOS Emulation • Each sensor networknode can run anemulated TinyOSapplication • Override drivers in TinyOS to communicate with DES • TinyOS applications think DES is just another hardware platform

  8. TinyOS Emulation (cont.) • Hardware interactions communicates with a CFP in the DES that: • Uses models to return values (i.e., get sensor readings) • Queue events in the DES (i.e., send packet)

  9. TinyOS Emulation (cont.) • DES considers emulated TinyOS applications asabstract nodes • Applications run as • Threads for fasterexecution and betterscalability • Processes so moreone application can be emulated in the same simulation

  10. Control Flow of TinyOS application • DES boots eachemulated application • Application executesuntil context switch • Replaced components inTinyOS (squares)interact with CFPs (roundedsquares) in the DES • When the application has finished executing for the current time step, it will context switch (italics) back to the DES until the next time step

  11. Models: Clocks Drift • From Stochastic model estimation of network time variance. White Paper, Symmetricom. f = fnom + ∆f0 + a (t – t0) + ∆fn(t) + ∆fe(t) • f – frequency of inaccurate clock • fnom – nominal frequency • t0 – starting time • a – aging rate • ∆fn(t) – noise term • ∆fe(t) – environmental term

  12. Models: Battery • Two models • Simple: deduct amount of power for every transaction (transmission, reception of a packet, carrier sensing) • Non-linear discharge (from D. Rakhmatov and S. Vrudhula. Energy management for battery-powered embedded systems.) • If I(t) is the current drawn from the battery at time t, then battery capacity used after time T is

  13. Models: Sensing Channel • Diffusive • Assigns value at each space-time coordinate • User defines values at specific coordinates and bilinear spine interpolation is used to calculate the values at the other points • Dynamic environment can be defined with 4 corner points • Mobility • Triggered when targets are in range • Random • Random value between user defined min and max

  14. Accuracy of TinyOS emulation • Compared results from SenSec to those from Harvard’s MoteLab and other TinyOS emulators • Application routes periodic packets from one sender to one receiver

  15. Accuracy of TinyOS emulation (cont.) • Other tools are unable to match the accuracy due to less accurate models or lack of certain models • Although they can be extended with more accurate models, it’s more efficient to leverage existing models in network simulators

  16. Attack Emulation • Attack Emulation System • Parses defined attacks from user or attack repository • Validates attack (i.e., attacking existing nodes) • Based on the current state of the simulation, it generates a sequence of events that is fed into the DES for each time step

  17. Attack Emulation (cont.) • Attack Repository • Organize test cases of attacks • Accommodate future scenarios • Stored cases can have parameters for changes in attack behavior (i.e., timing) • Object-oriented hierarchy facilitate creation of new attacks and composition of existing attacks

  18. Attack Emulation (cont.) • Attack Handlers • Interface between the DES and the Attack Emulation System (AES) • Feeds in the current state of the system • Translates the desired events from the AES and queues them in the DES

  19. Attack Emulation (cont.) • Limits the definition of attacks to realistic behaviors • Can only control the behavior of the attacker and compromised nodes • Can automatically change any node to a compromised node • When unconcerned about the initial compromise process • Compromised node can be simulated or running emulated TinyOS code

  20. Attack Models • Supports attacks at the network layer (no jamming) • Current models: Repetition attack, Delay attack, Data alteration, Packet spoofing, Selective Forwarding, Sybil, Wormhole, DDoS, HELLO flooding • Many attacks contain common elements which can be modularized and used by multiple attacks without repetition

  21. Case Study: Security of Routing Protocols • Evaluate the security of two routing protocols used in TinyOS: • Multihop: Creates routing trees based on the hop count to the base station • MintRoute: Creates routing trees based on the link quality and only changes routes if a new route is 75% better than the current route

  22. Case Study: Security of Routing Protocols (cont.) • Setup • 7 x 7 grid • Each node can only hear its 2 – 8 neighbors • The base station is at the top left • Simulated Attacks • Selective forwarding: attacker drops some packets • Sinkhole: attacker sends good link quality • HELLO flooding: attacker pretends to the base station • Sybil: attacker pretends to be more than one node to fill up neighbors’ routing tables • Variations of Sinkhole and Hello flooding that send spoofed messages that show poor neighbor quality

  23. Case Study: Security of Routing Protocols Results • Selective forwarding poses minimal risk since nodes switch away as soon as it starts dropping too many packets • Sybil works well in both cases but requires more packets than most of the other attacks

  24. Case Study: Security of Routing Protocols Results (cont.) • HELLO flooding works relatively well in both cases but only if the attacker keeps convincing neighbors that the real base station has a poor quality • A vulnerability that is easily exploited in both protocols is the use of old sequence numbers to determine poor link quality

  25. Case Study: Security of Routing Protocols Results (cont.) • All attacks work better the closer they are to the base station because they capture larger subtrees • Protocols that don’t change links easily (MintRoute) are naturally more resistant to HELLO 1 and Sinkhole 1 attacks

  26. Case Study: Security of Routing Protocols Results (cont.) • SenSec’s benefits • Detailed evaluation of sensor networks under controlled and repeatable environments • Support for unmodified implementations of protocols before deployment • Easily identify vulnerabilities design issues and environment conditions

  27. Future Work • Evaluation of existing sensor network security protocols • Test performance under various scenarios • Identify common strength and weaknesses between protocols • Create improved security measures (IDS, encryption schemes, etc.) • Emulation of specific hardware (i.e., Mica2, tmote sky, etc.)

More Related