1. Paper Review: On communication Security in Wireless Ad-Hoc Sensor Networks By Toni Farley

2. Paper Introduction [Sli] S. Slijepcevic, M. Potkonjak,V. Tsiatsis, S. Zimbeck, M.B. Srivastava. On communication Security in Wireless Ad-Hoc Sensor Networks, Eleventh IEEE International Workshops on Enabling Technologies: Infrastructure for Collaborative Enterprises (WETICE'02) June 10 - 12, 2002 Pittsburgh, Pennsylvania, USA

3. Research Contributions ©Assess communication security threats in SN ©Separate security based on sensitivity level of data -> efficient resource management ©Location-based scheme protects rest of network when parts are compromised

4. Approach ©Goal: Minimize security related energy consumption ©Principle: Data items must be protected to a degree consistent with their value ©3 types of network data and associated threat  Mobile code 4Malicious code can change network behavior in unpredictable ways  Locations of sensor nodes 4Acquiring this information may make physical location determination easier than w/ radio location  Application specific data 4Specific to security requirements of application (assumed low threat in this study)

5. WSN Architecture ©SensorWare (UCLA & Rockwell Science Center)  Localized Algorithms: Nodes only exchange messages within immediate neighborhood. One node aggregates data and sends to gateway node (proxy between user and network).  Local Broadcast: Used for communication.  Code Mobility: Supports mobile code. 4Do not need to keep all applications on node at all times 4Applications needed may not be known at deployment 4For reconfiguration after deployment

6. Security Threats ©Security requirements of threats in model  High: Insertion of malicious code to destroy or gain control of the network  Medium: Interception of location messages: attacker gains knowledge of node’s physical location  Low: Interception of application specific messages (confidentiality) ©Inject false messages  Incorrect information to user  Sleep deprivation torture

7. Communication Security Scheme ©Private key cryptography w/ group keys  Content of all network messages encrypted  Group keys (as opposed to pairwise) work with current network architecture (uses multicasting) ©Access to security API through SensorWare ©Assumes all nodes can access message content ©Various algorithms vs single algorithm w/ adjustable parameters  Single algorithm chosen as it takes less space

8. Keys ©All nodes share initial set of master keys  Set size based on expected network life span 4Known cyphertext attack  Dynamic key establishment won’t work 4Nodes do not keep track of neighbors 4Cannot guarantee all nodes get new key  One key active at any time 4Index of current key in master list is selected by pseudorandom generator w/same seed on each node 4Periodic, synchronous key changing

9. Encryption & Security Levels ©RC6 encryption algorithm  Adjustable parameter: # of rounds  More rounds = more security = more overhead ©For each data type: corresponding security mechanism  Level I (mobile code)  Level II (location information messages)  Level III (application specific messages)

10. Security Levels ©Keys for 3 levels derived from master key ©To access network, a user needs  Set of master keys  Pseudorandom number generator  Seed ©Level I  Mobile code messages are much less frequent  Can use stronger encryption w/ more overhead

11. Security Levels ©Level II  Locations of sensor likely to be in every message 4Means high overhead, esp. w/strong encryption 4Uses medium strength encryption (?)  Use location based keys for encryption 4Common keys within extended “cells” 4Isolates parts of the network 4Nodes must know exact location 4Uniform cell shape simplifies which cell a node is in 4Hexagonal cells ensure max 3 keys

12. Security Levels

13. ©Level III  Lots of application specific messages  Weakest encryption strength  Lower computational overhead  MD5 hash of master key

14. Implementation ©Using RC6 on Rockwell WINS sensor nodes ©Confirmed observation on # of rounds increase in computational overhead ©Rounds for levels I & III (% decreased overhead)  Level I: 32  Level III: 22 (23%) ©Same consumed energy as scheme with one encryption level  More Level II & III messages  Shift in security focus

15. Implementation

16. Issues ©Gaining control of one node grants access to all 3 things needed for accessing the network:  Set of master keys  Pseudorandom number generator  Seed ©Node synchronization must be exact to change master keys ©Knowledge of exact location needed for Level II is not practical if nodes are mobile