Security in Wireless Sensor Networks

1
Security in Wireless Sensor Networks
Professor Jack Stankovic Department of Computer
Science University of Virginia October 25, 2004
2
Security

• Complex, many aspects to consider
• General, complete solution is unlikely
• Opportunity to address this properly from the
  start!
• Targeted solutions for targeted attacks
• Reasonably secure WSN

3
Ad Hoc Wireless Sensor Networks

• Sensors
• Actuators
• CPUs/Memory
• Radio
• Minimal capacity
• 1000s

Self-organize
4
Mica2 and Mica2Dot

• ATMega 128L 8-bit, 8MHz, 4KB EEPROM, 4KB RAM,
  128KB flash
• Chipcon CC100 multichannel radio (Manchester
  encoding, FSK). 50 ft and up.

5
Sensor Board
6
Sensor Board
7
Applications
8
Outline

• Motivating Application
• Overview of the Security Problems in WSN
• Routing
• SPEED
• RAP
• IGF
• Denial of Service
• Jamming
• Potential Future Approaches
• Secure Group Management
• Exploit Physical Properties
• Summary

9
Energy Efficient Surveillance System
1. An unmanned plane (UAV) deploys motes
Zzz...
Sentry
2. Motes establish a sensor network with power
management
3. Sensor network detects vehicles and wakes up
the sensor nodes
10
General Security Issues

• New (severe) constraints (memory, bandwidth, cpu
  processing speeds, power, )
• Lightweight solutions required
• Symmetric cryptography (asymmetric crypto is too
  expensive)
• Physical Environment
• Faults versus attacks
• Cheap to attack

11
Specific Security Problems

• Routing and/or Backbone Disruption
• Denial of service
• Jam
• Prevent wake-up
• Prevent sleep (dies soon)
• Modify group management information

12
Specific Security Problems

• System Initialization (re-sync messages and
  centralized base stations)
• Clock Sync
• Neighbor Discovery
• Localization
• Etc.

13
Communication Scenarios

• Confidentiality (eavesdrop)

Node2
Base Station
Msg
Node1
Adversary
14
Communication Scenarios

• Integrity

Base Station
Msg1
Msg1
Node1
Adversary
15
Communication Scenarios

• Authenticity

I am the Base Station
Node 1
Base Station
Node 2
Adversary
Node 3
Reprogram system Reset system parameters
Node 4
16
Attack HELLO floods

• Hello packets to announce presence of a node
• Assumption the sender of a received packet is
  within normal radio range
• False! A powerful transmitter could reach the
  entire network
• Disrupts routing paths

17
Summary- Basic Problems

• Vulnerability of channels (eavesdrop and inject
  fake messages)
• Vulnerability of nodes (capture, modify messages,
  re-route)
• Absence of infrastructure (e.g., no centralized
  certification authorities)
• Dynamically changing topology (difficult to
  distinguish between dynamics and attacks)
• Minimum capacity devices
• Drain batteries
• Real-Time slow packets down

18
Routing - Network Assumptions

• Attacker has similar capabilities (HW)
• Nodes can be turned
• Tamper resistant nodes are not realistic

Many routing protocols have been proposed for
WSN, but (almost) none with security as a goal !
Examples GF, AODV, DSR, DD, SPEED, RAP, IGF,
19
Route Where

• Each node to base station
• Nodes to aggregation points and then from
  aggregation point to base station
• Between 2 (n) nodes (peer to peer)
• Between 2 (n) areas
• Among all members of a (dynamic) group

20
SPEED
USE VELOCITY
21
SPEED

• SPEED A Protocol for Real-Time Communication in
  Sensor Networks. Uses local neighbor tables

Strong Back-Pressure (Congestion)
Uniform Back-Pressure
22
SPEED
7
11
Packet
Destination
5
9
Packet
2
Delay
3
10
Source
Boo
23
Attacks

• Change neighbor information
• Change delay
• Change velocity set point
• Change last mile processing
• Slow down packets
• Inject false packets

24
Solutions for SPEED

• Authenticate neighbors as true neighbors and on
  routing table data updates (e.g., delays)
• Timestamp to prevent replay attack
• Confidentiality - Encrypt last mile information
• Encrypt data
• Global set point setup and any changes must be
  authenticated (or not permitted)
• Overhear to determine if attacker is acting like
  a sinkhole (black hole)

25
RAP Prioritized Packets
D
dis 90 m D 2 s V 45 m/s HIGH Priority
E
A
C
B
dis 60 m D 2 s V 30 m/s LOW Priority
26
RAP

• RAP A Real-Time Communication Architecture for
  Large-Scale Wireless Sensor Networks.

Respecting Deadlines and Priorities
Packets with Different Velocities
27
Attack

• Modify priorities
• Short deadline
• Long distance
• Inject packets with high priorities (denial of
  service)
• Etc.

28
Solutions for RAP

• Encrypt velocity field
• Authenticate packets and drop those from an
  attacker

29
Solutions

• SPINS
• uTESLA
• Provides authenticated streaming broadcast
• SNEP
• Provides data confidentiality, two-party data
  authentication and data freshness
• Routing protocol using the above building blocks

30
Solutions

• Mobile Ad hoc Wireless
• SEAD
• Ariadne
• SRP
• Etc.
• WSN
• TinySec link level encryption
• LiSP
• Etc.

31
Geographic Forwarding (GF)

• GF always chooses a node that is closest to the
  destination.
• Every node knows its location.

s
d
32
Implicit Geographic Forwarding

• Tackle the rapid dynamics found in WSNs
• To deal with
• Power Down Nodes (Sleep mode)
• Node Mobility
• Node Failure
• Scale
• Lazy Binding (to the nth degree)
• State Free no routing tables

33
IGF
Asleep
Moving Away

• IGF is a combined Routing/MAC protocol
• Eligible nodes - 60 degree cone (shift cone if
  necessary)
• RTS - set timer based on distance and energy
  remaining

34
IGF Implicit Security Properties

• No routing tables maintained
• Routes cannot be corrupted in this manner
• Impact of intruder limited to neighborhood
• No dissemination of route information

35
IGF Security Problems

• Intruder is unchecked in neighborhood
• Pretend it is in various places (Sybil attack)
• Answer multiple times
• Always answer first (will die more quickly)
• Jam cant do much (will die quickly)
• Eavesdrop (assume encryption)

36
Solution Approaches

• Prevent sender from choosing the adversary for
  the next hop
• Reduce probability of selecting the intruder

37
Solutions

• Soln 1 Authenticate after the CTS
• Soln 2 Overhear if attacker does not transmit
  or changes the packet choose another node
• Soln 3 Choose nth or random responder, not the
  first

38
Solutions

• Soln 4 Use Power level to detect Sybil attack
• Attacker could adjust power levels but could not
  be sure that they would reach the sender
• Soln 5 Select n nodes to forward message to
  (assume at most 1 attacker)

39
Solutions

• Soln 6 Verify location
• Angle of arrival
• Directional antenna (to send RTS and receive CTS)
• Use overhearing all nodes in cone should
  overhear each other and if intruder is outside
  cone some nodes will not hear

40
Solutions

• Soln 7 Omit destination altogether
• Intruder has to guess the right direction or send
  6 messages
• Intruder could infer proper forwarding sector
  based on history of message streams

41
Summary - Secure IGF

• Attack Model
• Insert or subvert a normal mote
• Inherently Good Properties
• No routing tables at all
• Contains attacks to limited area
• Attacks Possible
• Greedy attacker sends CTS immediately
• If it gets packet drop alter
• Sybil attack

42
Denial of Service
Ref Denial of Service in Sensor Networks Wood
Stankovic
43
The Jamming Problem

• Jamming disrupts communication around the source

44
Solution Overview - Mapping
45
Value - A Mapping Service

• Map jammed-area and export to other modules
• Value
• Report jammed area to base station
• Send in vehicle to find/destroy jammer
• Route around jammed area
• Lower duty-cycle to save energy
• Redirect any queries to services in the jammed
  area
• Expose area as programmer-accessible entity

46
Jam Detection

• Highly decentralized algorithm
• Loose group semantics, eager eavesdropping, uses
  local information, robustness to packet loss and
  failure, works with partial mappings
• Performance (example)
• When neighbor density is moderate, converges to a
  single mapping group in 1.5 5 seconds
• Function of size of jammed area
• Robust to failure rates of 20 25

47
Mapping Algorithm Overview

• Jamming Detection
• Group Formation
• Receive JAMMED message
• Receive BUILD message
• Coalescing groups
• Bridging groups

48
Algorithm Jamming Detection
Attempting to send a message

• if (wireless channel is busy for longer than
  250ms)
• if (near epicenter of jamming)
• sleep for awhile
• else
• send blind JAMMED msg to neighbors ltID
  locationgt

// Based on signal strength
49
Example Jamming Detection

• J1-3 and others are jammed by adversary
• Jamming is detected using heuristics
• Jammed nodes blindly report their IDs and
  locations

J3
M3
J2
JAMMED
M2
J1
M1
50
Algorithm Group Formation
Received JAMMED message from Ji

• for all local groups Gk
• if (Ji is compatible with Gk)
• add Ji to group Gk
• join Gk if not already a member
• if (no compatible groups found)
• create and join new group
• send BUILD msg after announce timer
• ltGk jammed nodes subsumed group IDsgt

// Find compatible group // or create new one
51
Example Group Formation

• Neighbors M1-3 receive JAMMED messages
• Neighbors create groups G1-3 and store direction
  vectors
• Mappers set announce timers
• M2 sends a BUILD message first, containing ltG2
  J2gt

J3
M3
ltG3 J3gt
J2
BUILD
M2
J1
ltG2 J2gt
M1
ltG1 J1gt
52
Algorithm Group Formation
Received BUILD for Gj (1/2)

• if (no local group Gj)
• create group Gj from msg
• else
• save group information in message
• wait

// Create Gj or save contents of message
53
Example Group Formation

• Neighbors store G2 information
• M1 also announces G1 to neighbors

J3
M3
ltG3 J3gt
J2
BUILD
ltG2 J2gt
M2
J1
ltG2 J2gt
BUILD
ltG2 J2gt
ltG1 J1gt
M1
ltG1 J1gt
54
Algorithm Group Formation
Received BUILD for Gj (2/2)

• if (Gj is compatible with another local group
  Gk)
• if (this node is a member of Gj or Gk)
• wait short delay and coalesce(Gj, Gk)
• else if (have heard a PROBE msg)
• wait long delay and coalesce(Gj, Gk)
• if (this node is on the edge of Gj)
• schedule PROBE msg to be sent
• if (this node is member of Gj AND msg not seen
  before)
• relay msg using current local state of Gj

// Compare neighbors // with direction vector
// Relay the message
55
Algorithm Coalescing
coalesce ( Gj, Gk)

• if (Gj, Gk still compatible and active)
• choose dominant group Gd group with highest
  ID
• merge jammed nodes into Gd
• merge subsumed group IDs into Gd
• if (not a member of Gi or Gk)
• join Gd as bridge member
• send BUILD msg ltGd merged jammed list
  merged group listgt

56
Example Coalescing
J3
ltG3 J3,J2 G2gt

• M3 compares direction vectors of G2 and G3
• M3 starts coalesce timer since they are
  compatible
• M3 sends a BUILD message containing the dominant
  group
• ltG3 J3,J2 G2gt
• M2 receives the BUILD and merges G2 into G3
• M2 is a member of G2, and so relays the message
• Neighbors also merge G2 into G3

M3
ltG3 J3gt
J2
BUILD
ltG2 J2gt
M2
J1
ltG2 J2gt
ltG3 J3,J2 G2gt
ltG2 J2gt
ltG3 J3,J2 G2gt
ltG1 J1gt
M1
ltG1 J1gt
57
Example Probing and Bridging
J3
ltG3 J3,J2 G2gt

• M1 eventually sends a PROBE, since it is on the
  edge of its group G1
• Neighbor receives the PROBE and sets a long
  coalesce timer for G1, G3
• B1 coalesces G1 and G3 when the timer expires,
  joining as a bridge node
• B1 sends a BUILD message containing the dominant
  group
• ltG3 J3,J2,J1 G2,G1gt

M3
J2
M2
J1
BUILD
ltG3 J3,J2 G2gt
B1
PROBE
ltG3 J3,J2 G2gt
ltG3 J3,J2,J1 G2,G1gt
ltG1 J1gt
M1
ltG1 J1gt
58
DoS Regions Mapped
59
Future Solutions/Directions

• Relaxed Group Semantics
• Exploit Physical properties

60
Group Management
IR Camera
Leader
Follower
Member
Node
61
Group Management
IR Camera
Leader
Follower
Member
Node
62
Examples Tracking andMap Regions
Base Station
63
Group Semantics - Operations
64
Group Management

• History
• Many group management and group communication
  protocols
• Internet community
• Infrequent membership changes
• Strong atomicity
• Ordering semantics
• Careful membership control
• Protocols operate under various fault models

65
Example Consensus

• Classical consensus all correct processes agree
  on one value
• No power constraints
• No real-time constraints
• Does not scale well to dense networks
• Approximate agreement (some work here) - on sets
  of values (physical quantities)
• New Solutions ?

66
New Concept of Consensus
Classical
New Definitions

• Termination every correct processor eventually
  decides some value
• Uniform Agreement no two processors decide
  differently
• Group Membership join/leave - everyone knows who
  is in the group

• Termination at least n correct processors
  decide some value by time t
• Group Agreement at least n processors decide the
  same value within epsilon
• Area/Function Membership join/leave an area or
  by function

67
Solutions - CLIQUES

• Assume knowledge of membership
• Uses a group controller to manage member
  additions/removals
• Each node does expensive computation and sends to
  others (multi-round)
• Each member provides a share of secret
• Final group key

68
New Solutions Required

• Without knowledge of membership
• Inexpensive computation
• One (or few) round(s)
• Can proceed with partial collection of shared
  keys
• Groups tied to geography ?

69
Exploit Physical Properties

• Physics
• Momentum, velocity, temperature readings, energy,
  continuity,
• Location
• You cant be where you say you are
• Time (and time validity of data)
• Protect for short periods of time
• Differentiated security
• Redundancy / High Density
• Cross check

70
Exploit Physical Properties

• Strength of Signal
• Too powerful looks like jammer or hello attack
• Overhearing
• Detect black holes, sink holes, changed messages
• Directional Antenna
• For localization
• Neighbor discovery (in correct direction)

71
Summary - Security

• New security models needed
• Efficient key distribution schemes required
  (static and dynamic)
• Solve data integrity, routing, secure groups,
  denial of service
• Can new solutions exploit physical properties?
• Provide multiple layers of defense
• Securing every message, data item etc is probably
  not possible however, the aggregate performance
  of the system needs to be secure

72
Two Perspectives

• WSN used for security
• WSN subject of attacks

73
Wireless Sensor Networks Unlimited
potential The next Internet (in terms of
impact)
If we can solve the Security and Privacy Problems