Security in Infrastructure Wireless
Sensor Networks
Emil	
  Lupu	
  
Imperial	
  College	
  London	
  
Security	
  in	
  WSN	
  is	
  difficult	
  because…	
  
•  Nodes	
  are	
  physically	
  
accessible	
  
•  Scale	
  is	
  unprecedended	
  
•  Resources	
  are	
  limited	
  
•  Opera?on	
  is	
  una@ended	
  
There	
  is	
  a	
  large	
  disconnect	
  between	
  
research	
  and	
  applica?on	
  
•  A	
  large	
  body	
  of	
  work	
  on	
  security	
  in	
  sensor	
  
networks.	
  	
  
–  …	
  a	
  cursory	
  search	
  on	
  Google	
  scholar	
  gives	
  34,400	
  
hits	
  since	
  2009	
  

•  Yet	
  security	
  remains	
  largely	
  ignored	
  in	
  
reported	
  deployments	
  	
  
Research	
  is	
  oPen	
  based	
  on	
  	
  
assump?ons	
  such	
  as	
  …	
  	
  
•  Random	
  choice	
  of	
  a@ack	
  targets	
  	
  
•  Nodes	
  are	
  placed…:	
  in	
  a	
  grid,	
  uniformly	
  and	
  
densely,	
  in	
  a	
  tree	
  with	
  a	
  single	
  sink,	
  in	
  pre-­‐
decided	
  groups.	
  	
  
•  Sensors	
  can	
  do…:	
  RSA	
  crypto,	
  puzzles,	
  Java,	
  
homomorphic	
  crypto,	
  game	
  theory….	
  as	
  well	
  
as	
  their	
  tasks.	
  	
  
•  All	
  nodes	
  are	
  coopera?ve	
  and	
  trustworthy,	
  
faults	
  always	
  occur	
  randomly.	
  	
  
Three	
  Guiding	
  Principles	
  
•  Link	
  studies	
  to	
  context	
  of	
  use	
  
•  Leverage	
  rela?onship	
  with	
  physical	
  
phenomena.	
  	
  
•  Security	
  as	
  part	
  of	
  resilient	
  design	
  
Link	
  studies	
  to	
  context	
  of	
  use	
  
• 
• 
• 
• 
• 
• 

Threat	
  Model	
  
Physical	
  Security	
  
Sensed	
  informa?on	
  
Topology	
  
Node	
  capabili?es	
  
Security	
  objec?ves	
  
Leverage	
  rela?onship	
  to	
  physical	
  
phenomena	
  
•  Key	
  genera?on	
  and	
  
distribu?on.	
  
•  Distance	
  bounding.	
  
•  Observability	
  of	
  
associa?on.	
  
•  A@esta?on.	
  
•  Verifica?on	
  of	
  correct	
  
behaviour	
  
Security	
  as	
  part	
  of	
  resilient	
  design	
  
•  Is	
  it	
  a	
  fault	
  or	
  a	
  
compromise?	
  
•  Redundancy	
  drives	
  
anomaly	
  detec?on,	
  
trust,	
  recovery	
  
•  Adapta?on	
  vs	
  security	
  
•  Resource	
  limita?ons	
  
Research	
  Themes	
  
•  Designing	
  solu?ons	
  in	
  concrete	
  se^ngs.	
  	
  
•  Establishing	
  confidence	
  in	
  the	
  correct	
  and	
  
secure	
  opera?on	
  of	
  devices.	
  	
  
•  Measuring	
  the	
  “health”	
  of	
  the	
  network	
  from	
  
within	
  itself.	
  	
  
•  Data	
  quality	
  and	
  trustworthiness	
  assurance.	
  
	
  
Designing	
  solu?ons	
  in	
  concrete	
  
se^ngs	
  
•  Key	
  management	
  for	
  the	
  
context	
  of	
  use	
  
•  Evaluate	
  and	
  improve	
  
exis?ng	
  algorithms	
  e.g.,	
  
applicability,	
  scaleability	
  
•  Mi?ga?on	
  and	
  recovery	
  
through	
  reconfigura?on	
  
Establishing	
  confidence	
  in	
  the	
  correct	
  
and	
  secure	
  opera?on	
  of	
  devices	
  	
  
	
  
•  Measuring	
  device	
  integrity.	
  
•  A@esta?on	
  techniques	
  
–  SoPware,	
  hardware,	
  physical	
  

•  Behavioural	
  fingerprin?ng	
  and	
  
valida?on.	
  
•  Reputa?on.	
  
Measuring	
  the	
  health	
  of	
  the	
  network	
  
from	
  within	
  itself	
  
•  For	
  a	
  sensor,	
  a	
  network,	
  a	
  city.	
  	
  
•  Assurances	
  and	
  Mechanisms	
  
•  Capability	
  of	
  restart/recovery/
response/degraded	
  opera?on	
  
•  Coordina?on/Topology/
Redundancy	
  
•  Effect	
  of	
  adapta?on	
  on	
  
security	
  and	
  resilience	
  
•  Learning	
  to	
  dis?nguish	
  
between	
  normal	
  behaviour,	
  
failures	
  and	
  security.	
  
•  Recognising	
  security	
  events	
  
in	
  constantly	
  evolving	
  
networks.	
  
•  A@ack	
  characterisa?on	
  
•  Compromise	
  epidemiology	
  
Data	
  Quality	
  and	
  Trustworthiness	
  
Assurance	
  

•  Data	
  protec?on	
  and	
  provenance	
  
•  In-­‐network	
  Data	
  Aggrega?on	
  
•  How	
  to	
  accept	
  data	
  degrada?on	
  to	
  maintain	
  
overall	
  system	
  goals.	
  	
  
•  Understand	
  and	
  mi?gate	
  against	
  data	
  
deteriora?on	
  effects	
  e.g.	
  node	
  compromise	
  
failures,	
  fouling	
  
•  Data	
  usage	
  control	
  
Polygonia	
  Interroga?onis	
  

WSN Security Research Directions

  • 1.
    Security in InfrastructureWireless Sensor Networks Emil  Lupu   Imperial  College  London  
  • 2.
    Security  in  WSN  is  difficult  because…   •  Nodes  are  physically   accessible   •  Scale  is  unprecedended   •  Resources  are  limited   •  Opera?on  is  una@ended  
  • 3.
    There  is  a  large  disconnect  between   research  and  applica?on   •  A  large  body  of  work  on  security  in  sensor   networks.     –  …  a  cursory  search  on  Google  scholar  gives  34,400   hits  since  2009   •  Yet  security  remains  largely  ignored  in   reported  deployments    
  • 4.
    Research  is  oPen  based  on     assump?ons  such  as  …     •  Random  choice  of  a@ack  targets     •  Nodes  are  placed…:  in  a  grid,  uniformly  and   densely,  in  a  tree  with  a  single  sink,  in  pre-­‐ decided  groups.     •  Sensors  can  do…:  RSA  crypto,  puzzles,  Java,   homomorphic  crypto,  game  theory….  as  well   as  their  tasks.     •  All  nodes  are  coopera?ve  and  trustworthy,   faults  always  occur  randomly.    
  • 5.
    Three  Guiding  Principles   •  Link  studies  to  context  of  use   •  Leverage  rela?onship  with  physical   phenomena.     •  Security  as  part  of  resilient  design  
  • 6.
    Link  studies  to  context  of  use   •  •  •  •  •  •  Threat  Model   Physical  Security   Sensed  informa?on   Topology   Node  capabili?es   Security  objec?ves  
  • 7.
    Leverage  rela?onship  to  physical   phenomena   •  Key  genera?on  and   distribu?on.   •  Distance  bounding.   •  Observability  of   associa?on.   •  A@esta?on.   •  Verifica?on  of  correct   behaviour  
  • 8.
    Security  as  part  of  resilient  design   •  Is  it  a  fault  or  a   compromise?   •  Redundancy  drives   anomaly  detec?on,   trust,  recovery   •  Adapta?on  vs  security   •  Resource  limita?ons  
  • 9.
    Research  Themes   • Designing  solu?ons  in  concrete  se^ngs.     •  Establishing  confidence  in  the  correct  and   secure  opera?on  of  devices.     •  Measuring  the  “health”  of  the  network  from   within  itself.     •  Data  quality  and  trustworthiness  assurance.    
  • 10.
    Designing  solu?ons  in  concrete   se^ngs   •  Key  management  for  the   context  of  use   •  Evaluate  and  improve   exis?ng  algorithms  e.g.,   applicability,  scaleability   •  Mi?ga?on  and  recovery   through  reconfigura?on  
  • 11.
    Establishing  confidence  in  the  correct   and  secure  opera?on  of  devices       •  Measuring  device  integrity.   •  A@esta?on  techniques   –  SoPware,  hardware,  physical   •  Behavioural  fingerprin?ng  and   valida?on.   •  Reputa?on.  
  • 12.
    Measuring  the  health  of  the  network   from  within  itself   •  For  a  sensor,  a  network,  a  city.     •  Assurances  and  Mechanisms   •  Capability  of  restart/recovery/ response/degraded  opera?on   •  Coordina?on/Topology/ Redundancy   •  Effect  of  adapta?on  on   security  and  resilience  
  • 13.
    •  Learning  to  dis?nguish   between  normal  behaviour,   failures  and  security.   •  Recognising  security  events   in  constantly  evolving   networks.   •  A@ack  characterisa?on   •  Compromise  epidemiology  
  • 14.
    Data  Quality  and  Trustworthiness   Assurance   •  Data  protec?on  and  provenance   •  In-­‐network  Data  Aggrega?on   •  How  to  accept  data  degrada?on  to  maintain   overall  system  goals.     •  Understand  and  mi?gate  against  data   deteriora?on  effects  e.g.  node  compromise   failures,  fouling   •  Data  usage  control  
  • 15.