Ultra-Lightweight Key Predistribution in Wireless Sensor Networks for Monitoring Linear Infrastructure

Transcription

1 Ultra-Lightweight Key Predistribution in Wireless Sensor Networks for Monitoring Linear Infrastructure Keith M. Martin Maura B. Paterson Information Security Group Royal Holloway, University of London 4 September 2009 Maura Paterson KPSs for linear WSNs 1/ 16

2 Outline Applications Requiring One-Dimensional WSNs Maura Paterson KPSs for linear WSNs 2/ 16

3 Traditional View of a WSN Maura Paterson KPSs for linear WSNs 3/ 16

4 Traditional View of a WSN Maura Paterson KPSs for linear WSNs 3/ 16

5 Traditional View of a WSN Maura Paterson KPSs for linear WSNs 3/ 16

6 Oil/Gas/Water Pipelines are Critical Infrastructure photo credit: Chris Sauerwald, Maura Paterson KPSs for linear WSNs 4/ 16

7 Pipeline Monitoring Pipelines are used to transport valuable resources over long distances through isolated areas. Maura Paterson KPSs for linear WSNs 5/ 16

8 Pipeline Monitoring Pipelines are used to transport valuable resources over long distances through isolated areas. Detection of leaks, measurement of seismic activity, detection of theft or sabotage... Maura Paterson KPSs for linear WSNs 5/ 16

9 Pipeline Monitoring Pipelines are used to transport valuable resources over long distances through isolated areas. Detection of leaks, measurement of seismic activity, detection of theft or sabotage... Also bridges, railway lines, roads... Maura Paterson KPSs for linear WSNs 5/ 16

10 Pipeline Monitoring Pipelines are used to transport valuable resources over long distances through isolated areas. Detection of leaks, measurement of seismic activity, detection of theft or sabotage... Also bridges, railway lines, roads... Perimeter surveillance requires sensors in a ring. Maura Paterson KPSs for linear WSNs 5/ 16

11 WSNs are Well-Suited to Monitoring Linear Infrastructure ease of deployment ease of maintenance increased reliability Maura Paterson KPSs for linear WSNs 6/ 16

12 WSNs are Well-Suited to Monitoring Linear Infrastructure ease of deployment ease of maintenance increased reliability Maura Paterson KPSs for linear WSNs 6/ 16

13 Characteristics of One-Dimensional WSNs We assume the network consists of an arbitrary number of nodes, each with communication range r. Restricted number of neighbours (proportional to r, rather than r 2 ) Location knowledge (In particular, the order of nodes along the pipe is known.) Density of node deployment (required to ensure adequate sensing coverage) Maura Paterson KPSs for linear WSNs 7/ 16

14 Key Predistribution {k 1, k 5, k 7 } {k 3, k 5, k 12 } k 5 key predistribution scheme (KPS) nodes are assigned keys before deployment nodes that share keys can communicate securely e.g. Eschenauer and Gligor: Each node draws m keys uniformly (without replacement) from a keypool K Maura Paterson KPSs for linear WSNs 8/ 16

15 Security Metrics for KPSs for Traditional WSNs Number of keys stored by each node. Connectivity: Pr 1 :=probability that two neighbouring nodes share a key Resilience: fail(s):=number of links between uncaptured nodes that are compromised when an adversary captures s nodes Maura Paterson KPSs for linear WSNs 9/ 16

16 Major Security Threat for One-Dimensional WSNs: Disconnection of the Network We need to prevent large sections of the network from becoming disconnected. Small numbers of isolated nodes are not a problem. Maura Paterson KPSs for linear WSNs 10/ 16

17 New Security Metric for One-Dimensional WSNs Definition A KPS for a one-dimensional network consisting of a set N of n nodes, where n is arbitrary, is s-fallible if Maura Paterson KPSs for linear WSNs 11/ 16

18 New Security Metric for One-Dimensional WSNs Definition A KPS for a one-dimensional network consisting of a set N of n nodes, where n is arbitrary, is s-fallible if 1. After the capture of any s 1 nodes, there exists a set E N of size at most O(1) such that N \ E is connected. Maura Paterson KPSs for linear WSNs 11/ 16

19 New Security Metric for One-Dimensional WSNs Definition A KPS for a one-dimensional network consisting of a set N of n nodes, where n is arbitrary, is s-fallible if 1. After the capture of any s 1 nodes, there exists a set E N of size at most O(1) such that N \ E is connected. 2. It is possible to choose s nodes whose capture partitions the network into two (or more) isolated networks of size Ω(n). Maura Paterson KPSs for linear WSNs 11/ 16

20 Upper Bounds For s-fallibility Theorem If a KPS for a one-dimensional WSN in which the nodes have communication range r yields a connected network, then it is s-fallible for some 1 s r. Theorem Suppose a KPS that yields a connected network assigns keys to nodes such that the largest distance between two nodes that share a key is b. Then it is s-fallible for some 1 s b. Maura Paterson KPSs for linear WSNs 12/ 16

21 Basic Construction We assign keys to the nodes such that: all pairs of nodes at distance r share a key; all pairs of nodes at distance 1 share a key. e.g. r = 3 Maura Paterson KPSs for linear WSNs 13/ 16

22 The Basic Construction is r-fallible (sketch...) Label nodes sequentially 0, 1,..., n. Suppose r 1 nodes are captured. There exists x Z such that no node with a label equivalent to x (mod r) has been captured. Let Ψ 1, Ψ 2 be nodes at distance at least r + 1 from some captured node. We can find a secure path between them: Maura Paterson KPSs for linear WSNs 14/ 16

23 The Basic Construction is r-fallible (sketch...) Label nodes sequentially 0, 1,..., n. Suppose r 1 nodes are captured. There exists x Z such that no node with a label equivalent to x (mod r) has been captured. Let Ψ 1, Ψ 2 be nodes at distance at least r + 1 from some captured node. We can find a secure path between them: 1. Take hops of length 1 from Ψ 1 to Ψ 2 until a node with label equivalent to x (mod r) is reached. Maura Paterson KPSs for linear WSNs 14/ 16

24 The Basic Construction is r-fallible (sketch...) Label nodes sequentially 0, 1,..., n. Suppose r 1 nodes are captured. There exists x Z such that no node with a label equivalent to x (mod r) has been captured. Let Ψ 1, Ψ 2 be nodes at distance at least r + 1 from some captured node. We can find a secure path between them: 1. Take hops of length 1 from Ψ 1 to Ψ 2 until a node with label equivalent to x (mod r) is reached. 2. Take hops of length r towards Ψ 2 until a node at distance less than r from Ψ 2 is reached. Maura Paterson KPSs for linear WSNs 14/ 16

25 The Basic Construction is r-fallible (sketch...) Label nodes sequentially 0, 1,..., n. Suppose r 1 nodes are captured. There exists x Z such that no node with a label equivalent to x (mod r) has been captured. Let Ψ 1, Ψ 2 be nodes at distance at least r + 1 from some captured node. We can find a secure path between them: 1. Take hops of length 1 from Ψ 1 to Ψ 2 until a node with label equivalent to x (mod r) is reached. 2. Take hops of length r towards Ψ 2 until a node at distance less than r from Ψ 2 is reached. 3. Complete the path by hops of length 1. Maura Paterson KPSs for linear WSNs 14/ 16

26 Construction for Ring Topologies Maura Paterson KPSs for linear WSNs 15/ 16

27 thankyou Maura Paterson KPSs for linear WSNs 16/ 16