Security in Wireless Sensor Networks Perrig, Stankovic, Wagner Jason Buckingham CSCI 7143: Secure Sensor Networks August 31, 2004 • Introduction to sensor networks • Security Issues – – – – – – – – – – Key establishment & setup Secrecy & Authentication Privacy Robustness to DOS Secure Routing Resilience to node capture Secure Group Management Intrusion Detection Secure Data Aggregation Secure Sensor Network Research Broad Range of Applications • • • • • • Wildlife Monitoring Machinery Performance Building Safety Military Applications Health Monitoring Countless other applications Most applications require some level of security! Traditional security cannot be applied directly • Sensor device limitations – Power, Processing, and Communication • Nodes are often physically accessible, allowing possible physical attacks • Sensor networks interact with their environments and with people Security Research Issues • These new problems present an opportunity to properly address network security • Security is NOT a standalone component – it must be integrated into every component Key Establishment & Setup • Why not use existing protocols? – Public key cryptography has too much system overhead for sensor networks – Key establishment techniques must scale well to hundreds or thousands of nodes – Sensor nodes have different communication needs Key Establishment - Potential Solutions • Establish a network wide shared key – Problem: if one node is compromised, the whole network is compromised • Instead use shared symmetric keys between each pair of nodes that are preconfigured – Problem: It doesn’t scale well! • For an n node network, each node must store n-1 keys, and n * (n – 1) / 2 total keys are needed. • Combine the above: use a network wide key to establish link keys, then erase the networkwide key. – Problem: New nodes cannot be added after initial deployment Key Establishment - Potential Solutions (cont.) • Bootstrapping Keys – each node shares a single key with the base station and the base station sets up keys between pairs. – Problem: requires a trusted base station that is the central point of failure • Random Key Predistribution – – choose a large pool of symmetric keys, and give each node a random subset of the pool – not all nodes share a common key, but the network will still be fully connected if the probability of two nodes sharing a key is sufficiently high – Problem: once compromising a sufficient number of nodes, attackers could reconstruct the entire pool and break the scheme Secrecy & Authentication – Cryptography • End-to-end cryptography – Provides high level of security, but requires that keys be set up among all end points – Incompatible with passive participation and local broadcast • Link-layer cryptography – Simplifies key setup – Supports passive participation and local broadcast – Problem? Intermediate nodes can eavesdrop and alter messages. Is this really a problem? Cryptography Issues • Performance Costs – Extra computation • Could be reduced by additional hardware but this increases node cost and will it really fix the problem? – Increases packet size – Recent research shows that most of the performance overhead is attributable to increased packet size, not additional computation • This limits how much dedicated cryptographic hardware will help Robustness to Denial of Service • Adversaries can simply broadcast a highenergy signal or violate the 802.11 MAC protocol to disrupt communication • Solutions? – Spread spectrum communication, but cryptographically secure spread spectrum radios are not commercially available – Automated defense, by simply routing around the jammed portion of the network Secure Routing • Current routing protocols suffer security vulnerabilities – DOS attacks, packet injection, replay attacks Resilience to Node Capture • Sensor networks are highly susceptible – the compromise of a single node usually compromises the entire network • This is more of a problem because sensor networks often lack physical security Solutions to Node Capture • Physical solution – tamper resistant packaging • Software: – Create algorithms that use majority voting schemes; send packets along multiple independent paths and check for consistency – Gather redundant data and analyze for consistency Secure Group Management • Groups of nodes perform data aggregation and analysis (tracking a moving object) • The group may change continuously and quickly • Protocols are needed for admitting new group members and supporting secure communication with the group • Solutions conserve time and energy Intrusion Detection • Classic intrusion detection is very expensive in terms of memory, energy, and bandwidth • To develop a solution, typical threat models must be analyzed • Secure groups may be a possible solution for decentralized intrusion detection Secure Data Aggregation • Tons of data can be collected from the various nodes in the network • How do we aggregate the data so as to reduce network traffic to the base station? • Aggregation locations must be secured Privacy • How do we prevent sensor networks from being used to violate privacy? – Devices are becoming smaller, cheaper, and more effective at surveillance • Solutions? – New laws, technological responses, and awareness Other Issues • What cryptographic algorithms are best suited for use in sensor networks? – – – – Public key cryptography? Too expensive! DES/Triple DES AES RC5 • We need something that fits the processing and memory requirements of our nodes Secure Sensor Network Research • How can we build security into sensor networks from the outset? • Advantages of sensor networks – Many applications will be deployed under a single administrative domain – It may be possible to exploit redundancy, scale, and physical characteristics