Problem Wensheng Zhang, Dr. Guohong Cao, and Dr. Tom La Porta Example: Battlefield Surveillance Challenges Small Sensing Range Limitations in sensor nodes Short Comm. Range Constrained Energy Supply Prone to Failure, being Compromised Requirements of Applications Mobile Data Consumer Dynamic Data Source Accurate Information Complete Information Energy Efficiency Collaboration Fault Tolerance Security Mobility Support Nodes individually detect their surrounding area, and generate low-level sensing data. In many applications, a wireless network needs to detect and track mobile targets, and disseminate the sensing data to mobile sinks. Individual sensing data are collected and fused to form more complete and concise high-level sensing data. High-level sensing data are stored in network. High-level sensing data are disseminate to sinks when being queried. Security Mechanisms Our Solution: A Data-Centric Framework Dynamic Convoy Tree-Based Collaboration (DCTC) Nodes detecting the same target form a tree. Relying on the tree, Root can collect data from other nodes, and process the data. As the target moves, the convoy tree is reconfigured by adding some nodes and pruning some other nodes. The tree is reconfigured as its root becomes faraway from the target More Research on DCTC Conservative vs. Optimistic (Prediction-based) Tree Expansion and Pruning Schemes Complete vs. Interception-based Tree Reconfiguration Schemes Sequential (Global) vs. Localized Tree Reconfiguration Schemes (when nodes can adapt their transmission power) Motivations for DCTC Why Dynamic? -- The set of nodes involved in collaborations is dynamic Why Tree? -- To efficiently facilitate data collection and fusion Basic Idea of DCTC Localized Collaboration-Based Security Mechanisms Collaborative Misbehavior Detection Individual detection: each node monitors its neighbors Discover misbehavior ? Opinion collection: initiator sends out request to other neighbors (co-detectors) of the suspect to collect their opinions on the suspect. Diagnosis and result notification: collected opinions are processed to decide whether the suspect is misbehaving, and then sends out the result. YES Data Dissemination with Adaptive Ring-Based Index (ARI) data source sink source sink Why not Previous Schemes? External storage-based: unnecessary transfers of data. (DCS has the similar problem) Directed Diffusion: flooding availability information. (TTDD has the similar problem.) Type k target Query Response (data) Detecting node Storing node register sink Type k index node node closest to Loc=hash(k) On-demand data transferring Type k target Storing node Query Response (data) Detecting node Type k index node Query Supporting sink mobility sink move! Why Index-Based Scheme? Supporting source mobility LkLk Type k index nodes are connected via forwarding nodes to form an index ring encircling the index center (Lk) The reliability of the index nodes can be improved due to adaptive replication. Why Ring? B A CD E F LkLk B’C’D’ G’ E’ F’ H’ LkLk to source Adapting the index ring is transparent to sinks and sources. Dealing with cluster failures LkLk to source <r LkLk Source B A CD E F LkLk B’C’D’ Overloaded! Load Balance TransparencyFault Tolerance Index node sink Query Response (data) target Register target target move! Ongoing Work: Heterogeneous Strategy for storing Data in Network Node-group Pair-wise Key-based Data Authentication (and False Data Filtering) keys known by a group keys known by a node key space each node-group pair shares a key a sender uses the keys it known to validate outgoing messages a receiver or an intermediate node verify messages using the keys it known polynomials are used to generate and maintain keys. Collaborative Key Updating key points: Keys of future versions are preloaded avoid updating-time key transmissions The future keys of a node are collaboratively maintained by the node and its neighbors prevent a single node from stealing future keys Random organization of keys prevent off-line analysis. sink