Edge-Enabled Tactical Systems Cached Results (User graph, blacklists, precomputed) Fact Databases (DBPedia, Freebase) Rumor Corpus (Snopes.com) Filter Stream Credibility Score Calculate Score Event and Noise Classifiers (event-specific spam) Structural Content Extract Features Temporal Data Stream User Query Receiver operating characteristic (ROC) curve ROC curve (area = 0.97) 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 User connects mobile device to the cloudlet, and upon visual confirmation the admin starts the pairing process Mobile Device connects to router, validates server credentials, and authenticates with RADIUS server Communication between the mobile device and the cloudlet is encrypted at the transport and message level The implementation pipeline for credibility calculation. Word cloud represents input data from a shooting event Word cloud represents input data from a non-shooting event Fusion Architecture ISE implementation Cooperative Context Sharing Receiver operating characteristic (ROC) curve for shooting event detection C&C view of Edge Analytics The Edge Analytics prototype Admin logs into the Cloudlet Manager to start the Bootstrapping process Step 1: Bootstrapping • Generation of Server Credentials using IBE (Identity-Based Encryption) • Setup of RADIUS Server with Server Credentials Step 2: Pairing • Generation of Device Credentials using IBE • Transfer to device using Bluetooth or USB, plus visual confirmation • Transfer to RADIUS Server Step 3: WiFi Authentication RADIUS Server implements Wi-Fi WPA2-Enterprise 802.1X EAP-TTLS with PAP • Device receives server credentials and validates • Devices sends its credentials for validation Step 4: API Requests • Device exchanges encrypted messages with the server • Each exchange is validated against authorized device list Termination • Automatic due to timeout: Bootstrapping requires setting up mission length • Manual due to known loss or compromise: Server Management component has revocation option ROUTER Edge environments are characterized by dynamic context, limited computing resources, high levels of stress, and poor network connectivity. Edge-Enabled Tactical Systems (EETS) adapts, extends, and innovatively investigates architectures and technologies that provide efficient and easily deployable mobile solutions for teams operating in edge environments. Goal for FY15: Efficient and trusted integration between the edge and the enterprise Trusted Nodes: Establishing Trusted Identities in Disconnected Tactical Environments Method and prototype to establish trust between mobile devices and cloudlets in disconnected tactical environments Confidence in Information: Fusion of Social and Physical Sensor Data Fusion of local sensor information, gathered cooperatively and opportunistically, with streaming social media and Open Source Intelligence (OSINT) to inform strategic support and improve tactical response. Possible Relationships • Tweet + location (actual or inferred) cues GCF sensors • Trending topic + similar mission keywords cues ISE sensor (events) • ISE sensor/event + location cues GCF sensors Scenarios (objective) • Geo-tagged tweet triggers GCF sensors for collection • Trending keyword matches with ISE event description • Use ISE to task GCF for additional sensor data • ISE event triggers GCF sensor collection Confidence in Information: Assigning Credibility Scores to Social Media Streams in Real-Time Prototype and algorithm to determine the reliability of information derived from social media. 1 2 3 4 Individual Context Unit Context Enterprise Context Consolidated Context Model TOC ENT File System File System File System Play HTTP Server JVM Browser EA Analysis Engine Twitter EA REST Proxy ElasticSearch ElasticSearch Native connector File IO Websocket connector Mongo Connector Redis Protocol EA System boundary HTTP//REST Redis Redis External System (NICS) Browser (Chrome) Legend Components Connectors Twitter Twitter File System MongoDB Data Store Edge Analytics Front End (Play App) Command Center Front End (Play App) Redis Database Integrated Analysis Twitter Publisher and Controller GCF Publisher and Controller GCF ISE Manager ISE Node Twitter MQTT Server GCF Device Edge Analytics Fusion GCF Integration ISE EA Batch Jobs Back End System Software Component Request Data Legend Legend Sensor Management View User Interface Layer Application Layer I/O Layer Context Data View Application Manager Context Engine Data Manager Battery Bluetooth GPS Sensor Manager Set of Android Sensors Mediated Broadcast TCP/ IP Communications Manager Task View User View Alert View Sensor Config Layer Logical Component Data Source Synchronous Call-Return Asynchronous Callback Data Read/Write AIDL Interface File App- Specific Rule Set Config Config Config Default Rule Set App Data SEI Research Review 2015 Contact: Grace Lewisand Jeff Boleng [email protected], [email protected] Distribution Statement A: Approved for Public Release; Distribution is Unlimited