Network Science and National Security Challenges and Opportunities
Total Page:16
File Type:pdf, Size:1020Kb
Network Science and National Security Challenges and Opportunities Robert A. Bond GraphEx 2018 DISTRIBUTION STATEMENT A. Approved for public release. Distribution is © 2018 Massachusetts Institute of Technology. unlimited. Delivered to the U.S. Government with Unlimited Rights, as defined in 25 April 2018 This material is based upon work supported by the Assistant Secretary of DFARS Part 252.227-7013 or 7014 (Feb 2014). Notwithstanding any Defense for Research and Engineering under Air Force Contract No. FA8702- copyright notice, U.S. Government rights in this work are defined by 15-D-0001. Any opinions, findings, conclusions or recommendations DFARS 252.227-7013 or DFARS 252.227-7014 as detailed above. Use expressed in this material are those of the author(s) and do not necessarily of this work other than as specifically authorized by the U.S. reflect the views of the Assistant Secretary of Defense for Research and Government may violate any copyrights that exist in this work. Engineering. Outline • Introduction • National Security Network Science Applications • Case Study: Command (or Intel) Center of the Future • Summary GraphEx National Security Keynote- 2 RAB 04/25/2018 National Security Challenges US national security challenged by diverse and distributed threats Russia • Motivated by lost Rogue Nations greatness, perceived • North Korea, Iran Homeland Security NATO threats, etc. • Growingly empowered by • Terrorism • Revanchism potential nuclear & missile programs • Law enforcement to become revisionism? China Global Terrorism • Massive power shift from North Atlantic to Pacific • Many nations, environments, Belt of Unrest • Assertiveness now, aggressiveness motivations, etc. • Large, young, frustrated population in 20 years? • Desire to better control destiny & restore Muslim greatness Network Science and Graph Algorithms have crucial roles to play in mitigating these threats GraphEx National Security Keynote- 3 RAB 04/25/2018 Why in the World Graphs? • Graphs provide a natural and expressive representation for networks & relational data • Often, it is precisely the relationships between data that matter (e.g. influence ops) • Often, graphs can be the most robust representations in the face of adversarial actions (e.g. for cyber attacks) We make sense of the world by combining inter-dependent data-sets Dense data (maps, images) + Sparse data (networks, relationships) GraphEx National Security Keynote- 4 RAB 04/25/2018 National Security Involves Networks Everywhere Example: Networks in a Bioterrorist Attack Early Indications and Warnings Response to an Attack 7 4 2 Bio Coordination across 17 networks 5 Terrorist Attack 6 3 1 Query Results Example from XLab Program 7 Travel events GraphEx National Security Keynote- 5 RAB 04/25/2018 Network Science and Graph Algorithms in National Security – Some Examples National Security Area Network Science Roles What’s at Stake Homeland Security Cyber security Defending against major cyber attack against U.S. CT operations Defending against terrorist attack against U.S. WMD disaster response/forensics Countering the spread of disease (pandemics) National infrastructure ops Resilience in air traffic control, power grid, internet, etc. Russian Resurgence Counter influence operations Protecting democratic values at home & abroad Chinese Emergence Strategic and tactical intelligence Maintaining international political influence Conventional warfare C4ISR Preserving global deterrence Enhancing global political stability Improving warfighting effectiveness Preparedness and options in international crises Global Terrorism Sentiment analysis Increased global stability Belt of Unrest CT/CI operations (I&W) Maintaining international political influence CT/CI operations (forensics) Reducing danger to civilians abroad and allies Rogue Nations Indications and Warnings (I&W) Maintaining / providing options Missile defense Preserving global deterrence Preventing nuclear attack / WMD on the homeland GraphEx National Security Keynote- 6 RAB 04/25/2018 CT/CI : counter terrorism / counter insurgency WMD: weapons of mass destruction Outline • Introduction • National Security Network Science Applications – Cyber Security – Influence Operations – Brain Research & Bioinformatics • Case Study: Command (or Intel) Center of the Future • Summary GraphEx National Security Keynote- 7 RAB 04/25/2018 A Few Specific National Security Network Science Applications Cyber Security (Reconfigurable Networks) Influence Operations (Macron Leaks Retweets) Intel Analysis (Chinese Social Network) Bioinformatics (Disease Network) Brain Science (Mouse Neocortex) Tactical Operations (Threat Networks) • Social Networks • Ecological Networks • Electrical Power Grid • Pandemic, Sexual, etc. Networks • ATC Networks • The Internet • Circuit Networks • Metabolic Networks • Financial Networks • Communications Networks • Military Logistics Networks • Publishing Networks GraphEx National Security Keynote- 8 RAB 04/25/2018 Cyber Adversarial SCenario modeling and Automated Decision Engine (CASCADE) Network defense through dynamic reconfiguration Measure Risk Combined CASCADE Iterations Prototype Capabilities Technical Challenges • Dynamically quantifies risk in the face of an • Co-optimization of mission and network at huge adaptive cyber adversary network scales in real-time • Considers mission context to select optimal • Adversary awareness, incomplete information network segmentation • Attributed, time-varying, multigraph topology GraphEx National Security Keynote- 9 RAB 04/25/2018 Influence Operations and Grey Zone Warfare 1 Countering Information Operations • Detection of information operations • Assessment of intent and impact • Assessment and execution of Counteractions Adversary IO Tactics Exploit Cyber Social Media • Information manipulation; bias, slant, distortion, omission, and amplification • Hacking accounts and leaking information Technical Challenges • Data Collection & Network Construction − Data observability, volume, speed & variety − Building the relevant networks (fusion) • Network Sampling and IO Detection − Subsampling for statistical analysis at scale − Detection techniques Nation State adversaries using the internet and social • IO Characterization and Response media to wage sophisticated influence operations − Creating the IO narrative and assessing impact − Modeling and simulation of CoA 1excerpted from: Smith et al., Influence estimation on social media networks using causal inference, in Proc. IEEE SSP, to appear GraphEx National Security Keynote- 10 RAB 04/25/2018 Bioinformatics for National Security A few Connections between Graphs and Bioinformatics2 • Biological relationships can be captured as graphs, • pathway data (e.g., metabolic pathways, signaling pathways, gene regulatory networks), genetic maps and structured taxonomies. • Even laboratory processes can be represented as workflow process model… • Graphs and can be used to support formal representation for use in laboratory information management systems • Graphical databases can be used to link the large number of enormous biological related databases A few National Security Graph Applications for Bioinformatics and Synthetic Biology • Rapid development of countermeasures against novel bio and chem threats • Detect and defeat engineered pathogens and biologics 1 Human disease network (such as proteins containing non-natural amino acids) • Read/write DNA for novel targeting/tracking/locating 1from: http://erj.ersjournals.com/content/44/3/775.figures-only • Custom protein design and production for therapeutics, 2 excerpted and modified from: https://www.ncbi.nlm.nih.gov/books/NBK25464/ accelerated vaccine development, decontamination GraphEx National Security Keynote- 11 RAB 04/25/2018 Brain Science: Anatomy of the Brain Precursor to Understanding Operation High Resolution Example Graph Processing Neuron Dataset Results Challenges Z • Graph Formation: X Ø Dense clutter of intertwined axon fibers Ø High accuracy is required to trace long distance neural connections Ø Large volume (entire mouse brain ~1 PB) • Graph Processing Ø Brain structural organization; information flow and localization Y Ø Future experiments to investigate relationships between physiology and MIT Chung Lab and Lincoln Laboratory developing structure techniques for specimen preparation and automated long Ø Efficient algorithms at huge range neuron tracing scales (~100B neurons, >100T synapses) GraphEx National Security Keynote- 12 RAB 04/25/2018 Summary of Graph Application Drivers • Very large graphs – Information Operations à millions of edges, thousands of vertices – Brain Science à trillions of synapses, billions of neurons • Large graph analysis in real-time – Cyber security à ~seconds (at the speed of the effect propagation) – Tactical operations à seconds to hours (inside the opponent’s OODA loop) • Graphs of all types, combined – Information Operations à multiple sources of data, hidden components – Bioinformatics à attributed graphs, uncertainty and noise • Algorithms over dynamic graphs – Tactical and strategic operations à temporal emergence of threat, noise, uncertainty – Cyber security à reasoning across the time-frame of attack, active deception • Complex graph processes over attributed multigraphs – Cyber security à co-optimization of graph and mission – Tactical and strategic ops à graph topology analyses, causality, threat propagation, learning, … GraphEx National Security Keynote- 13 RAB 04/25/2018 OODA: Observe, Orient, Decide, Act Outline • Introduction