Case Study: Making use of satellite data and big-data analytics
Download as PPTX, PDF3 likes1,330 views
This document summarizes how Analyze Corporation uses satellite data and machine learning to identify illegal fishing activity. They analyze 500 million messages from 125,000 vessels over 5 years to detect anomalous behavior patterns compared to expected travel and working patterns. Their predictive models identify vessels likely engaged in illegal, unreported, and unregulated fishing to provide intelligence for authorities.
Case Study: Making use of satellite data and big-data analytics
- 1. Brian Waechter Strategic Associate Analyze Corporation www.analyzecorp.com Case Study: Making use of satellite data and big data analytics OR… Copyright 2015 Analyze Corporation All rights reserved.
- 2. www.analyzecorp.com Not a Fish Story: Illegal Fishing for All to See via Big Data Behavioral Analytics Copyright 2015 Analyze Corporation All rights reserved.
- 3. Question: In an increasing socially- conscious world, do you know from whence your seafood entree came? Copyright 2015 Analyze Corporation. All rights reserved. 3
- 4. The Problem: Illegal, Unreported and Unregulated (IUU) Fishing 4 Illegal fishing takes place where vessels operate in violation of the laws of a fishery. This can apply to fisheries that are under the jurisdiction of a coastal state or to high seas fisheries regulated by regional organizations. Unreported fishing is fishing that has been unreported or misreported to the relevant national authority or regional organization, in contravention of applicable laws and regulations. Unregulated fishing generally refers to fishing by vessels without nationality, or vessels flying the flag of a country not party to the regional organization governing that fishing area or species. According to a 2008 UN report, titled “The Sunken Billions: The Economic Justification for Fisheries Reform”, the world's fishing fleets are losing $50 billion USD each year through depleted stocks and poor fisheries management. Copyright 2015 Analyze Corporation. All rights reserved.
- 5. • Goal: Identify vessel behavior activity using Satellite Automatic Identification System (AIS) data • Challenge: Illegal and threatening behavior covers many different motions and activities Requires advanced and complex computer learning Involves voluminous amount of data IUU Fishing--Behavioral Analytics Match 2009 to 2013 AIS Data 125,000 Unique Vessels 500 million Messages = 3 Billion Data PointsCopyright 2015 Analyze Corporation. All rights reserved. 5
- 6. The Collaborative Team 6 Data Sharing Analysis Collaboration Insight Credit: Ken Campbell Digital Globe Technology Sensor Data Analytic Capabilities System Integration Social Networks Crowd Sourcing Copyright 2015 Analyze Corporation. All rights reserved.
- 7. Classic “big data” analytics strategy Data Acquisition – Satellite AIS Data Data Extraction - Raw AIS sensor data is dirty Data Analysis using a variety of statistical and machine learning techniques Predictive Analytics involves creating unique prediction functions Objective: Identify motion behavior through motion analytics The Process Copyright 2015 Analyze Corporation. All rights reserved. 7
- 8. Timely, Continuous Global Coverage = Timely, Continuous Global Data 8 Ship locations with shore-based AIS receivers Ship locations with space-based AIS receivers Copyright 2015 Analyze Corporation. All rights reserved.
- 9. Travel Work Travel Technical Approach Identify basic vessel pattern based on economic incentive and expected activity for the vessel type One can understand a single vessel’s behavior… but 125,000 vessels every day? Copyright 2015 Analyze Corporation. All rights reserved. 9
- 10. Automated model Geometric & pattern matching Machine learning Adjusts over time via weighted algorithms based on data Determines which vessels are “troublesome” (IUU behavior) over five-year period Provides intelligence for later interdiction and/or legal evidence Behavioral Modeling Copyright 2015 Analyze Corporation. All rights reserved. 10
- 11. Behavior Patterns Travel Travel Work Copyright 2015 Analyze Corporation. All rights reserved. 11
- 12. Satellite AIS Data Points KEY: Green = Travel Blue = Somewhat Erratic Red = Work Pink = Extremely Erratic Copyright 2015 Analyze Corporation. All rights reserved. 12
- 13. Possible Working Activity Denotes Working Activity Copyright 2015 Analyze Corporation. All rights reserved. 13
- 14. Working Activity Identified Copyright 2015 Analyze Corporation. All rights reserved. 14
- 15. The following video shows a dynamic visualization of vessels engaged in fishing activity changing over a one year period Tracks of fishing and other vessels are detected with satellite-based AIS receivers operated by SpaceQuest Vessel tracks are classified by Analyze to rate the likelihood that a ship was actively engaged in fishing activity at each point Each detection is highlighted using a visualization technology developed jointly by Google and SkyTruth https://www.youtube.com/watch?v=-TOB3X-rw3U Baseline activity is compared to what is expected and the differences are highlighted as potential threats for interdiction and/or legal action Results 15 Collaboration Insight Copyright 2015 Analyze Corporation. All rights reserved.
- 16. Summary Conclusion Vessel behaviors can be identified through motion analytics using satellite AIS data alone Lat/Long and time all that is needed (VMS/LRIT/SAR) Insights can now be pushed to operations Way Ahead Display real time results Integrate into user-specific platforms Flag potential anomalous activity long before it become a threat Other Applications Created station-keeping algorithms with Mapquest—where vehicles stop and why Profiled insider threat behavior (mouse/keyboard activity) for Verisign and Zone Fox Copyright 2015 Analyze Corporation. All rights reserved. 16
- 17. Analyze and Illegal Fishing in the News 17 http://www.wired.com/ 2014/11/plan-map- illegal-fishing-space/ http://www.wsj.com/arti cles/google-partners- target-illegal-fishing- with-new-technology- 1415927971 Copyright 2015 Analyze Corporation. All rights reserved.
- 18. Analyze® Technology Application Copyright 2015 Analyze Corporation. All rights reserved. 18 3) Text1) Geospatial 2) Sociometric PLATFORMSPRODUCTSCUSTOMERS
- 19. Analyze™ a big data analytics company software-as-a-service (products) custom advanced analytics (services) http://analyzecorp.com/for-everyone-to-see/http://www.analyzeclients.com/http://www.analyzeimpact.org/ Copyright 2015 Analyze Corporation. All rights reserved. 19
- 20. your data our science Founded February 2013 Headquarters: Fairfax, VA Technology Office: Boston, MA 20 a “big data” “analytics” company Field Office: San Antonio, TXCopyright 2015 Analyze Corporation. All rights reserved.
Editor's Notes
- #6: Every message a file Message every 6-8 seconds 24 parameters per message
- #7: With big data analytics, it’s a team sport Google—searching for use case—to put on Google Ocean Oceana—global non-profit
- #11: Model—algorithm—prism—find behavior in a range—put geo-fence around where not supposed to go
- #17: Previous analysis tried to build boolean expressions to identify typical fishing grounds, ships traffic, ports, etc.