Skip to content

System to detect abnormal vessel behavior based on historical AIS data

License

Notifications You must be signed in to change notification settings

dma-ais/AisAbnormal

Repository files navigation

AisAbnormal

Introduction

AisAbnormal is a system to work with maritime data received via the Automatic Identification System, abbreviated AIS. AIS data contain information about vessel's identities, positions, draughts, manouvering intentions, and more. An AIS data stream therefore makes it possible to maintain a detailed picture of maritime traffic.

The purpose of AisAbnormal is to detect "abnormal" vessel behaviours in real-time, so as to aid marine traffic controllers and others to determine which of many screen targets to focus their attention to. AisAbnormal can also be used in off-line mode to batch analyse historic, non real-time data.

Architecture

AisAbnormal consists of three executable modules:

  • The statistics builder (stat-builder) which is capable of building grid-based statistics for vessel movements inside a of grid of cells. The input is a stream of raw AIS data from file or network, and the output is a single binary file containing the calculated statistics.

  • The event analyzer (analyzer) which is capable of receiving raw AIS data (from replay from file or live from network) and analyzing these AIS data to detect abnormal vessel behaviours. Some analyses of vessel behaviour are based on statistical data produced by the stat-builder, while others are based on vessel behaviour alone. If an analysis gives rise to a vessel being perceived to exhibit "abnormal behaviour" than an event is recorded and stored in a database together with information about the vessel, the event type, the involved parties, the time and location, and the statistic used as a basis for the event.

  • The web front-end (webapp) which is a browser based web application which makes a user capable of working with charts (including nautical charts available via WMS), vessels, grid/cells, statistical data and detected abnormal events.

The architecture can be visualized like this:

The big picture

In the diagram the "AIS sources" can be files with recorded AIS data, or network streams with live or recorded data. The bold ellipses depict executables, which are described in further detail below.

Prerequisites

The only prerequisite for running the executables in AisAbnormal is a standard instance of the Java 8 runtime environment (JRE) which can be downloaded from Oracle's website. Until March 2014 only early access versions are available.

Download Java 8 runtime.

Dependencies

AisAbnormal is programmed in:

  • Java 8
  • Javascript

and standing on the shoulders of amongst others:

unit and functional tested using

and built with:

Building

mvn clean install

Development builds are available on CloudBees.

Releases

The source code is stored in a public repository on GitHub. The development line of code is kept on the master branch and some short-lived feature branches.

There is also a special release branch in the GitHub repo. Whenever a commit is merged or committed onto this branch it will be detected by the linked Docker Hub. Docker Hub will then perform an automated build of two new docker images; one for the analyzer (called dmadk/ais-ab-analyzer) and one for the web application (called dmadk/ais-ab-web).

The two automatically built docker images are public and can be used to launch docker containers running the analyzer and web application (see below).

Developing

AisAbnormal is IDE agnostic and can be developed using any text editor. However, it has been successfully built and run with both Eclipse and IntelliJ IDEA.

To use Eclipse the M2 plugin or the Maven eclipse target are recommended:

mvn eclipse:eclipse

Running

Executable jars will be created for the command line applications.

Statistics builder

Starting

The stat-builder is kept in the ais-ab-stat-builder folder, and can be invoked like this:

$ cd ais-ab-stat-builder/
$ java -jar target/ais-ab-stat-builder-0.1-SNAPSHOT.jar

This will produce a helpful output like this:

The following options are required: -input -output
Usage: AbnormalStatBuilderApp [options]
  Options:
    -downsampling
       Downsampling period (in secs).
       Default: 60
    -gridsize
       Grid resolution (approx. cell size in meters).
       Default: 200
  * -input
       Glob pattern for files to read. '.zip' and '.gz' files are decompressed
       automatically.
    -inputDirectory
       Directory to scan for files to read
       Default: .
    -mt
       Run multi-threaded.
       Default: false
  * -output
       Name of output file.
    -r
       Recursive directory scan
       Default: false

The -mt option is currently experimental and may be subject to be removed.

As an example: To launch the stat-builder where it reads AIS input from a compressed plain text file named aisdump_dk.txt.gz, downsamples these messages by 10 seconds, uses a cell size of 200 metres for the grid system, and stores the output in a file called test-stats.statistic - use this command line:

java -jar target/ais-ab-stat-builder-0.1-SNAPSHOT.jar -downsampling 10 -gridsize 200 -inputDirectory /data/ais -input aisdump_dk.txt.gz -output /data/stats/test-stats

Output

The stat-builder produces two files: One with the extension .statistics and one with the extension .statistics.p:

$ ls -l statistics/
 -rw-r--r--+ 1 tbsalling  staff      131072 20 Feb 15:06 test-stats.statistics
 -rw-r--r--+ 1 tbsalling  staff     6291456 20 Feb 15:06 test-stats.statistics.p

It is important to keep these two files together and keep them together as a pair if moving or copying them around in the file system.

Analyzer

Design

The analyzer is designed, so that the the data flow through main classes looks like this:

Analyzer data internal flow

Starting

The analyzer can be started like this:

$ cd ais-ab-analyzer/
$ java -jar target/ais-ab-analyzer-0.1-SNAPSHOT.jar

This will produce a helpful output like this:

The following option is required: -config 
Usage: AbnormalAnalyzerApp [options]
  Options:
  * -config
       Name of configuration file to use.

The -config parameter is mandatory, and points to a file containing the all configuration parameters for the analyzer. If a non-existing file is pointed to, then the analyzer prints a configuration file template on the console:

$ java -jar ais-ab-analyzer-0.1-SNAPSHOT.jar -config non-existing.file
2014-09-18 08:30:29,593 INFO  main             AbnormalAnalyzerAppModule        AbnormalAnalyzerAppModule created (dk.dma.ais.abnormal.analyzer.AbnormalAnalyzerAppModule@20fa23c1).
2014-09-18 08:30:29,597 ERROR main             AbnormalAnalyzerAppModule        Could not find configuration file: non-existing.file
Use this template for configuration file:
-----------------------------------------
#
# This is the configuration file for ais-ab-analyzer.
#

...
    
# Comma-separated list of MMSI numbers which will not cause any events to be raised
blacklist.mmsi = -1

...

Starting using public Docker image

The analyzer could be started in a Docker container from a publicly available Docker image, which requires no local checkout or build of the software.

Stopping existing analyzer

If the analyzer is already running in a previously started Docker image, then this must be identified and stopped like this:

$ sudo docker ps
[sudo] password for tbsalling:
CONTAINER ID        IMAGE                          COMMAND                CREATED             STATUS              PORTS                    NAMES
855a6bf9d20d        dmadk/ais-web:latest           "java -jar /ais-web.   2 days ago          Up 2 days           0.0.0.0:8082->8080/tcp   cranky_heisenberg
e06b8fe80cef        dmadk/ais-view:latest          "/bin/bash /start.sh   2 days ago          Up 2 days           0.0.0.0:8090->8090/tcp   grave_fermi
87fcc5c6ab96        dmadk/ais-ab-analyzer:latest   "/bin/bash /start.sh   5 days ago          Up 5 days                                    mad_carson
04e9d9da9685        dmadk/ais-ab-web:latest        "/bin/bash /start.sh   2 weeks ago         Up 2 weeks          0.0.0.0:8080->80/tcp     sleepy_rosalind
$

Here it is noted, that the analyzer is running in container id '87fcc5c6ab96' with image name 'dmadk/ais-ab-analyzer:latest'.

Pulling latest Docker image

Before the running container is stopped, we pull the latest Analyzer Docker image from the Docker Hub like this:

$ sudo docker pull dmadk/ais-ab-analyzer:latest
Pulling repository dmadk/ais-ab-analyzer
030207c89f49: Download complete
511136ea3c5a: Download complete
5e66087f3ffe: Download complete
...
bcc720fde3fb: Download complete
$

Stopping existing Docker image

Now we stop the running Docker container of the Analyzer by typing:

$ sudo docker kill 87fcc5c6ab96

Where '87fcc5c6ab96' is the previously noted container id.

Starting latest Docker image

The Docker newly pulled and latest version of the Analyzer is then spawned in a new Docker container this:

$ sudo docker run -e CONFIG_FILE=/data/analyzer.properties -v /home/aisstore/AisAbnormal/data:/data dmadk/ais-ab-analyzer:latest 2>&1 > ~/logs/ais-ab-analyzer.out &

The /home/aisstore/AisAbnormal/data folder is a local folder on the host machine, which contains the analyzer.properties configuration file and the statistics file to use for statistical analyses.

Output

The Analyzer, when configured to store to an H2 database, will produce an event database file like this:

$ ls -l /data/events/
-rw-r--r--+  1 tbsalling  staff     53248 21 Feb 10:34 test-events.h2.db

This is a file, that contains the events and which can later be used with the webapp.

Diagnostic output

A normal startup of the analyzer produces output in line with this:

10:32:45,352 INFO  main             AbnormalAnalyzerAppModule        AbnormalAnalyzerAppModule created (dk.dma.ais.abnormal.analyzer.AbnormalAnalyzerAppModule@1f89ab83).
10:32:45,591 INFO  main             AbnormalAnalyzerApp              java.version: 1.8.0-ea
10:32:45,591 INFO  main             AbnormalAnalyzerApp              java.specification.version: 1.8
10:32:45,591 INFO  main             AbnormalAnalyzerApp              java.vm.version: 25.0-b66
10:32:45,591 INFO  main             AbnormalAnalyzerApp              java.home: /Library/Java/JavaVirtualMachines/jdk1.8.0.jdk/Contents/Home/jre
10:32:45,591 INFO  main             AbnormalAnalyzerApp              java.vendor: Oracle Corporation
10:32:45,592 INFO  main             AbnormalAnalyzerApp              os.arch: x86_64
10:32:45,592 INFO  main             AbnormalAnalyzerApp              os.name: Mac OS X
10:32:45,592 INFO  main             AbnormalAnalyzerApp              os.version: 10.9.1
10:32:45,592 INFO  main             AbnormalAnalyzerApp              user.name: tbsalling
10:32:45,592 INFO  main             AbnormalAnalyzerApp              AbnormalAnalyzerApp created (dk.dma.ais.abnormal.analyzer.AbnormalAnalyzerApp@4524411f).
10:32:45,602 INFO  main             AbnormalAnalyzerAppModule        Created AisReader (Thread[Thread-1,5,main]).
10:32:45,604 INFO  main             AppStatisticsServiceImpl         AppStatisticsServiceImpl created (dk.dma.ais.abnormal.analyzer.AppStatisticsServiceImpl@78186a70).
10:32:45,606 WARN  main             StatisticDataRepositoryMapDB       File IndreDKtilStat-200-10.statistics already exists!
10:32:45,674 INFO  main             AbnormalAnalyzerAppModule        Opened statistic set database with filename '../../../data/statistics/IndreDKtilStat-200-10.statistics' for read.
10:32:45,694 INFO  main             AbnormalAnalyzerAppModule        Statistic data repository is valid.
10:32:45,696 INFO  main             AbnormalAnalyzerAppModule        Created Grid with size 200.0 meters.
10:32:45,697 INFO  main             TrackingServiceImpl              TrackingServiceImpl created (dk.dma.ais.abnormal.tracker.TrackingServiceImpl@351d0846).
10:32:45,780 INFO  main             AbnormalAnalyzerAppModule        Created ReplayDownSampleFilter with down sampling period of 10 secs.
10:32:45,780 INFO  main             PacketHandler                    PacketHandlerImpl created (dk.dma.ais.abnormal.analyzer.PacketHandlerImpl@2f7c2f4f).
10:32:45,907 INFO  main             Version                          HCANN000001: Hibernate Commons Annotations {4.0.2.Final}
10:32:45,911 INFO  main             Version                          HHH000412: Hibernate Core {4.2.8.Final}
10:32:45,913 INFO  main             Environment                      HHH000206: hibernate.properties not found
...

While running, the analyzer will periodically emit statistical output about its state of operation:

10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== Abnormal analyzer statistics ====
10:32:47,098 INFO  Thread-1         TrackingServiceImpl              Now processing stream at time Tue Nov 12 12:21:41 CET 2013
10:32:49,051 INFO  Thread-1         SuddenSpeedChangeAnalysis        Tue Nov 12 12:34:09 CET 2013: Detected SuddenSpeedChangeEvent for mmsi
10:31:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Pos message count                      0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Stat message count                     0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Track count                            0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Message rate                           0 msg/sec
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== Application statistics ====
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== Abnormal analyzer statistics ====
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== Abnormal analyzer statistics ====
10:32:47,098 INFO  Thread-1         TrackingServiceImpl              Now processing stream at time Tue Nov 12 12:21:41 CET 2013
10:32:49,051 INFO  Thread-1         SuddenSpeedChangeAnalysis        Tue Nov 12 12:34:09 CET 2013: Detected SuddenSpeedChangeEvent for mmsi
10:32:46,968 INFO  main             AppStatisticsServiceImpl         Statistics service started.
10:32:46,968 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== AIS Abnormal Behavior application statistics ====
10:32:46,968 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== Application statistics ====
10:32:46,968 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Unfiltered packet count                0
10:32:46,968 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Filtered packet count                  0
10:32:46,968 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Message count                          0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Messages out of sequence               0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Pos message count                      0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Stat message count                     0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Track count                            0
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         Message rate                           0 msg/sec
10:32:46,969 INFO  pool-1-thread-1  AppStatisticsServiceImpl         ==== Application statistics ====

And also, periodically it will output lines like this:

10:32:47,098 INFO  Thread-1         TrackingServiceImpl              Now processing stream at time Tue Nov 12 12:21:41 CET 2013

Which tells the timestamp of the last processed AIS message. This is particularly useful when working with AIS datasets from files, as it helps to estimate the time of completion.

Choice of database system

Using an H2 database has the advantage, that no separate RDBMS system need to be running. The software for the H2 database is embedded into the analyzer, so it can readily store in H2 format to a single file. The disadvantage is, that only one process at a time can access the H2 event database file. This prohibits running the analyzer and the webapp at the same time.

To run the analyzer and the webapp simultaneously requires a more advanced RDBMS than H2, and for this purpose the analyzer can connect to and use a separately installed, configured, and running Postgres database accessible via the designated IP, port, database name, user name and password.

Web application

Starting

The web application contains an embedded web server, and it does therefore not need to be deployed in a traditional Java middleware server. Instead, the web application can be started like this:

$ cd ais-ab-web/
$ java -jar target/ais-ab-web-0.1-SNAPSHOT.jar

This will produce a helpful output like this:

The following options are required: -statistics -eventDataRepositoryType
Usage: ais-ab-web [options]
  Options:
    -eventDataDbFile
       Name of RDBMS file to hold event data.
    -eventDataDbHost
       Name of the event database RDBMS host.
    -eventDataDbName
       Database name to use for the event database with the RDBMS host.
    -eventDataDbPassword
       Password to connect to the event database RDBMS host.
    -eventDataDbPort
       Remote port of the event database RDBMS host.
    -eventDataDbUsername
       Username to connect to the event database RDBMS host.
  * -eventDataRepositoryType
       Type of repository used to hold event data ('h2', 'pgsql').
  * -statistics
       Filename of statistic data file to read.
    -port
       Port no. for listening to HTTP.
       Default: 8080

To launch the web application where it reads statistical data from the test-stats files, reads events from the test-events H2 file - and exposes the web application on port 8080, issue the following example command line:

     java -jar target/ais-ab-web-0.1-SNAPSHOT.jar -statistics data/statistics/test-stats -eventDataRepositoryType h2 -eventDataDbFile data/events/test-events -port 8080

Once the web application has launched it can be accessed from a browser, e.g. locally, using this URL:

     http://localhost:8080/abnormal/

Once the web application server part is launched and a browser pointed to it, the expected initial browser screen looks like this:

Web application startup screenshot

Note! Until further, it is necessary to launch the application with the current directory being ais-ab-web (so that static assets assumed by the program to be in src/main/webapp are available).

Starting using public Docker image

The web application image can be run in a Docker container from a publicly available Docker image, which requires no local checkout or build of the software. The Docker container is spawned like this:

sudo docker run -p 127.0.0.1:8080:80 -e DB_HOST=<ip> -e DB_PORT=8432 -e DB_USER=xxx -e DB_PASS=xxx -e DB_INST=xxx -e STAT_FILE=/data/stats-2014-dwn5-grid200 -v /home/ais/AisAbnormal/data:/data dmadk/ais-ab-web:latest

Here port 80 in the container is mapped to port 8080 on the host machine. There is no DNS client configured in the docker image, so DB_HOST must point to an IP address running a Postgres server. Remaining DB_* parameters are for connection to Postgres. STAT_FILE points to the statistics data file to use, and /home/ais/AisAbnormal/data is a local folder on the host machine which is mapped to /data in the docker container. This folder should contain the statistics data file.

Modules

##ais-ab-common##

This module contains all common resources for the project. Examples are the Tracker and the Application Statistics which are used by both the stat-builder and the analyzer.

##ais-ab-stat-builder##

Command line application building/updating a statistics database using historical AIS data from a file. See detailed description above.

##ais-ab-stat-db##

This module hols the interface and implementations of a statistics database API. This is used by both the stat-builder, the analyzer, and the web application.

Currently the only provided implementation of the statistics database is based on the MapDB storage engine.

##ais-ab-analyzer##

This module holds a command line tool to analyze AIS data for abnormal events. See detailed description above.

##ais-ab-event-db##

Interface and implementation of an event database API, which is used by both the analyzer and the web application.

Currently two implementations of the event-db interface are provided:

  • H2
  • Postgres

##ais-ab-web## This module holds the web application for presenting statistical data and abnormal events. See detailed description above.

About

System to detect abnormal vessel behavior based on historical AIS data

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •