Maritime Surveillance has been identified as of top priority at EU level.
State-of-the-art maritime surveillance systems combine multi-sensor data, including satellite Synthetic Aperture Radar (SAR) images, AIS (Automatic Identification System) transmissions and feeds from radars/coastal sensors, and process/fuse them in near-real-time in order to produce a maritime situational picture as integrated as possible. Especially in large-scale deployments, compute, memory and storage resources become critical in order to improve the accuracy of the integrated picture and reduce the processing time. In this context, the aim of the maritime surveillance pilot in EVOLVE is to assess the value which the EVOLVE technologies can bring to the sector.
Using the in-house-developed ACRITAS maritime surveillance platform as a starting point, Space Hellas is adapting and re-engineering its main components in order to benefit from the EVOLVE technology propositions. This adaptation includes the transformation the initial monolithic architecture of the platform in a workflow-based containerised logic, with the ultimate aim of taking advantage of the EVOLVE accelerations, such as parallelisation, fast storage and hardware-accelerated distributed processing.
The maritime surveillance workflow includes the following blocks:
- AIS data acquisition: ingesting data from private network of receivers or from AIS data provider. AIS data are broadcast from all legitimate vessels, reporting their type, position, course, speed etc.
- SAR image acquisition: downloading SAR scenes either from satellite or from a patrolling aircraft. Satellite Aperture Radar is the most common type of imaging for detection of vessels, even at night or under low visibility conditions (clouds).
- Filtering and pre-processing: data geo-selection (region), ortho-rectification, noise reduction, data cleansing in SAR images
- Vessel detection: Detection of vessels in the SAR image, using e.g. a constant-false-alarm-rate (CFAR) detector
- Vessel classification: Classification of the vessel type, using feature extraction
- Fusion/Correlation: Correlation of AIS and SAR vessel targets, following temporal alignment.
This results in:
- the identification of SAR-detected vessels;
- the identification of the vessels who do not emit an AIS signal (non-cooperating vessels) and
- the (offline) improvement of the vessel classification algorithm.
AIS anomaly detection: this stage operates solely on AIS data and identifies anomalies, such as intermittent AIS signals, changes in the vessel identity, abrupt route/speed changes, which could:point to suspicious vessel behaviour. This step combines rule-based and Machine Learning approaches to improve accuracy. Convolutional Neural Networks (CNNs) is the primary technique used.
Visualization: Integration of maritime integrated situational picture (ISP) in a GIS-based GUI, where the vessels and all the associated metadata, along with the satellite imaging scenes, are displayed on a map.
So far, in EVOLVE, all the workflow stages have been ported and deployed in the NOVA computing platform and are being dynamically orchestrated using the EVOLVE software stack. Overall, compared to the legacy configuration, the EVOLVE technologies and infrastructure have so far resulted in a considerable (71%) reduction in the processing time. Further improvements are still to be expected, by taking more advantage of the EVOLVE acceleration features.