adobestock_557389202-resized

IPMSRL: Autonomous cyber defence for maritime operational technology

A simulation environment for training multi-agent reinforcement learning (MARL) agents to defend maritime industrial control systems from cyber threats.

IPMSRL
Contact us

Transforming maritime cyber defence with AI

This BMT innovation is the result of a strong, multi-disciplinary collaboration between BMT’s AI, Data Science, and Cyber Security teams. By combining deep domain expertise with cutting-edge technology, the teams have developed a powerful capability that enhances maritime cyber resilience through intelligent simulation and autonomous response strategies.

IPMSRL is a configurable simulation platform designed to model cyber-attacks and defences within maritime Operational Technology (OT) systems, such as an Integrated Platform Management System (IPMS). It enables the training of autonomous agents using Multi-Agent Reinforcement Learning (MARL) to respond to and recover from cyber incidents. The environment reflects real-world OT challenges, including partial observability, legacy infrastructure, and complex network topologies.

While the current focus is on maritime applications, the technology is highly adaptable to other key domains — including OT-based sectors such as offshore wind, energy, and ports, as well as IT-related environments. This flexibility supports broader innovation and cross-sector resilience.

From simulation to resilience: IPMSRL in practice

Exploring cyber defence response strategies

IPMSRL offers a streamlined simulation environment for testing how AI agents can respond to cyber threats in maritime Operational Technology systems. It supports early-stage prototyping of autonomous defence strategies in realistic OT scenarios.

Attack scenario simulation

Simulates complex cyber-attacks using the MITRE ATT&CK® ICS framework, with future alignment to models such as the Cyber Kill Chain and NIST guidelines for incident response and OT security.

Autonomous agent development

Researchers can develop and refine multi-agent reinforcement learning models for coordinated cyber incident response.

Reinforcement learning research

IPMSRL provides a controlled environment for experimenting with MARL algorithms such as MAPPO and IPPO in OT contexts.

1  / 
buttonarrow
Show more

Features and benefits

simulation

Realistic maritime OT simulation

IPMSRL replicates the architecture of Integrated Platform Management Systems (IPMS), including critical and non-critical nodes, to reflect real-world vulnerabilities in shipboard control systems.

security

Collaborative multi-agent defence

Supports training of multiple reinforcement learning agents that coordinate actions to contain, eradicate, and recover from cyber-attacks, improving response effectiveness.

threat_intelligence

Advanced threat modelling

Integrates the MITRE ATT&CK® ICS framework and aligns with industry standards such as NIST SP 800-61 (Incident Response) and NIST SP 800-82 (OT/ICS Security) to simulate sophisticated attacker behaviours and lateral movement across OT networks.

notifications

Partial observability and alert dynamics

Models realistic alert systems with configurable success probabilities, enabling agents to learn in environments with incomplete or delayed information.

award_star

Customisable learning and reward systems

Offers flexible reward shaping and hyperparameter tuning to optimise agent performance and encourage efficient, strategic decision-making.

arrows_outward

Scalable and adaptable architecture

Designed to support varied network topologies, attack scenarios, and operational priorities, making it suitable for research, training, and future real-world deployment.

adobestock_199788327-resized

IPMSRL has shown promising results in enabling autonomous cyber defence for complex OT systems. The ability to simulate realistic attack scenarios and train agents to respond effectively is a game-changer for maritime cyber security.

Dstl Cyber Defence Enhancement Programme

Meet the expert behind the tech

Alec Wilson - BMT

Alec Wilson

Senior AI/ML Research Specialist

Alec is a Senior AI/ML Research Specialist, specialising in Machine Learning (ML) with a focus on Reinforcement Learning. Alec is experienced in technically leading R&D projects funded under Dstl’s ARCD Programme exploring autonomous cyber defence strategies in the maritime OT domain.

Connect
Alec Wilson - BMT

Alec Wilson

Senior AI/ML Research Specialist

Alec is a Senior AI/ML Research Specialist, specialising in Machine Learning (ML) with a focus on Reinforcement Learning. Alec is experienced in technically leading R&D projects funded under Dstl’s ARCD Programme exploring autonomous cyber defence strategies in the maritime OT domain.

Contact us