Responsive security system protects internet of things devices
To offer joined-up, real-time services, smart devices use internet of things (IoT) technologies, making them more complex and more reliant on active interfaces across multiple platforms. But collecting and analysing data across interoperable platforms increases security vulnerabilities. The SecureIoT project, guided by reference models such as the Industrial Internet Security Framework and the OpenFog Reference Architecture, has developed more responsive IoT systems security. “We enhanced well-established security building blocks with additional modules to more proactively detect vulnerabilities and boost the preparedness of cybersecurity teams,” says Ioannis Soldatos, one of the coordination team members at Intrasoft International, the project host. The EU-supported project’s platform for end-to-end data-driven IoT security, includes support for data collection via modular and configurable probes, developed for both open-source platforms like FIWARE and commercial platforms like SIEMENS MindSphere. The platform also integrates security algorithms to automate risk mitigation.
The ‘specify-build-test-learn-improve’ approach
After analysing IoT security scenarios to identify vulnerabilities and threats, the team defined the knowledge and functions needed by the platform. The build phase resulted in prototypes which were tested with the help of security experts. SecureIoT developed an assessment process which can identify, grade and, in some cases, mitigate IoT security lapses. The team also devised a compliance auditing service to ensure devices meet security standards, and a service to help IoT programmers incorporate the latest security functions. Their security knowledge base stores security information from external sources, such as the Common Vulnerability and Exposures, Common Attack Pattern Enumeration and Classification and Common Platform Enumeration. Platform modules, such as the predictive analytics algorithms, were tested individually. These include process mining techniques that analyse event sequences to detect abnormal behaviour, such as traffic anomalies. They also apply deep learning techniques, such as variational autoencoders, to identify attack patterns. The whole system was field-tested in controlled conditions for different sectors. This included: Industry 4.0 scenarios in a pilot plant with automation platforms and cyber-physical production systems; connected car interfaces carried out in test-driving environments; and humanoid robots used to support children with autism. Various attacks – such as malicious software updates, vehicle network attacks and denial-of-service attacks – were simulated to trigger the response of SecureIoT services. “The rapid detection of cyberattacks, along with accurate predictions, reduced overall exposure to risk. For example, with our Industry 4.0 testing, cyberattack detection was 70 % accurate, with secure access control 100 % effective in reducing the spread of cyber infections from untrusted IoT systems,” explains Spyros Evangelatos, a coordination team member.
Securing next-generation IoT applications
In the era of IoT, the impact of security breaches could be financially and socially severe. “As well as mitigating risks and boosting preparedness, SecureIoT could also increase the trust of users, key for widespread adoption,” adds Soldatos. The team are currently further validating their technology in real-life environments before commercialisation, with plans to include on-premises deployment for larger enterprises, as well as a cloud-based security-as-a-service option for SMEs. “Our smart tools go beyond state of the art. As IoT systems don’t typically collect data for training security algorithms, we didn’t develop models based on standard open-source data sets, but rather used our own data collected by smart objects in realistic environments,” concludes Evangelatos.
Keywords
SecureIoT, cybersecurity, cyberattack, data, algorithms, internet of things, IoT, Industry 4.0, robots, risk, mitigation
