TL;DR
This systematic review examines the current state of cybersecurity research for water systems, highlighting a growing interest in securing these critical cyber-physical systems. The paper identifies key research trends, common vulnerabilities, and the need for more comprehensive security frameworks and industry collaboration to protect water infrastructure from increasingly sophisticated cyber threats.
What Problem Does This Paper Address?
Water systems, like other critical infrastructure, are rapidly evolving from isolated, bespoke systems to interconnected cyber-physical systems (CPS) that leverage IoT, edge computing, and AI. While this modernization enhances efficiency and control, it also significantly expands the attack surface for malicious actors. These actors can aim for industrial espionage, sabotage, or disruption of essential services, with potentially catastrophic consequences for public health, the environment, and the economy. This paper addresses the need to understand the current landscape of cybersecurity research specifically within the water sector to identify existing efforts, gaps, and future research directions.
Methods
The authors conducted a systematic literature review following the PRISMA guidelines. They defined six research questions (RQ1-RQ6) to guide their analysis, focusing on publication trends, geographic distribution, author affiliations, publication venues, covered security aspects, and classification of security aspects. They searched multiple academic databases (Springer Link, IEEE Xplore, ACM, Science Direct, ASCE Library) and Google Scholar using specific keywords related to water and cybersecurity. Inclusion and exclusion criteria were applied to filter relevant peer-reviewed publications from 2000 to 2020. Data was extracted and analyzed to answer the research questions.
Datasets
The paper discusses several testbeds and datasets used in the reviewed research. Notably:
- SWaT (Secure Water Treatment) testbed: A six-stage water treatment process testbed with real cyber and physical equipment.
- WADI (Water Distribution) testbed: An extension of SWaT, designed to simulate water treatment, storage, and distribution.
- BATADAL dataset: Used for detecting cyber-physical attacks on water distribution networks.
The paper also mentions that many studies rely on simulated data or data generated from these testbeds, and that the availability of these resources has spurred research, particularly in detection mechanisms.
Key Insights and Ideas
- Growing Interest, but Still Nascent: While research interest in water system cybersecurity has significantly increased since 2016, it still lags behind other critical infrastructure sectors like electricity.
- Focus on Drinking Water: The vast majority of research is concentrated on drinking water distribution and treatment systems, with less attention paid to wastewater, stormwater, or irrigation systems.
- Testbeds are Crucial: The development and availability of testbeds like SWaT and WADI have been instrumental in advancing research, enabling realistic simulations and evaluations of security measures.
- Detection is King (for now): The most common research focus is on developing detection mechanisms for cyber-attacks, likely driven by the availability of datasets and the need to identify threats.
- Industry-Academia Gap: A significant gap exists in collaborative research between industry and academia, hindering the practical applicability and deployment of security solutions.
- Need for Holistic Frameworks: Current research often focuses on specific aspects of security. There's a clear need for comprehensive security frameworks that integrate prevention, detection, response, and mitigation strategies.
Key Results
- Publication Trends: Research has seen a significant surge, with most publications appearing after 2015.
- Geographic Distribution: Singapore, the USA, and Israel are leading in research output, often linked to projects funded by the EC.
- Affiliation: The overwhelming majority of research (85.1%) originates from academia, with a small percentage from private organizations and research institutes. Notably, no papers were co-authored with water companies.
- Publication Venues: Conferences (54.7%) are the primary venue for publishing, followed by journals (31.2%) and workshops (14.1%).
- Security Aspects Covered: Detection mechanisms are the most studied aspect (31 papers), followed by attack models (10 papers) and simulation/testbeds (9 papers).
Limitations and Future Work
- Testbed Limitations: While valuable, existing testbeds primarily focus on drinking water systems. More diverse testbeds for wastewater, stormwater, and irrigation are needed. Building and maintaining these are resource-intensive.
- Attack Model Sophistication: Current attack models often focus on single-point attacks. More research is needed on stealthy, multi-point attacks that can evade detection.
- Detection Model Comparability: A lack of standardized performance metrics and diverse datasets makes it difficult to compare the effectiveness and generalizability of different detection models. Deep learning shows promise but is vulnerable to adversarial attacks.
- Real-World Applicability: Many detection models are evaluated using simulated data or limited real-world scenarios. More research is needed on the usability and maintainability of proposed detection models in operational environments.
- Root Cause Analysis and Response: Current research often stops at detection. Future work should focus on identifying the root cause of anomalies, locating compromised devices, and developing effective response and mitigation strategies.
- Industry Collaboration: There's a critical need for increased collaboration between academia and industry to ensure research is applicable and addresses real-world challenges.
- Comprehensive Frameworks: Developing a holistic security framework that integrates all aspects of security (prevention, detection, response, mitigation) is essential.
Summary updated