Sunandan Adhikary

Research Areas

Formal Methods, Control Systems, Safe Reinforcement Learning, Cyber-Physical Systems (CPS) Safety and Security

PhD Research Summary

Modern safety-critical Cyber-Physical Systems (CPS), such as autonomous vehicles and industrial microgrids, require robust designs to ensure reliable and safe operation despite hardware limitations, runtime faults, and external vulnerabilities. With increasing reliance on learning-enabled components, guaranteeing absolute safety is essential. ... This doctoral research focuses on comprehensive design-space exploration to quantify and enhance the resilience of embedded CPS through structured resilience analysis algorithms, resource-aware network and task scheduling, secure execution strategies, and safe-by-design fallback control architectures. Early work addressed the gap between control performance and optimal resource utilisation using fine-grained dynamic scheduling, where control task executions adapt based on real-time performance metrics. This ensures desired control quality and efficient resource sharing under varying hardware and network conditions. Ongoing research develops a resilience quantification method for real-time hybrid systems using reachability analysis, evaluating safe behaviour under network delays and platform noise across operating regions to optimise system parameters. Building on this, the research introduced secure scheduling techniques to mitigate schedule-based timing inference attacks on automotive Controller Area Networks (CAN). By synthesising skipped and shifted control executions, predictable timing patterns are hidden, enabling protection against side-channel attacks and supporting intrusion detection while maintaining strict real-time performance requirements. Recent work focuses on synthesising backup safe controllers using Lyapunov-based barrier functions. When the system approaches unsafe boundaries—due to attacks or unpredictable behaviour in learning-based controllers—it switches smoothly to a mathematically guaranteed fail-safe control mode, with adaptive resource reallocation based on criticality. Overall, the research establishes a rigorous foundation for structurally resilient, resource-efficient, and provably safe next-generation CPS.

Publications

1. Safe Controller Synthesis Using Lyapunov-based Barriers for Linear Hybrid Systems with Simplex Architecture
   S. Adhikary, & S. Dey, arXiv (2026)
2. Mitigating Timing-Based Attacks in Real-Time Cyber-Physical Systems
   A. Sain, S. Adhikary, & S. Dey, arXiv (2026)
3. Adaptive Parameterisation for Efficient Detection of False Data Injections
   A. Bhattacharya, S. Adhikary, I. Koley, V. Loya, & S. Dey, ACM Transactions on Embedded Computing Systems (2026)
4. Detecting Stealthy False Data Injections on Cyber-Physical Systems using Temporal Distance Metrics
   A. Bhattacharya, S. Singh, S. Adhikary, & S. Dey, ACM/IEEE 16th International Conference on Cyber-Physical Systems (ICCPS) (2025)
5. MAARS: Multi-Rate Attack-Aware Randomized Scheduling for Securing Real-time Systems
   A. Sain, S. Adhikary, I. Koley, & S. Dey, ACM/IEEE 16th International Conference on Cyber-Physical Systems (ICCPS) (2025)
6. RSU Placement Optimization for Securing Vehicle Platoon Against False Injection Attacks
   A. Roy, I. Koley, S. Adhikary, A. Sarkar, & S. Dey, ACM Transactions on Cyber-Physical Systems (2025)
7. Enhancing Attack Resilience in Real-Time Systems through Variable Control Task Sampling Rates
   A. Sain, S. Adhikary, I. Koley, & S. Dey, arXiv (2024)
8. Revisiting Dynamic Scheduling of Control Tasks: A Performance-Aware Fine-Grained Approach
   S. Adhikary, I. Koley, S. K. Ghosh, S. Ghosh, & S. Dey, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (2024)

Google Scholar