2021 Fiscal Year Research-status Report
Stochastic Control-Theoretic Approach to Development of Simultaneously Cyber-Secure and Energy-Efficient Randomized Transmission Methods for Dependable IoT
| Project/Area Number |
20K14771
|
|---|
| Research Institution | National Institute of Informatics |
|---|
Principal Investigator |
CETINKAYA AHMET 国立情報学研究所, アーキテクチャ科学研究系, 特任研究員 (60851730)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | Control systems / Network attacks / Multi-agent systems / IoT |
|---|
| Outline of Annual Research Achievements |
There were four main research results, in the second term (2021) of this project.
1) A resilient quantization method was designed to change transmission data rate depending on malicious network packet losses. This method is designed to be able to use randomization to improve security and is data-efficient.
2) Stability and security of nonlinear systems with linerization-based controllers was analyzed.
3) Discrete-time multi-agent systems and attack-recovery sequences on inter-agent communication links were studied using tools of game theory.
4) A new data-driven control approach was proposed to design controllers using input-output data from a nonlinear system. In addition, conditions were provided to check when a stochastic system cannot be stabilized with constrained controllers.
|
| Current Status of Research Progress |
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
Reason
A key aspect of this research project has been to realize secure and dependable IoT systems, which face computation and communication constraints. The second term of the research project was very successful in development of new methods that can handle these constraints. The newly developed quantization and data-driven control methods as well as the proposed analysis techniques for control constraints and nonlinear systems are integral to developing dependable IoT systems.
|
| Strategy for Future Research Activity |
There are three targets to reach in the final phase of this research project,
1) Jamming attacks on systems under disturbance: It is essential to study jamming attacks on systems that face other detrimental problem such as disturbance.
2) Analysis of rolling-horizon security games in multi-agent systems: Multi-agent systems provide useful models for IoT systems. Securing multi-agent systems can be characterized as a game-theory problem. Computational capabilities for decision making in those games will be analyzed through so-called rolling-horizon games.
3) Quantization schemes for nonlinear systems: Stabilization of nonlinear systems under data-rate limitations and denial-of-service attacks are part of the research targets.
|
| Causes of Carryover |
Academic conferences in this fiscal year were held virtually as online meetings due to Covid-19 pandemic. Consequently, the budget reserved for the travel expenses were not used. In the next fiscal year, the incurring amount will be used for the purchase of experiment devices, books, and articles as well as the payment of academic conference registrations and processing charges of journal articles.
|
