Research

  • Game Theory to Combat Evasive Adversary
  • Deception-as-Defense Framework
  • Deceptive Signaling Framework
  • Deception-Proof Mechanisms – Autonomous Intersection Control
  • Sensor Fusion – Distributed Online Learning (and Optimization)
  • Adaptive Filtering (Prediction and Estimation) Theory

  • Game Theory to Combat Evasive Adversary
    Recently, the arm race between defensive measures and adversaries for the security of digital systems has gained significant pace in favor of adversaries. Digitalization of our personal information and financial assets has made them vulnerable to threats throughout the world. It is no longer surprising to see news about data breach in world-wide organizations. Such organizations are expected to be secure against cyber threats by deploying the state-of-the-art defense mechanisms. However attacks are becoming more and more sophisticated, more stealthy, and with far wider attack surfaces. Signature based defense mechanisms can no longer defend against such advanced threats effectively. Adaptation of attacks against existing defense measures or brand new attacks have made it necessary to consider zero-day vulnerabilities in digital systems and to take precautions beyond signature based defenses. In the following (selected) papers, we provide cohesive analytical framework to break the vicious cycle of cat-and-mouse-game-like interaction in computer security through a game-theoretical lens that anticipates attacks’ adaptation:

  • M. O. Sayin, D. Sahabandu, R. Poovendran, and T. Başar, “A Cohesive Game Theoretical Framework for APT Detection: From Networks to System-Calls,” Working Paper.
  • M. O. Sayin, C.-W. Lin, E. Kang, S. Shiraishi, and T. Başar, “A Game Theoretical Error-Correction Framework for Secure Traffic-Sign Classification,” IEEE Transactions on Intelligent Transportation Systems, submitted for publication, available at arXiv:1901.10622, 2019.
  • M. O. Sayin, H. Hosseini, R. Poovendran, and T. Başar, “A Game Theoretical Framework for Inter-Process Adversarial Intervention Detection,” in Proceedings of International Conference on Decision and Game Theory for Security, GameSec, on Lecture Notes in Computer Science, Seattle, WA, Oct. 2018.
  • Go to top


    Deception-as-Defense Framework
    Cyber-physical systems, incorporating both physical and cyber parts together, e.g., process control systems, robotics, smart grid, and autonomous vehicles, have resulted in new and distinct, e.g., security related, challenges for control system design. Due to the asymmetry of information, how information flows in-between attackers and the control system (also including defense mechanisms) plays significant role in their success/failure. As an example, attackers need to use system-related information in order to learn the system dynamics, to design the best (or successful) attacks, and to evade the intrusion detection systems. Therefore, defenders can filter the system-related information to control the attacker’s perception strategically so that he/she can detect and mitigate the attacks. In the following (selected) papers, we introduce secure sensor design framework that filters the sensors’ outputs strategically and we address the resiliency of control systems prior to detection of any infiltration into the controllers:

  • M. O. Sayin and T. Basar. Deception-As-Defense Framework for Cyber-Physical Systems. In A. Teixeira and R. Ferrari (Eds.). Safety, Security, and Privacy for Cyber-Physical Systems, Springer International Publishing, submitted for publication, available at arXiv:1902.01364.
  • M. O. Sayin and T. Başar, “Robust Sensor Design Against Multiple Attackers with Misaligned Control Objectives,” IEEE Transactions on Automatic Control, submitted for publication, available at arXiv:1901.10618, 2019.
  • M. O. Sayin and T. Başar, “Secure Sensor Design for Cyber-Physical Systems Against Advanced Persistent Threats,” in Proceedings of International Conference on Decision and Game Theory for Security, GameSec, on Lecture Notes in Computer Science, S. Raas, B. An, C. Kiekintveld, F. Fang, and S. Schauder, Eds., vol. 10575, Vienna, Austria, Oct. 2017.
  • Go to top


    Deceptive Signaling Framework
    Data-driven engineering applications, e.g., machine learning and artificial intelligence, build on data, i.e., information. However, this implies that information has tremendous power on decision making, and correspondingly, information providers have influential power on decision makers. Importantly, the information providers can be deceptive such that they can benefit, whereas the decision makers can suffer, due to the strategically filtered information. To be able to deceive the decision maker, the information provider should anticipate the decision maker’s reaction while facing a trade-off between deceiving at the current stage and the ability to deceive in the future stages. In the following (selected) papers, we address how a deceptive information provider can filter the information to control the decision maker’s decisions:

  • M. O. Sayin and T. Basar. Deception-As-Defense Framework for Cyber-Physical Systems. In A. Teixeira and R. Ferrari (Eds.). Safety, Security, and Privacy for Cyber-Physical Systems, Springer International Publishing, submitted for publication, available at arXiv:1902.01364.
  • M. O. Sayin, E. Akyol, and T. Başar, “Hierarchical Multi-stage Gaussian Signaling Games in Noncooperative Communication and Control Systems,” Automatica, in print.
  • Go to top


    Deception-Proof Mechanisms – Autonomous Intersection Control
    Classical traffic lights, even the adaptive ones equipped with sensors, are not efficient for the quality of transportation. In that respect, communication based intersection control can be a novel alternative to the classical traffic lights. However, diversification in the drivers’ objectives and possibility of malicious (or selfish) ones result in non-cooperative multi-agent environments, where incentives play significant role in the agents’ actions. In the following (selected) papers, we address those issues by designing strategy-proof mechanisms in a game-theoretical perspective:

  • M. O. Sayin, C.-W. Lin, S. Shiraishi, J. Shen, and T. Başar, “Information-driven Autonomous Intersection Control via Incentive Compatible Mechanisms,” IEEE Transactions on Intelligent Transportation Systems, vol. 20, no. 3, pp. 912-924, 2019.
  • M. O. Sayin, C.-W. Lin, S. Shiraishi, and T. Başar, “Reliable Intersection Control in Non-cooperative Environments,” in American Control Conference, ACC, Milwaukee, WI, USA, June 2018.
  • B. Zheng, M. O. Sayin, C.-W. Lin, S. Shiraishi, and Q. Zhu, “Timing and Security Analysis of VANET-based Intelligent Transportation Systems,” in IEEE International Conference on Computer-Aided Design, ICCAD, Irvine, CA, USA, Nov. 2017.
  • [Patent] M. O. Sayin, C.-W. Lin, S. Shiraishi, “Managing roadway intersections for vehicles,” US Application No 15/924,979 filed on 3/19/2018.
  • Go to top


    Sensor Fusion – Distributed Online Learning (and Optimization)
    A network of agents equipped with monitoring, processing, and communication modules, brings in a new dimension for information processing applications. As an example, in remote sensing applications, each sensor can monitor certain phenomena, process the measurements, and enhance the processing performance by communicating with other sensors. In energy harvesting sensor networks, distributed processing algorithms with limited computational complexity and communication load are desirable since computation and communication can consume substantial amount of power. In the following (selected) papers, we, specifically, address the issues related to the excessive communication load:

  • M. O. Sayin and S. S. Kozat, “Compressive Diffusion Strategies Over Distributed Networks for Reduced Communication Load,” IEEE Transactions on Signal Processing, vol. 62, no. 20, pp. 5308-5323, Oct. 2014.
  • M. O. Sayin, N. D. Vanli, S. S. Kozat, and T. Başar, “Stochastic Subgradient Algorithms for Strongly Convex Optimization over Distributed Networks,” IEEE Transactions on Network Science and Engineering, vol. 4, no. 4, pp. 248-260, 2017.
  • N. D. Vanli, M. O. Sayin, I. Delibalta, and S. S. Kozat, “Sequential Nonlinear Learning for Distributed Multi-Agent Systems via Extreme Learning Machines,” IEEE Transactions on Neural Networks and Learning, vol. 28, no. 3, pp. 546-558, 2017.
  • M. O. Sayin, S. S. Kozat,  and T. Başar, “Team-Optimal Distributed MMSE Estimation in General and Tree Networks,” Digital Signal Processing, vol. 64, pp. 83-95, 2017.
  • Go to top


    Adaptive Filtering (Prediction and Estimation) Theory
    Adaptive filtering (prediction and estimation) has been extensively studied in the literature and has been (and is going to be) used extensively in the industry due to its adaptability (or flexibility) to changes in the problem parameters and scalability for large-scale problems. There are various algorithms that can lead to superior/inferior performances depending on the specifics of the problem, e.g., stationary structure, computational complexity, or large-scale data. In the following (selected) papers, we, specifically, address the issues related to stability and robustness of the adaptive algorithms while seeking to achieve improved trade-off in terms of adaptability (i.e., convergence) rate and steady state performance:

  • M. O. Sayin, N. D. Vanli, and S. S. Kozat, “A Novel Family of Adaptive Filtering Algorithms Based on the Logarithmic Cost,” IEEE Transactions on Signal Processing, vol. 62, no. 17, pp. 4411-4424, Sept. 2014.
  • M. O. Sayin, Y. Yilmaz, A. Demir, and S. S. Kozat, “Krylov Proportionate Normalized Least Mean Fourth Approach: Formulation and Performance Analysis,” Signal Processing, vol. 109, pp. 1-13, Apr. 2015.
  • N. D. Vanli, K. Gokcesu, M. O. Sayin, H. Yilmaz, and S. S. Kozat, “Sequential Prediction Over Hierarchical Structures,” IEEE Transactions on Signal Processing, vol. 64, no. 23, pp.  6284-6298, 2016.
  • Go to top