Partially Observable Games for Secure Autonomy*

@article{Ahmadi2020PartiallyOG,
  title={Partially Observable Games for Secure Autonomy*},
  author={Mohamadreza Ahmadi and Aruna Viswanathan and Michel D. Ingham and Kymie M. C. Tan and A. Ames},
  journal={2020 IEEE Security and Privacy Workshops (SPW)},
  year={2020},
  pages={185-188}
}
Technology development efforts in autonomy and cyber-defense have been evolving independently of each other, over the past decade. In this paper, we report our ongoing effort to integrate these two presently distinct areas into a single framework. To this end, we propose the two-player partially observable stochastic game formalism to capture both high-level autonomous mission planning under uncertainty and adversarial decision making subject to imperfect information. We show that synthesizing… 
1 Citations

Figures from this paper

Stochastic Dynamic Information Flow Tracking Game using Supervised Learning for Detecting Advanced Persistent Threats

Hierarchical Supervised Learning algorithm that integrates a neural network, to predict the value vector of the game, with a policy iteration algorithm to compute an approximate equilibrium is proposed.

References

SHOWING 1-10 OF 17 REFERENCES

The Partially Observable Games We Play for Cyber Deception

A two-player partially observable stochastic game (POSG) framework is used, wherein the deceiver has full observability over the states of the POSG, and the infiltrator has partial observability, to find robust strategies for the deceivers using mixed-integer linear programming.

Dynamic Programming Approximations for Partially Observable Stochastic Games

This work studies the POSG model and its cooperative counterpart, DEC-POMDP, and provides bounded approximation techniques, which enable it to scale POSG algorithms by several orders of magnitude.

Safe Policy Synthesis in Multi-Agent POMDPs via Discrete-Time Barrier Functions

This paper uses barrier functions to design policies for MPOMDPs that ensure safety and forms sufficient and necessary conditions for the safety of a given set based on discrete-time barrier functions (DTBFs) and demonstrates that the formulation also allows for Boolean compositions of DTBFs for representing more complicated safe sets.

Risk-Averse Planning Under Uncertainty

A method based on bounded policy iteration for designing stochastic but finite state (memory) controllers, which takes advantage of standard convex optimization methods, leading to sub-optimal solutions with lower coherent risk.

Intelligent Autonomous Agents are Key to Cyber Defense of the Future Army Networks

The industry is beginning to embrace approaches that may contribute to technologies of autonomous intelligent agents for cyber defense of the Army networks, and a growing body of research in Government and academia demonstrates promising steps towards solving some of the challenges.

Finite state control of POMDPs with LTL specifications

This work limits the search of policies over finite state controllers of a fixed size which leads to a Markov chain with free parameters, over which the probability of satisfaction of the specification can be maximized.

Finite-state Strategies in Delay Games

This work presents a very general framework for computing finite-state strategies in delay games: they exist for all winning conditions that are recognized by automata with acceptance conditions that satisfy a certain aggregation property.

Anatomy of unmanned aerial vehicle hijacking with signal spoofing

  • Sait Murat Giray
  • Computer Science
    2013 6th International Conference on Recent Advances in Space Technologies (RAST)
  • 2013
Anatomical analysis of UAV hijacking attack delivers insights about vulnerabilities, methods to exploit, mechanisms to detect, how to deter the attack and evaluates the impacts.

A Survey on Game-Theoretic Approaches for Intrusion Detection and Response Optimization

This article reviews the works relying on decision-making techniques focused on game theory and Markov decision processes to analyze the interactions between the attacker and the defender, and classify them according to the type of the optimization problem they address.