Fast heuristic approach for control of complex authentication systems

In secure networks, the authentication process, which verifies a user's identity, is a sensitive operation. Balancing the authentication process's security and usability while keeping operating costs low is a major challenge. We investigate this process in a secure system that has multiple authentication methods available. The goal is to find a fast and easy-to-implement approach to assign incoming requests to authentication methods in a way that balances cost, usability, and security. We model the system as a continuous-time, infinite-horizon Markov Decision Process. Then, we propose a new, fast heuristic approach to solve the problem, as the curse of dimensionality limits the effectiveness of exact methods. The proposed heuristic uses closed-form approximations to define threshold policies that are easily implementable. Numerical experiments show that our approach yields robust, near-optimal solutions, with high accuracy, for a wide range of problem instances.