Efficient Timeout Synthesis in Fixed-Delay CTMC Using Policy Iteration
We consider the fixed-delay synthesis problem for continuous-time Markov chains extended with fixed-delay transitions (fdCTMC). The goal is to synthesize concrete values of the fixed-delays (timeouts) that minimize the expected total cost incurred before reaching a given set of target states. The same problem has been considered and solved in previous works by computing an optimal policy in a certain discrete-time Markov decision process (MDP) with a huge number of actions that correspond to suitably discretized values of the timeouts. In this paper, we design a symbolic fixed-delay synthesis algorithm which avoids the explicit construction of large action spaces. Instead, the algorithm computes a small sets of "promising" candidate actions on demand. The candidate actions are selected by minimizing a certain objective function by computing its symbolic derivative and extracting a univariate polynomial whose roots are precisely the points where the derivative takes zero value. Since roots of high degree univariate polynomials can be isolated very efficiently using modern mathematical software, we achieve not only drastic memory savings but also speedup by three orders of magnitude compared to the previous methods.