Optimizing Mission Abort Policy With Threshold Voting of Imperfect Shock Detectors

Shock count is a key parameter used in designing mission abort policies (MAPs) for systems executing their operations under random shock conditions. Existing models mostly assume a perfect mechanism of detecting shocks. In practice, the shock detection system may fail to detect shocks that have occurred (false negative) or flag non-existent shocks (false positive), both leading to wrong shock count and misleading MAP designs. This work's contribution lies in modeling a single-attempt mission system with a fault-tolerant shock detection system that applies threshold voting among multiple imperfect detectors to contribute to the mission abort decision based on shock count and system operation time. A probabilistic approach is put forward for assessing mission performance of the considered system in the form of task success probability (TSP), survival probability of system (SPS), and expected losses of mission (ELM). An ELM minimization problem is further formulated and solved, which aims to determine the optimal tri-parametric MAP, achieving a balance between TSP and SPS. We analyze a drone-based surveillance system to showcase the suggested model. We also examine the impact of key parameters (cost, shock occurrence rate and detection probability) on mission performance metrics and on the best-obtained MAPs, leading to important managerial recommendations.