Reliability analysis of mobile agent control system with multiple alternative plans

With the advancement of artificial intelligence technologies, mobile agents are becoming more commonly used in a variety of industries that require high reliability from their control systems. In an uncertain environment, if the mobile agent control system’s state transition includes only one plan, the system will enter the fault state immediately after the plan fails. Therefore, multiple alternative plans can be provided during the system design process to improve system reliability. First, this paper studies and describes the factors associated with the proposed multiple alternative plans, namely the success rate and plan implementation cost. Second, a Policy Generation Algorithm for identifying an appropriate execution sequence of those alternative plans is proposed. Furthermore, we propose a formal method-based pipeline framework for verifying the reliability of a mobile agent control system equipped with multiple alternative plans: we invoke the probabilistic model checking technique to create a Discrete-Time Markov Chain formal model of the mobile agent control system, convert the required properties into Probabilistic Computation Tree Logic formulae, and verify the model using the advanced probabilistic model checker PRISM. A case study is provided to demonstrate the applicability of the suggested methodological framework. The experimental results demonstrate that the proposed mobile agent control system with multiple alternative plans can improve system reliability while also meeting the least expected operational cost under the alternative plan set.