Using timed automata to check space mission feasibility in the early design phases

Abstract

According to the model-based systems engineering paradigm, all engineers contribute to a single centralized data model of the system. The German Aerospace Center (DLR) develops a software tool Virtual Satellite which enables the engineers to store, exchange and alter their corresponding subsystem data on base of a distributed system model and thus contribute to the overall mission design during concurrent engineering (CE) sessions. Each engineer has their own scope of responsibilities, e.g. satellite trajectory, communication, or thermal analysis. Tracking implications of design changes on the whole system and feasibility aspects of the design is not trivial. Having an automated feasibility checking mechanism as a part of CE which would run iteratively after each design change provides a useful feedback mechanism for engineers and for the spacecraft client. For the purpose of mission feasibility checking a domain specific language (DSL) has been implemented using the Xtext Java framework. The extended parametric data model defined in the DSL serves as an executable representation of the spacecraft mission. The idea to use such an executable model to create a preliminary mission plan and hence confirm missions feasibility during conceptual study has already been introduced by Schaus et al. at the DLR. However, the vector of values of system variables was assumed to be equivalent with the currently active component, implying that component activities are mutually exclusive. This led to over-constraining of the execution model. Our work argues that concurrency considerations are critical from the earliest design phases. Since satellite is coupled with its environment and concurrency is an intrinsic property of the physical nature, considering concurrency allows for more realistic mission plans. The contributions of this paper are the introduction of concurrency considerations at the early space mission design phases and the use of timed automata tool (UPPAAL) for the mission feasibility check during concurrent engineering sessions. As a result, with almost no overhead, the planned mission can be analyzed in a more realistic way. Furthermore, the run-times of the feasibility check amount to 10–100 milliseconds or less, which is also a significant improvement with respect to the previous work. This allows for more precision and fine granular modeling, and is a promising basis for model refinements in the consecutive mission design phases.