Towards solving the software dependency and update execution order problem for industrial plants

Upcoming security regulations and shorter product cycles increase the need for software updates in industrial plants. Instead of today's "never change a running system" agility similar to IT standards regarding software updates is needed in the future.

This requires supporting tools to handle the heterogeneous landscape of industrial plants. In this paper an approach for automatic dependency solving and update execution order determination is presented. An ontology based knowledge graph is used to unify software bills of material, security ratings, functional changes and the plant topology. The graph is translated to a package management/dependency solving problem allowing the use of existing solving algorithms to generate feasible update states while maximizing the update value. The update order per device is translated to the shortest path problem and solved by a proven algorithm. Following this, the necessary update steps are ordered considering topological constraints.

The approach is implemented and evaluated on two plant configurations with five test cases, covering optimal, parallel-execution, and unsolvable scenarios. ... mehrCorrect update selection, optimized ordering, and clear diagnostics in unsolvable cases are demonstrated.