A workflow management framework for the dynamic generation of workflows that is independent of the application environment

Abstract

Workflow is a well-known and widely used technology in business management. Traditional workflow solutions are designed for humans and generally use a graphical representation of workflow elements that reflect the involvement of human factors. Additionally, in a situation where workflow execution is not possible, human intervention is necessary. This means that current workflow design is limited in flexibility, in terms of tasks supported, and that it cannot be easily scaled or adopted. Furthermore, current workflow design is limited in efficiency and efficacy, especially in modern environments (e.g. 5G and IoT) where problems can be complex and solutions unpredictable. This paper proposes a workflow management framework that uses dynamically generated workflows to control a managed environment. Exception detection and handling in workflow generation produce recommendations for mitigating incidents that might occur. The key characteristics of the proposed framework are its ease of implementation, flexibility and scalability. These characteristics allow for the quick definition of new tasks, known and unknown, and to assess the quality of the generated recommendation through feedback from the managed environment. Experiments performed in two different environments, robotics and networking, demonstrate the elasticity and functionality of the proposed method to dynamically generated workflows.