Abstract:Smart devices in an environment can be programmed and coordinated by a workflow in advance to achieve a user's goal. No matter how advanced or smart the devices are, devices can fail during workflow execution. In this paper, we describe an approach to remedy such situations. We apply existing concept of adaptive workflow management to a collection of devices, called a device ecology. Information about the devices are kept in a device hierarchy so that a suitable substitute device that can perform similar task can be retrieved to replace a failed device in order to ensure the workflow can continue execution. A prototype has been implemented as proof of concept.