Automated risk detection for medical device networks with hard real time requirements

Recent research projects [1], [2], [3] show an increasing trend from standalone medical devices to heterogeneous networks of medical devices. This poses new challenges in the field of admission and asks for novel methods to verify the network system's correct functioning and guarantee for the patients safety. The traditional, monolithic admission approach is not feasible in this case because detailed information about the concrete medical device interconnection would be needed to ensure the complete system's functioning. However, this information is unknown until the individual devices of different manufacturers are integrated into a specific medical device network. Manufacturers can only specify and guarantee the individual devices' standalone real time properties which can be distributed via ISO/IEEE 11073 medical device description format in a machine readable way. In this paper, we propose a method to automatically check whether certain real time requirements of medical device networks will be met at runtime. Our method works by building a directed, acyclic graph based on the information supplied along with the device description files as well as the specified network configuration. We map properties of the annotated information flow graph to the cyclic nature of TDMA (Time Division Multiple Access) real time networks, provide an algorithm to extract timing information from this graph and formulate three automatically verifiable constraints which are vital to the proper functioning of the network.