A Workload Characterization for the Internet of Medical Things (IoMT)

摘要

We perform an extensive study of medical applications that will potentially execute on the Internet of Medical Things (IoMT), from an edge computing perspective. Using this study, we perform a workload characterization of potential IoMT applications and explore the microarchitecture implications of these applications. Our study includes workloads spanning a variety of medical applications including medical image processing algorithms, inverse Radon transform, and implantable heart monitors. We compare these workloads' characteristics to an existing embedded systems benchmark suite, MiBench, to reveal their differences and similarities. The analysis presented herein will enable the study and design of right-provisioned microprocessors for the IoMT, and provide a framework for studying the execution characteristics of workloads in other emerging Internet of Things application domains.