As the number of Internet of Medical Things (IoMT) increases, the need for performing on-premises tasks within hospitals or medical centers also increases. Many healthcare organizations are progressively embracing or adopting an edge computing paradigm such that computationally intensive tasks can be processed at the edge of the network in order to avoid latency and network reliability issues associated with offloading tasks to the cloud. The problem, however, hospitals or medical centers may not be equipped with sufficient computing resources that can process advanced ML or AI tasks efficiently. In addition, some tasks may not be easily offloadable or contain sensitive patient healthcare data which increases the risks of having malicious attacks. In this paper, we extend our Edgify resource provisioning framework to consider the task offloading of healthcare applications' involving patients' data as a multiple criteria decision making (MCDM) process that often involves multiple conflicting criteria (e.g. data privacy risks, costs, latency, network reliability, among others). We evaluate our proposed framework through a number of experiments which demonstrate the usefulness and effectiveness of employing our optimization approach within hospitals or medical centers.
翻译:随着医疗物品互联网(IOMT)数量的增加,在医院或医疗中心内执行现场任务的需求也有所增加,许多保健组织正在逐步接受或采用一种边际计算模式,这种边际计算模式可以使计算密集的任务在网络的边缘处理,以避免与向云端卸载任务有关的潜伏性和网络可靠性问题;然而,问题可能没有为医院或医疗中心配备足够的计算机资源,从而无法有效地处理先进的ML或AI任务;此外,有些任务可能不容易卸下,或含有敏感的病人保健数据,从而增加恶意攻击的风险;在本文件中,我们扩大我们的强化资源提供框架,以考虑将涉及病人数据的保健申请卸载任务作为多重标准决策程序(MCDM),这往往涉及多种相互矛盾的标准(例如数据隐私风险、成本、惯用率、网络可靠性等)。我们通过一些实验来评估我们提议的框架,这些实验表明在医院或医疗中心内采用优化方法的效用和有效性。