Smart Healthcare: RL-Based Task Offloading Scheme for Edge-Enable Sensor Networks

With the wide application of Internet-of-Medical-Things (IoMTs) or sensor nodes which equipped with sensors. These networked sensors are used to gather enormous data from different smart healthcare applications, and this collected data process for making appropriate decisions. Edge computing is an efficient platform that provides computational resources to collect sensor data. In the meantime, intelligent and accurate resource management by Artificial Intelligence (AI) has become the center of attention, especially in healthcare systems. With the help of AI, IoMT based healthcare devices will remarkably enhance the computational speed and range. But the challenging issue in these energy-hungry, short battery life, and delay intolerant portable devices is inappropriate and inefficient classical trends of fair resource allotment. Thus, this paper proposes Computation Offloading using Reinforcement Learning (CORL) scheme to minimize latency and energy consumption. We first formulate the problem as a combined latency and energy cost minimization problem, satisfying the lack of limited battery capacity and service latency deadline constraints. Moreover, proposed algorithm search optimal available resources node to offload task towards the trade-off between energy and latency. The experimental results show the benefits of the proposed scheme in terms of saving energy, minimizing latency, and maximum utilization of node resources in edge-enabled sensor networks. We are using an iFogSim simulator to validate the proposed scheme under realistic assumptions.

I. Introduction

Nowadays, we are moving to the era of the smart healthcare system. Smart healthcare refers to the use of technologies such as IoMTs or Sensor Nodes (SNs), cloud computing, edge computing, and artificial intelligence (AI) to enable a convenient, personalized, and efficient healthcare system [1]. Such advanced technologies facilitate real-time health monitoring using medical-care applications on wearable gadgets or cell phones, encouraging people to be in charge of their health [2], [3]. A research and markets report predicted that the IoMT healthcare market is expected to grow from USD 41.22 billion in 2017 to USD 158.07 billion by 2022, at a compound annual growth rate of 30.8% [4]. These IoMT devices are equipped with sensors that generated enormous data and are transported over the Internet infrastructure to a central computing facility (typically a cloud), where intensive data processing is carried out. In the end, health information is sent back to end-users (such as doctors, nurses, and patients) for further diagnosis. Together with AI, it can be used in health screening, early diagnosis of diseases, and treatment plan selection. However, a centralized cloud data center lacks low latency, flexibility, and inability to support mobility for end-users [5], [6].