Gaussian Distribution-Based Machine Learning Scheme for Anomaly Detection in Healthcare Sensor Cloud

1. Introduction

These days, various smart systems have been developed to facilitate monitoring and management of physical and human environments in many ways. Such smart systems are also known as Internet of Things (IoT) systems. Sensor based IoT systems have various applications such as healthcare monitoring, battlefield monitoring, street monitoring, disaster management, military applications, forest fire detection, unmanned vehicles and manufacturing industries (Bessis 2011; Lounis et al. 2016). Such IoT applications generate a huge amount of data that is usually stored at cloud to increase usefulness of the resources (Thilakanathan et al. 2014). Sensor networks are integrated with cloud to improve the effectiveness of the applications. This integration is termed as sensor cloud which is beneficial for both sensor networks and cloud. Various sensor networks store their sensed data at the cloud. These physical sensors are mapped with virtual sensors at cloud. Sensor cloud administrator integrates the sensed data from various sensors into the unified standard with help of virtualization at cloud. Thus, cloud can provide sensor as a service with help of virtualization to the multiple users according to their choice and demand. Any genuine end user can access the data of one or all authorized sensor networks just in one click with help of this integration (Dwivedi et al. 2019). Figure 1 presents a healthcare monitoring system where each human body behaves as a sensor network. Here, data from various wearable body sensors of many patients have been stored at cloud through base station such as mobile phone. Different types of authorized users viz., doctors, nurses, medical students and researchers can access the health records of the patients using their credentials. Doctors can provide medical support to the patients anytime and from anywhere with this system. They can help the patients instantly if the emergency case is observed. Data owners may also earn money for providing their data at cloud in some cases. Cloud can provide sensor as a service to the authorized students and researchers by providing them various types of data. Thus, legitimate end users can get data of one or more patients easily and quickly. Doctors, nurses, students, researchers and patients may belong to either same or different hospitals in this healthcare system. In this way, everyone is benefitted with this sensor cloud integration.

Figure 1.

Healthcare monitoring sensor cloud