Article Preview
Top1. Introduction
Nowadays, computing is not just limited to a single system; in fact, every device is being redesigned to be connected via the Internet to provide facilities and on-demand computing. This technology is known as the Internet of Things (IoT). In this emerging technology, various types of computing devices and techniques have been integrated to facilitate the users in many ways (Gubbi et al., 2013). The sensor cloud is another such integration. The sensor cloud is an integration of sensor networks with the cloud. The sensor cloud facilitates its end-users to get data from various sensor networks, just in one click (Dwivedi et al., 2019). It is possible due to the process of virtualization performed in the cloud.
IoT applications are based on sensor networks. There are two types of sensor networks namely general-purpose and special-purpose. Body sensor networks are the special-purpose networks used in healthcare systems. Cloud computing is a technology that is based on pay-per-use policy. Users have to pay only for what they use. Clouds can be categorized into three types, viz., public, private, and hybrid. Cloud provides mainly three types of services namely software-as-a-service (SaaS), platform-as-a-service (PaaS), and infrastructure-as-a-service (IaaS). The sensor cloud architecture enables for one more type of service called sensor-as a-service (SeaaS).
When sensor networks are integrated with the cloud, they facilitate the sensor owner, cloud service provider, and end-users in many ways. This integration also improves the effectiveness of sensor applications. To overcome the storage limitation of sensor networks, the cloud can provide an effective and economical space for storing huge amounts of data. Data analytics tools can be applied to get insights on this huge data.
Sensor clouds find their utility in several information systems such as smart healthcare, forest fire detection, street monitoring, battlefield monitoring, military applications, disaster management (Bessis et al., 2011; Thilakanathan et al., 2014; Lounis et al., 2016). A huge amount of data is generated in such applications. These data are very crucial. Therefore, security becomes a prime concern with these applications (Ahmed et al., 2016; Petrakis et al., 2018).
Figure 1. The architecture of a smart healthcare sensor cloud
A smart healthcare system is a system where doctors can monitor the health status of their patients and prescribe the suggestions remotely. The architecture of a smart healthcare sensor cloud is shown in Figure 1. Various real-time facilities can be provided to the patients using this smart system. Emergency cases can be handled quickly and required services can be provided instantly using this system. This system allows remote treatment as well as reduces the treatment cost. In this system, patients are equipped with various body sensors. Several types of patients’ data such as SpO2, heart rate, body temperature, and blood pressure can be sensed and transferred to the cloud for further processing. Nurses, junior doctors, medical students, researchers, and other authorized users can access the patients’ health data as per their requirements. Patients’ health data are very crucial and any false information or outlier can cause various problems. Hence, the sensed information must be precise and accurate. Outliers are the data that are quite different from the other members of a set or group. It might be caused due to any malicious activity by an intruder or some failures in the hardware. This paper focuses on outlier detection in sensor cloud data of the smart healthcare system.