Article Preview
Top1. Introduction
Sensors are autonomous and are used to sense temperature, light, humidity, pressure, sound, vibration, etc. Sensor Cloud is a combination of wireless sensor networks (WSNs) and the cloud to serve the end-users in a better way. These sensor networks might be of similar or different types. These days, sensor clouds are created in many applications viz., forest fire monitoring, healthcare monitoring, disaster management, military applications, and various Internet-of-Things (IoT) based applications (Tembhare et al. 2019; Kashyap et al. 2019; Pachauri 2015; Bessis et al. 2011). Sensors produce a huge amount of data in these applications that may be stored in the cloud for further processing (Khanam et al. 2022; Petrakis et al. 2018). This sensor data can be used anytime and from anywhere as per the requirements. Clouds can enable sensor-as-a-service to its users on basis of pay per use technique. A user can get data from any of such sensor networks very quickly and at the lowest cost. Users can also get data from multiple WSNs at the same time. It is possible due to virtualization within the cloud. Cloud creates virtual sensor networks that are mapped with multiple physical WSNs to respond to the user’s query that needs data from multiple WSNs. Several users can get various types of services from multiple WSNs at the same time with this integration. Thus, everyone gets benefited from this integration. WSN can get storage for its huge data, the cloud can earn for providing sensor-as-a-service, and users can access a variety of data from different networks instantly and easily (Dwivedi et al. 2018). Figure 1 presents the architecture of a sensor cloud which has three layers. The bottom layer is for physical WSNs, the middle layer is for cloud while the upper layer is for end-users.
Figure 1. An Architecture of Sensor Cloud
Sensor data is crucial in many applications. So, it must be secure, and the integrity of the data must be maintained (Dwivedi et al. 2021; Carpen-Amarie et al. 2012). Data generated by the sensors might have some unexpected behavior patterns due to outliers. The outlier is an anomaly that is generated because of malicious activities, low-quality sensor nodes, or harsh environment conditions (Gil et al. 2016; Bosman et al. 2017; Branch et al. 2013; Fawzy et al. 2013). Detecting outliers is an important research issue which has been become part of the interest of the researchers these days. By resolving this issue, we can enhance the performance of the sensor cloud (Yenke 2017; Lin et al. 2015; Xu et al. 2013; Xu et al. 2016). Various machine learning-based approaches have been developed for the detection of such outliers (Dwivedi et al. 2021; Aleksandrova et al. 2019; Rath et al. 2019; Ahmed et al. 2016; Snoussi et al. 2015; Zhang et al. 2013). Machine Learning helps the machine to learn from the surroundings. The machine itself improves its performance due to learning (Lin et al. 2019).