Underground coal mines are harsh in the working environment, which leads to more causality. Therefore, modern mines are using sensor networks to monitor the mines regularly to minimize mine accidents. Hence, very limited real-time monitoring systems are developed using a Real-time operating system (RTOS). Contiki-NG is a popular embedded application development framework for Wireless sensor networks and the internet of things (IoT). This paper provides a detailed roadmap to design a real-time monitoring system for underground coal using Contiki-NG. The System design comprises designing sensor node with modified RPL protocol and border router using TI CC2650 launchpad and raspberry pi 3. The Node-RED middleware collects data from the application environment using a sensor node. Store the data into the MongoDB database and visualizes sensed data in a web browser in the monitoring center.

Hierarchical routing protocols are the most prominent energy-efficient routing protocol to attain longer life time in wireless sensor network. In the literature, many cluster-based routing protocols have been proposed by considering energy is the most important parameter. These techniques have been implemented using machine learning algorithms and simulated for approximated network model but not in real time scenarios. In this paper, we have implemented centralized clustering protocol in real time scenario and analyzed energy elapsed during different phases of the nodes. Texas Instruments CC2650 ultra-low power wireless microcontroller is used as target hardware platform and indoor environment monitoring as target application.

To distinguish the classes of unknown data, classification is an important technique. If implemented on hardware these techniques provide more accuracy, faster response in real time predictions than any software implementation. Existing hardware implementations are not at all cost effective, and require specific knowledge of hardware platform. In this paper a cost effective implementation of a statistical classifier has been proposed on Raspberry Pi, an open source hardware platform. This proposed implementation provides on board training of the classifier and real time class prediction. Moreover, it does not require knowledge about the hardware platform to classify the data.