Author: Ivana Škender
Mentor: Matej Rojc
Comentor: Grega Močnik
Degree: Level 2 study program Electrical Engineering, course Electronics
Date: June, 2020
DKUM: IVANA ŠKENDER
In the master thesis we developed an IoT system on a Microsoft Azure cloud platform. On the Microsoft Azure cloud platform we have used several services that are very useful in the field of the Internet of Things. The IoT system in the thesis thus represents an advanced solution for the development of smart IoT environment, where hardware monitors the humidity, temperature and pressure in the faculty. The Microsoft Azure cloud platform enables the use of many services that are crucial in the IoT systems. The proposed IoT system presents many possibilities for the development of advanced services based on algorithms for advanced analytics and artificial intelligence in the Microsoft Azure cloud platform.
Azure IoT platform is a cloud computing platform developed by Microsoft. The platform supports a wide range of services that enable the design, development and management of applications in a large global Internet network. The cloud service provider takes care of the maintenance of the basic infrastructure, and the user manages the cloud resources as if the resources were in his own data center. The Azure IoT platform provides more than a hundred services, including applications on virtual devices, support for artificial intelligence, and mixed reality.
In this thesis, we want to develop an IoT system on the Azure platform, in which we would obtain real data from the IoT environment in the premises of the faculty and transmit them to the cloud. Within the presented system, we limited ourselves to battery-powered sensors in rooms that have the ability to measure temperature, humidity, and air pressure. In the proposed IoT system, each gateway represents an IoT device on the side of the Azure cloud platform, thus enabling the control of the included gateways, adding new gateways, etc. Data captured in the premises are also available for further use, e.g. to turn the air conditioner on/off automatically. The presented solution does not necessarily require interaction with users, consequently IoT devices can also perform activities on the basis of what was instructed by another IoT device. This is also the basic idea of the Internet of Things. The Microsoft Azure cloud platform allows us to store and process the captured data in the cloud in the proposed system, and consequently the data can be accessed even when we are not in the premises of the IoT environment or we are not even in college. The Microsoft Azure platform ensures adequate data security when storing data through the use of its services. The proposed solution also opens up various possibilities for upgrades and extensions.
In the proposed system, we need a few communication protocols to transfer data from the IoT environment to the Azure cloud. We used the BLE protocol to send data from sensors to the physical gateway. BLE is especially suitable when battery life is more important than data transfer speed. BLE data transfer is basically one-way communication. For example, the RuuviTag platform uses BLE communication and has its own MAC address, based on which we know which RuuviTag sends specific data. In the proposed system for connecting the client and broker, or Raspberry Pi and Azure IoT Hub platform, we use an encrypted TLS connection. Because MQTT messages are organized according to suggested topics, the developer has the ability to determine which clients should receive and send only certain messages. For example, sensors can send data to a broker by publishing their data in the “sensor_data” topic, and the broker then passes the data to those data storage and processing applications that are subscribed to the “sensor_data” topic. On the other hand, there is a console application through which the system administrator sends commands to adjust sensor configurations, such as sensitivity and sampling frequency, with changes posted in the “config_change” topic. The sensors are subscribed to the “config_change” topic, and can therefore obtain information about new events in this topic.
We need hardware and software for the proposed system, as well as cloud services on the Microsoft Azure cloud platform. For the hardware we used: RuuviTag sensor platforms for measuring temperature, humidity and air pressure in the room. We used the Raspberry Pi as a gateway between the physical and digital worlds. The Raspberry Pi uses BLE communication to connect to sensors and the MQTT communication protocol to send messages to the Microsoft Azure cloud platform. We wrote the code in the Python programming language, which reads the sensor data, connects to the cloud platform, and sends the sensor data to the Microsoft Azure cloud platform. In order to achieve the best possible flexibility and easy upgrade of the system, we also set up a web application that allows you to add new sensors and new transitions to the system. We distributed the RuuviTag sensor platforms in several rooms at the FERI faculty. Thus, two platforms are located in space G216, two in space G215, and two in the Bacquerel (BQ) laboratory. All services on the Microsoft Azure cloud platform are accessed through the Azure Portal user interface. Azure Portal allows us to use and manage all the services offered by the Microsoft Azure cloud platform. The first service we used was the IoT Hub. The IoT Hub is a hub in the cloud that connects the physical gateway of the IoT system and other cloud services. Stream Analytics is available for fast and easy data transfer within the Microsoft Azure cloud platform. This service is not free (Free, or Students account), and if it was used continuously, its price would be around $ 87 for a period of one month. Therefore, we used the Azure Functions service to transfer data. With the help of the Power BI tool, we finally visualized the data that the sensor platforms send to the Microsft Azure cloud platform. In the Power BI Desktop environment, we first connected to an SQL server created on the Azure platform and obtained data using the ‘DirectQuery’ connection method. We then processed and used that data to produce a report that we published on the Power BI Service. Within the Microsoft Azure cloud platform, we can also create a quality weather forecasting system with the help of Stream Analytics, Azure Machine Learning Studio and artificial intelligence AI. For this purpose, the program code on the Raspberry Pi was also adapted for reading data from only one RuuviTag platform, namely only for reading temperature and humidity sensors, because in this case we only need these parameters. The Stream Analytics tool captures the data we send to the IoT Hub and then forwards it to the Blob Storage repository and also to the Power BI Service. On a dashboard created in the Power BI Service tool for real-time data, we can then monitor data on current temperature, humidity, and rain probability. Temperature and humidity data are obtained from sensors in the environment, and the probability of rain is calculated based on this data in the Machine Learning Studio service.
Conversational agents in intelligent environments by using the Home Assistant...