Evaluating Thread network performance, locating and strengthening weak radio links

Sammanfattning: In the fast-developing world we are living in, a tech phenomenon known as the Internet of Things (IoT) has taken hold. It has seen a lot of development over the past few decades, and today there are an estimated 30 billion IoT devices active. IoT is a machine-to-machine network that senses the world with the help of sensors. The collected data is then either acted upon by controlling actuators and controllers or displayed in an easily accessible format for the end user. These networks have taken hold in many areas of our societies, such as smart cities, agriculture, industries, smart homes, and even monitoring of our well-being. With such a wide application range, many protocols have been developed for different environments and applications. This report is co-developed with a startup company Devair, which has developed a device to better calibrate and optimize air ventilation systems. A solution that could improve air quality and potentially save money regarding calibration and operational costs. Their device uses Thread as their IoT protocol, a mesh networking protocol that uses the 2.4 GHz frequency band. They have been experiencing connectivity issues with devices being disconnected from the network or not receiving packages. A common problem when using IoT devices, which normally want to use as little transmission power as possible since they are resource-constrained battery-powered devices. Therefore our aim with this work is to research and develop a method of evaluating radio links in a Thread network. We will also look at ways to strengthen these links without changing Devair’s current system. From our findings, we will in this report present a method that extracts both software and hardware metrics for all radio links in a Thread network. This is done to analyze the weak radio links between all devices in the network. The proposed method does not need to establish any DNS server or any other IP addressing service. Instead, each device’s unique RLOC16 identifier is used, which is a feature of the Thread protocol. When weak links have been located using this evaluation method, the report presents two methods to strengthen the weak links. The first which did not yield a favorable outcome is changing to a sub-GHz system compared to the 2.4 GHz system that Thread uses. The sub-GHz system used was a Texas instrument CC1310 using the 868 MHz frequency band, which was compared with the Thread system running on an MGM210PA32JIA SoC. In our tests, we tested both line-of-sight performances as well as signal penetration performance. The tests showed that the Thread system performed substantially better in line-of-sight conditions and marginally better in signal penetration performance. How-ever, we can not conclude that 2.4 GHz systems perform better than 868 MHz systems. For that, further tests need to be done on multiple systems of both frequency bands, and this work should be seen as a comparison between the CC1310 and MGM210PA32JIA systems. The second method utilizes the intrinsic message-hoping feature of a mesh network such as Thread. By placing an intermediary device between two devices that experience a weak radio link, the message can be relayed by the intermediary device that we call a connectivity router. This method improved the link quality by dividing the weak link into two shorter links. When placing the connectivity router, we found that placing it so that the signal can travel the furthest without encountering any obstacles along the way yields the best result. The connectivity router can also be used in such a way to guide the signal around obstacles with a known high attenuation rate.