In recent years, we have seen a growing trend for edge computing, in which data processing tasks are carried out closer to the devices and sensors that generate them. The leads to lower latency and better performance, as the data do not have to be transferred through distant networks to be processed in remote servers. In addition, as IoT technology continues to evolve, it is increasingly common to see computing incorporated directly in the IoT layer. In other words, IoT devices are not only capable of gathering data and sending them but can also process and analyse data and take decisions based on them, without the need for transferring the data to other data processing systems.
The data gathered by IoT devices must be increasingly accurate and reliable, as they are critical for applications. Therefore, it is vital to ensure that the data are accurate and have not been modified by malicious third parties. This is what is known as “trusted data”. In addition, it must be possible to transport the data from the internet to the IoT device and vice versa, and for the data routing to be carried out through the IoT network so that it reaches its destination correctly. Routing in IoT is a complex task, as the network is comprised of nodes that can be moved and may be temporary. Therefore, these must be able to communicate efficiently.
As the amount of data generated by IoT devices increases, local data processing at source is required to reduce latency and cut the consumption of energy required for data transmission. This is where TinyML technology comes into play. TinyML devices are small computer devices equipped with machine learning models that carry out data processing tasks without the need for network connection. In addition, federated learning technology can be used to train models in a distributed way so that they do not need to share data with a central server. In this context, TinyML devices require a reliable data routing protocol for the transmission of data between IoT devices with low energy consumption and broad coverage.
The research group Computer Networks and Distributed Systems (CNDS) of the Universitat Politècnica de Catalunya – BarcelonaTech (UPC) is leading the LoRaMesher project, whose aim is to provide an advanced wireless communication service LoRa for the interconnection of applications in the IoT layer. IoT devices connected by LoRa construct a mesh network, so that they can communicate with each other through the same network nodes, as occurs on the internet. In the framework of this project, various applications are being prototyped, such as the monitoring of wind at music festivals, advanced detection of forest fires or alternative communication to WiFi networks in case of emergency. The LoRaMesher technology can also be used to offer connectivity for low traffic without an operator in rural, remote areas where traditional networks do not exist. Thus, this technology can have a significant impact in many sectors, including agriculture and the forestry sector.
Image: Connection of IoT devices using LoRa wireless communication network.