A factory where every machine, every sensor and every vehicle is connected, sharing information in real time to optimize processes and ensure any moving around is done safely. This is one of the promises behind the Internet of Things (IoT) concept, which, when combined with 5G, can transform the industry and transport sectors. The FITNESS project is supported by PEPR ‘5G and Future Networks’, and involves several IMT schools, the CEA, the CNRS and Inria, aiming to develop this vision by providing solutions that are tailored to the challenges we are facing today and will face in the future.
The IoT systems developed for this project must be in line with mission-critical needs for industry and transport, where reliability, low latency and resilience are essential. “The Fourth Industrial Revolution requires robust communications to guarantee the continuity of critical processes such as production line management. A breakdown could have a serious impact on security”, reveals Olivier Boissier, member of the Laboratory of Informatics, Modeling and Optimization of Systems (LIMOS CNRS UMR 6158), Professor in Computer Science at Mines Saint-Étienne, and joint project leader of FITNESS. Likewise, in the transport sector, IoT systems need to be reliable in high-density environments such as public transport and self-driving vehicles.
Massive IoT: how can density be managed?
Massive IoT, or massive scale IoT, refers to environments where the concentration of connected objects is very high, such as logistics warehouses, smart cities and transport infrastructure. The range of these IoT objects is vast and includes sensors, actuators, robots and communications devices. In massive IoT, there can be thousands or even millions of connected devices operating at the same time.
Despite the density of connected objects, providing enhanced services can be a great challenge in terms of energy management, interoperability and reducing interferences. The goal is to guarantee that each object operates independently with efficient communication, without causing any disturbances within the network.
A typical example is the maintenance facilities of large firms (such as SNCF), where hundreds of sensors monitor the trains. “These facilities are huge, and the large quantity of IoT objects puts considerable demand on the physical resources that support communication. It is therefore necessary to distribute the frequency bands and coordinate what communicates and when, so that everything works optimally”, explains Olivier Boissier.
Challenges related to energy and the optimization of resources
For large firms like SNCF, energy management is key, and massive IoT is no exception to this rule. IoT objects can sometimes be restrictive devices, particularly in terms of energy resources. They are often battery-powered, and one of the key challenges is to make sure this battery lasts as long as possible. A solution is therefore to operate them intermittently: “Objects are put on standby when they are not in use, and are only powered back on when they need to produce signals”, specifies Olivier Boissier. Another option is for objects to communicate over shorter distances, using a less powerful signal. The information therefore needs to be received close to the object, using a mobile robot for example.
These optimization issues also apply to calculation and communication resources. IoT objects cannot process all data, some devices are only used to receive and send information, and others, such as mobile robots, can move around and act on their environment. Therefore the requirements vary depending on the scenarios, and so transmission strategies must be adjusted accordingly.
This means that there needs to be a whole range of solutions, capacities and energy consumption. “Some techniques can provide very low latency as long as the volume of data isn’t too significant, others will conversely process enormous volumes, which can be detrimental to speed”, explains Guillaume Lozenguez, who is also a computer science researcher at IMT Nord Europe and involved in one of the FITNESS project work packages. “What is interesting is just how dynamic the IoT networks are, and how switching from one technique to another depending on the requirements will impact the architecture”, he adds.
A diverse range of communication protocols
Once the need and associated constraints have been identified, it’s necessary to choose the right technology. Each protocol – including Wi-Fi, LoRa (Long Range Wide Area Network), Narrowband IoT (NB-IoT) and 5G – has its own specific characteristics in terms of scope, energy consumption and bandwidth. For example, LoRa is ideal for long-range communications with low energy consumption, whereas Wi-Fi is preferable for exchanging large volumes of data over short distances.
“Historically, firms have often used different technologies to meet specific needs, each one presenting its own advantages and disadvantages”, explains Guillaume Lozenguez. With its advantages over other protocols in terms of latency, throughput, management of widespread communications, and security, 5G could be the ideal standard. But the question of usage remains key: “We have to ask ourselves whether it’s worth deploying these new features, if they are not used afterwards”, stresses Olivier Boissier. “As such changes come at a cost, both financial and to the environment.”
Dynamic switching to ensure optimal connectivity
The FITNESS teams are therefore looking to combine these different modes of communication, and are working on developing systems that can switch from one protocol to another in real time depending on needs. “If a robot needs to send high resolution images, it can switch between Wi-Fi or 5G connections to guarantee adequate throughput, then return to less energy-intensive communication once the information has been sent”, explains Guillaume Lozenguez.
How does AI fit into all this? Of course, AI has a role to play in this dynamic optimization. It can be deployed both in the lower layers to improve bandwidth management, and in the upper layers for applications to make best use of the transmission capacities. AI therefore helps the systems to better coordinate resources thanks to techniques such as machine learning or knowledge graphs. It provides an additional, powerful tool to improve the performance of IoT networks and makes them more flexible in meeting the dynamic needs of industry or the transport sector.