If you witness a pile-up on the motorway and imagine its consequences, you probably don’t contemplate all the telecommunication needs generated by such an incident. And yet, emergency services must be coordinated, health services must be informed in real time, transport operators must manage traffic, and so on. All these operators need to communicate efficiently, not only internally but also with one another. The systems used by these different organizations are often compartmentalized, slowing down the decision-making process and intervention on the ground.
Such situations highlight the need to set up network and service infrastructures that are capable of simultaneously meeting the demands of several operators from different sectors of activity. Among the flagship projects of the ‘5G and Future Networks’ PEPR, the NF-MUST project aims to partly satisfy this need by elaborating a service architecture conducive to cooperation in multi-operator and multi-sector operations and communications.
A real challenge because, in addition to the specific technical requirements and the diversity of communication and security protocols, end-to-end coordination between sectors means calling upon different operators. “Bearing in mind that 5G and 6G operators already benefit from heterogeneous technological segments, that they must concatenate in order to set up end-to-end infrastructures [free from interruptions/discontinuities] capable of meeting user needs”, emphasizes Djamal Zeghlache, professor in Networks and Services at Télécom SudParis, who is holder and leader of the NF-MUST project. “In order to interconnect different operators, we therefore have to create a network of networks! Which means broadening the concept of slicing to the multisectoral approach.”
“Like a mille-feuille in which each layer represents a part of the network”
Borrowed from the cloud computing approach, ‘slicing’ has developed alongside 5G, driven by the virtualization of services and networks. Virtualization consists in splitting a single hardware base into separate virtual environments, which makes it possible to pool physical resources. A router, for example, can be split into compartments, each compartment corresponding to a virtual router serving a particular user (tenant). Several tenants can then share the same equipment while remaining separate from one another. These virtualized environments provide a flexible, pooled base on which various applications and services can be operated – typically, health and emergency service communication – without the need for specific hardware.
Slicing takes this concept a step further, by splitting up the entire network, virtualized resources included, into ‘vertical’ slices, such as energy, health care and transport. “The notion of pooling resources such as calculation, storage or memory, is extended to the sharing of network infrastructures and services”, Djamal Zeghlache continues. In concrete terms, from their 5G and 6G network infrastructures, operators propose slices with customized configurations to meet the demands of the different vertical sectors of a specific tenant. “Like a mille-feuille in which each layer represents a part of the network, dedicated to a specific sector or tenant”, the researcher adds.
Each gets their own slice, which includes not only the virtualization of network infrastructures, but also the dynamic allocation of specific resources (such as bandwidth, traffic flow priority, etc.) in accordance with their needs. This makes it possible, for example, to implement 5G infrastructures at a smart port, where the operators concerned (port authority involving ship-owners, crane drivers, transport companies, etc.) share a network while benefiting from slices reserved specifically for their own respective operations. “These infrastructures are linked together, in order to provide services suitable for the different operators and tenants present in the port environment”, Zeghlache adds.
The challenge of multisectoral coordination
One of the goals of the NF-MUST project, therefore, is to extend the slicing concept to the multisectoral level. Different sectors (energy, health, transportation, etc.) could then share a common infrastructure while each benefiting from specific slices to accommodate their individual needs.
Firstly, this requires a good understanding of the context of a particular multisectoral use case: identifying the immediate needs of all operators in order to transform them into a single request to be sent to all networks. These networks will then provide the environments in which the services required for the specialized applications of the different sectors will be fulfilled. To use the car accident example once again, the NF-MUST architecture will thus provide an end-to-end slice to ensure cooperation between the forces of law, fire brigade, emergency services, motorway patrol, etc. In other words, all communication services and networks associated with the specific involvement of each of these operators.
However, while sharing network resources, these different operators must be able to isolate their own sensitive data. The NF-MUST end-to-end architecture must allow each of them to operate in their own environment, while communicating with the others in a flexible manner. “We should think of it as a building in which each floor or space (stairwell, room, hallway) is reserved for a different tenant, linked to the others by walkways or separated by partitions. In this environment, it is possible to change the configuration – add partitions, move them, make them impermeable, etc. – in such a way that each operator can function independently or, conversely, cooperate with a trusted partner”, illustrates Djamal Zeghlache.
Toward dynamic management of resources
The other goal of the NF-MUST project is to make slicing dynamic and automatic, in order to adjust resources to the immediate needs of users. Currently, slices are often predesigned and predefined before being made available in a catalog of services. These predefined slices have been identified as meeting specific needs and chosen to satisfy the customer’s requirements. The slicing dynamic aims to go beyond this predefined approach.
For this, it needs to analyze in real time the available infrastructure and verify those resources yet to be allocated. These resources include virtualized entities – which can be used simultaneously by different clients – as well as physical ones, which cannot be virtualized and must be shared in succession between users. The slices must then be configured to accommodate the expressed needs.
A catalog of customizable services
The following step involves making the service catalogs accessible to users, so that they can compose their own slices to match their specific needs, like modular building blocks. When a client expresses a need, whether in technical jargon or everyday language, the system has to translate, then analyze this query in order to identify the necessary components from the catalog. It therefore functions as a knowledge base containing elementary blocks from which slices can be built. Each slice is constructed by assembling different elements found in the catalog, while ensuring that users are uniquely authenticated and hold the required rights of access to their services. Djamal Zeghlache compares this to “combining LEGO® bricks to meet the client’s request”.
In short, the composition of a slice entails knowing which services are available, authenticating and understanding the client’s request, breaking it down into elementary blocks, and finding the compatible components in the services catalog. Although the catalog is at the core of this process, progress remains to be made to enhance this knowledge base and adapt it to multisectoral uses. “Although lots of us are working on the topic, service catalogs are not yet very well stocked. The procedure isn’t fully automated. So for now, this is a medium-term goal”, says Djamal Zeghlache.
