Carriers planning 5G deployments should not think about the project in isolation. These networks must coexist with LTE and other platforms. Moreover, they should be developed in a context in which other barrier-pushing new technologies, such as artificial intelligence and big data, will coalesce to provide end users with what they need, according to Arnaud Vamparys, Orange Network Labs' senior vice president for radio networks.
That's the plan. Carriers have a long way to go to achieve vital steps such as figuring out how to effectively move computing to the edge, implement network slicing and integrate AI, big data and other technologies. Expanding standards will be a key to doing this successfully, Vamparys told Telco Transformation in the following conversation, which has been lightly edited for length and clarity.
Telco Transformation: How important is automation for the buildout of 5G networks, services and applications?
Arnaud Vamparys: In 4G we are implementing some automation tools on our network. One of them is the SON tool, the self-organizing network. We have moved, step-by-step, from 2G to 3G to 4G, and here now is 5G. There are many parameters. It's more and more complex and we have to make sure that we have the right set of tools for our engineers.
We are preparing SON on our entire footprint, from basic parameter settings to complex stuff. We can do such things as stop use of some equipment at night when there are fewer customers. That of course consumes less energy on the network side. That's just an example, and 5G more automation will be needed.
TT: In general, what should carriers be thinking about in terms of 5G deployment?
AV: We need to take into account the already installed network, adding 5G to existing 4G. It's very important to have a step-by-step approach based on the actual operation, actual supervision of our networks. So it's very important not to start as a green field.
TT: What role will be played by platform-as-a-service?
AV: Here there still is a long journey. In the standard, on the core network side there will be quite a rupture with the arrival of NG SDN, the next-generation core network. We still have lots of work on the service-based architecture. So I do think that we have some steps to take starting with orchestration ... and then [evolving to the] next-generation core network and a service-based architecture.
So the agility will come from the fact that we can have step-by-step insight on the network to enable the adaptation of the network to serve the type of objects [on it]. For instance, supporting a smartphone, an autonomous car, a fridge or a sensor differently. We are putting in something flexible. On the other hand, we also need to be a customer-focused, to understand which kind of SLAs are really needed. [The key question:] Are we able to guarantee that on the network during each [step of the] transition?
TT: What must be done to bring computing to the edge?
AV: We still have the need to have more standardization. I think service-based architecture enabling edge computing, edge storage. And we all know that today we have a lack of standards, a lack of open APIs. So we are contributing at Orange because we will deploy more local points of presence, that we can benefit by adding storage and computing. We believe that one of the beneficiaries will be low latency applications and B2B local site high reliability [applications], which is in the first iteration of 5G.
TT: How should carriers think about network slicing, which is very promising but very complex to the O and M layer?
AV: It's quite complex. We all have these technical visions of slicing, but it's still [just a] technology vision. In our reality, we need to understand from a customer point of view. Today, we are offering mobile service and a fixed service. You really need to understand what the real needs. We did one experiment for connected cars with Ericsson and PSA. And we will show some results soon on that describe how to do a slice for vehicles … I think it's a bit like smartphones because cars [also] are objects that are moving a lot. But we are still very far from real end-to-end implementation. The results also will depend on regulation: what is authorized and what is not authorized.
TT: How can all of this be adroitly managed?
AV: Today we have already probes on the networks, meaning that we aggregate data. We anonymize all of the data we have. We are even selling that to some B2B customers through Orange Business Services. And we start to use that for the O&M element of our radio tools to get things right. We imagine that we can use artificial intelligence to do predictive analytics.
Today, we want in a closed loop to be able [to predict], let's say, when part of a site is down, or when there's an issue because something is happening in an area for the network to have some pre-set parameters and to react. And as well for us to have a bit of help with decision-making on the network.
The complexity of the network is growing, so you need some tools to reduce the complexity for your day-to-day work. We also want to make sure with data that it can enable us to achieve the right reliability for the customer. There are more and more objects connected to our network, so we want to have some isolation capabilities to be able to make sure we can handle all the situations. Big data aggregation and AI will help has well to maintain this reliability.
— Carl Weinschenk, Contributing Writer, Telco Transformation