ETSI recently announced it was launching a number of new ISG research groups, including one focussing on Integrated Sensing and Communications [ISAC]. Amid all the excitement and hype in recent years about general purpose AI, metaverse, Open RAN, the ‘internet of senses’ (remember that?), non-terrestrial networks and so on, this is a topic which has flown relatively under the radar. However, the ITU’s 6G definition includes expanding on 5G’s three aspects, but then layers on non-terrestrial networks; integrated AI and communications; and ISAC.

So how come this has gained such influence and its own ISG? Is it just very effective hype for a small group of influencers? 6GWorld was able to catch up with the newly-appointed chair of this new group, InterDigital’s Alain Mourad, to understand what all the fuss is about.

Where’s The Sense in ISAC?

It might have escaped some caught up in the hype cycles, but using the electromagnetic spectrum to sense things isn’t exactly new; we’ve been doing that with the visible spectrum for millions of years and created sensors decades ago for exactly that. Is there a good reason to build sensing into our communications infrastructure when it’s already quite prevalent elsewhere and XR developers are working on things like volumetric video?

Mourad is enthusiastic about all these different kinds of sensing – he sees no need to reinvent the wheel.

“You have one level of integration where the sensing and communication are being done potentially separately. We would be looking at a situation, for example, where there are certain frequencies for the sensing versus the communication… even different modes, different systems,” he explained.

“So basically the sensing doing its own thing, and the results of the sensing are fed into the communication in a way.”

This level of integration – coexistence – is something which could be done today using, for example, already-installed CCTV cameras operating in the visible and infrared spectrum linked with 5G mobile signals. By why would we even want to do that? Mourad explained.

“The sensor data is useful to the network to improve the network performance – better beamforming, if you like, or better delivery of a service. But it also works the other way around. So, you know, the communications can also feed information into the sensor system.”

If you think about that, it makes sense. While a camera, for example, can sense objects’ locations with a very detailed resolution, the communications network can do other things very well, such as identify and track the movement of objects (through identifying doppler shifts in reflected signals) and, to different degrees depending on frequency, penetrate through solid objects, allowing ‘sight’ beyond what is visible. Using something outside of the visible spectrum could also help compensate for visual mistakes such as misinterpreting a shape due to camouflage or a trick of the light.  

As a result, there is a good deal of complementarity here which could potentially open up some different kinds of services in, for example, security (scanning for concealed weapons) while also improving the network’s understanding of the surroundings for more accurate channel modelling and more efficient communications. This might be one of those elusive win-win situations.

What ETSI’s Doing With ISAC

According to Mourad, essentially they are looking into everything right now. As well as the ‘coexistence’ style of ISAC where we have two separate systems for sensing and for communications, he described two other types that are more ambitious.

“One of the scenarios of integration is to say if I have one system doing both, I may be able to do the communication on certain frequency ranges like sub-6GHz, and then perhaps the sensing on millimetre wave.”

This would do away with the need to coordinate data streams with an external system, which is liable to simplify life for the telco commercially, if not technically. There is more, however.

“The ultimate integration level that everybody would like to see is where it’s exactly the same frequency, same spectrum, the same transceiver chain, the same system and delivering both capabilities,” Mourad explained.

“Because if you speak to operators, they will tell you “We would love to see that sensing capability in our network, but we want it for free.” They don’t want to pay any extra for it in terms of spectrum or they don’t want to slice the existing spectrum for communication to enable the sensor.”

Unsurprisingly, the research and development needed to make this ‘ultimate integration’ work is very different from a coexistence model or even the two-frequency model. The questions and the challenges are different.

This opens up space for ETSI, as opposed to 3GPP, to do some valuable research work.

“We know that 3GPP can only do so much with the time allocated to this topic, and it’s exclusively focused on an existing model and trying to patch it, not finding a better one,” Mourad commented. This is largely down to the evolutionary process that the different releases take.

“If you take 3GPP RAN, they have to go one baby step at a time,” he explained.

“They’ll take, you know, a couple of use cases and take them not as being provided from external sensor sources, but more from assuming that we’re going to get them out of the radio signal. So what implications would these use cases have on the channel model? This is what they’re going to be focusing on in the next 1.5 years.”

At the same time 3GPP’s Systems Architecture group will also be following an incremental approach. Said Mourad: “It will look at ways of changing or adjusting the architecture in 3GPP to embed in the sensing capability.”

This is, in his eyes, why ETSI’s work is crucial.

“We would like to really look at the bigger picture,” Mourad enthused.

“I mean, first of all, we’ve got 32 use cases from 3GPP. There are many more that people are talking about. Let us put out a roadmap – short term, mid-term or long term, in terms of the level of ambition and advancement of these use cases. So what can I do for some use cases already today with 5G?And what can I do if I make small, small enhancements? What kind of use cases can I actually enable in the mid-term and which use cases are going to be most challenging and most advanced, which will require significant changes to all layers of the system?”

As well as the use cases and roadmap they will be looking at three other areas:

• Channel modelling for all kinds of different services and scenarios, as outlined above, with testing through partner labs.
• The system architecture, other than 3GPP’s work: “If you relax the constraint of backward compatibility you might actually come up with a different architecture, or the sensing to be natively embedded as a capability into the system. So how does it look different from an evolutionary architecture, you know, adding the sensing capabilities on top? These kind of studies will be looked at in that work.”
• Privacy, security and sustainability: “Well, of course, between sustainability and privacy there is not much in common but in our work scope for the ISG it makes sense for us to combine them just for the sake of optimizing the number of deliverables.”

While the work items still need to be officially approved there is a fairly clear pattern emerging, which Mourad spelled out:

“You see how we are positioning the ISG really to be in a position to feed inputs to 3GPP Release 20. We are not trying to duplicate the work in Release 19, but… we will keep a close eye on what 3GPP is doing and then really complement it and then go a little bit further.”

Invasion of the Privacy Snatchers

On the topic of ‘going a bit further,’ Mourad described something a bit more demanding than the coexistence scenario at the top of this article, but still something well within scope for 5G.

“Imagine your base station becoming a radar. And for that radar, you need not to do anything in terms of changing the signal, changing the base station hardware. It’s just still using the 5G radio signal. What meaningful sensing you can achieve with that?”

This is an object of research, but only up to a point. Mourad described two types of sensing: monostatic, which is where a transmitting station will ping and then receive back the reflected signal; and bistatic, where the transmitting station is different from the receiving one. Telecoms networks today are very much set up for a bistatic environment – that’s essentially communications – but not for monostatic. That would require a hardware upgrade.

“If you focus only on [bistatic], we can already actually enable that sensing in 5G Advanced, we don’t have to wait for 6G.”

Observant readers will have noticed that, up to this point, there has been no mention of user devices. While signals being sent and received between base stations is an obvious opportunity, surely the same goes for base stations and smartphones, or smartphone to smartphone?

“These are a little bit more advanced than the near term deployment scenarios that operators would be looking at, but certainly on the roadmap for the moment because that needs new devices, new hardware,” Mourad explained.

He also pointed out that device makers including Apple and Samsung are engaged with the ISG because of this.

Some people might have flashbacks to the end of the film The Dark Knight, where Batman deploys a controversial technology to turn all the mobile phones in Gotham City into radars – something considered too intrusive for even Batman to use safely. Although this is fiction, this kind of situation is exactly why ETSI is involving privacy as one of their main work items. How would anyone know if their phone radio was becoming a sensing device? We can cover up a camera, there are indications if other sensors are on… but what about this?

“Whenever I make a presentation, I start with that and people freak out,” Mourad nodded.

“There’s much more actually that the network operators could do without us knowing and that’s why we have the security, the privacy aspects. I think they definitely need to look at those.”

There is one very interesting option which might assist in managing privacy, and this lies in the methods to reduce the data burden from the sensing. Rather than simply having all the sensing data going back to a central node, processing could be performed on the device which simply picks out what is relevant – as Mourad described it, ‘the sensing result’ rather than the sensing data.

If I am looking in a cupboard for a bottle of ketchup, I don’t need to know where the tuna or the cereal are. All I need in that moment is to find that one ingredient. In the same way, sensing should have an objective.

“Typically the sensing result is light from the data point of view,” Mourad confirmed.

“The sensor data is obviously huge, but not all of it is useful. So definitely that’s going to be a consideration, you know – compressing the data, preprocessing the data, extracting different sensing results, different nodes analysing that data, exposing that sensing result or sensing data to which entity, which node, which device.”

This will be a topic not only in the ISG, of course, but 3GPP and many other environments. It will be fascinating to see how legislation such as GDPR adapts (or not) to this – are previous scandals over listening from Samsung smart TVs or Amazon’s Alexa about to be repeated? Building safeguards into the process will be… interesting to keep an eye on, as the interplay between different stakeholders shapes up.

Orange CTIO Bruno Zerbib commented recently that they would not be marketing 6G, given that capabilities would hopefully be added much more incrementally. With ISAC, it looks as though this may well be the case.