Ethernet in the fronthaul opening up 5G potential
Research into ways to extend intelligence and monitoring capabilities into fronthaul connectivity may be a key enabler for developing new 5G network architectures, according to Kent University’s Nathan Gomes.
Kent University has been researching the use of Ethernet for fronthaul, and the challenges associated with that, as lead member of the iCIRRUS project, which is part of the EU’s Horizon2020 programme. Gomes’ team is also investigating where to site intelligence in the network – whether to have smart or dumb Remote Radio Heads, for instance – in order to best optimise signalling and control in C-RAN architectures.
Coordination technologies such as CoMP, which is defined for 4G, already require very tight synchronisation between remote radioheads (RRHs). Those requirements will only increase with 5G’s greater throughput (and lower latency) requirements. That could require a more intelligent and controllable fronthaul network – meaning that the iCIRRUS work could go forward to enable the dense network aspect of 5G.
Other connectivity technologies, such as Device-to-Device and “Clone-to-Clone”, could also benefit from this increased understanding of how to monitor and synchronise fronthaul connectivitiy. A key aspect here will be understanding signal and control flows to optimally route traffic between the “cellular” layer and Device-to-Device mode, or even between cloud-based “clones” of devices.
Gomes takes Inside5G through the background to the project:
“The iCIRRUS project came about from discussions we had had, first with JDSU, looking at putting intelligence and monitoring into the front-haul. JDSU had been doing this with mobile networks in the back-haul but we had particular interest in the front-haul. And then we thought that if we needed to do that we needed a transport facility in the front-haul which could be monitored – so something like an Ethernet network in the front-haul. That led to other talks about why you would use Ethernet and the advantages of using it in the front-haul.
We thought of using all of the aspects that could be gained by intelligently monitoring a fronthaul
“The monitoring side led us to think about how it could link in to other works which were being done at Kent on device-to-device communication and how that could alleviate some of the load upon the front-haul, to offload some of the traffic. We spread the ideas out from there and gathered other partners in who it turned out were also thinking about Ethernet use in the front-haul for different reasons, and were thinking about things like offloading not just to device-to-device communication but actually offloading it to clone-to-clone; so cloning devices in the cloud and moving traffic between clones in the cloud rather than physically between devices. So we thought of using all of the aspects that could be gained by intelligently monitoring a fronthaul for this device-to-device and clone-to-clone communication as well.
This is where iCIRRUS came from.”
However, using Ethernet in the fronthaul is challenging. Gomes says “at least half of the project is looking at the challenges” whilst the rest of it is looking at the advantages of what can be gained for things like device-to-device and clone-to-clone communications.
Ethernet fronthaul challenges
“Once you try to have statistical multiplexing you’re then inviting a number of challenges and potential problems because the fronthaul is traditionally for transporting radio signals and those require continuous transport: if you do statistical multiplexing then you have timing issues which don’t lend themselves very well to transporting continuous radio signals.”
Gomes says this means having to look at different subdivisions of the functions of base stations and the remote radio-head, something that is already being discussed in various standards and industry bodies.
“Even if you do the different subdivision it doesn’t get over the issue of timing and synchronisation for sending signals to radio-heads and so we have to look at how to use the transporters, not just of signals, but also of frequency and time synchronisation information.
“With a different subdivision we also have to look at how to support functions such as co-ordinated multipoint which is defined for use in 4G – but we’ll be seeing that this is very important for 5G for providing very high throughput. This will require very tight synchronisation between different radio-heads and therefore places a lot of demand on the front-haul. So providing for that very tight synchronisation will be a big challenge for fronthaul transport using ethernet .
“The other issues are how to use the monitoring of the network to provide for Device-2-Device communication, where the control of that should be in order to optimise the signalling and the offloading of the traffic from the network.”
The notion of dividing up functionality between base units and radio heads is part of research Kent is undertaking as part of another research project, known as RAPID. RAPID is focussed on dense user environments for 5G, and focusses on a different approach for the front-haul transmission.
Gomes says, “It’s looking at a much more centralised approach in which you have even more simplified radio-heads. In iCIRRUS we’re looking at a different subdivision of the functions between the central unit base station and the radio-head, where in fact we would be putting a few more functions into the radio-head than has been done up to now. That is to reduce bit-rate requirements in the front-haul.
“The RAPID project actually looks at simplifying the radio-head and moving more functions back into the base station unit. This is looking at a sort of analogue, radio-fibre approach for the front-haul and particularly concentrates on mmWave radio use, with beam-forming for the mmWave use and the use of analogue optical transport to try to maximise the use of the front-haul.”