Hutchison Global Communications (HGC) will be using active optical equipment from Infinera to support a C-RAN deployment that it will be offering on a wholesale basis to Hong Kong’s mobile operators.
The first customer will be “sister” operator 3 Hong Kong (3HK), but the aim is to offer the C-RAN platform to other operators in the territory. Jon Baldry, Director of Metro Marketing, Infinera, said that HGC’s wholesale model influenced the technical selection it made for handling fronthaul transport.
These requirements led to the operator choosing an active transponder so that multiple CPRI channels can be sent over the same number of wavelengths. In this instance Infinera offers six transponder functions on a card, deployed in a 1u chassis – so enabling a single 6-sector site or 2×3-sector sites from a single blade. Baldry said the deployment was “good verification of our performance – being able support all CPRI Specification v6.1 rates – including going to 12 Gigabit.” The company also offers passive, semi-passive optical and active muxponder products in its fronthaul portfolio.
Although the active transponder solution fits HGC’s business model, for C-RAN deployments where operators own all their own infrastructure many are likely to use passive optical solutions, usually equipped with some additional monitoring capabilities. This reduces latency across the system.
For example, Ulrich Kohn of rival vendor ADVA Optical, told TMN that his team decided that passive transport of CPRI interfaces with a specially developed monitoring capability allowed the most efficient way to transport CPRI signals and monitor access needs. That’s because ADVA understood fronthaul to be a very cost-sensitive market.
Active solutions offering the electrical aggregation of CPRI – with demaraction at the RRH and BBU ends of the RAN link – also “burn latency”, Kohn’s colleague Antony Magee added. However he did add that if you were looking to monetise the link, such as in a wholesale model, then active deployment would be an option, with “every vendor” having an active model of some sort available.
Infinera’s Baldry acknowledged that the active electronics introduce delay into the system. Infinera says that its active Muxponder (multiple signal conversion onto a single wavelength) option increases latency by the equivalent of 1.5km of fibre – which Baldry said is an acceptable loss where LTE’s latency budget mandates a theoretical maximum 20km.
In other words, with an active Muxponder solution, an operator loses 1.5km of “reach” to meet its overall latency budget across the BBU and transport. For comparison, a transponder – a single signal to wavelength conversion such as that being used by HGC – introduces only 1m-worth of delay.
He also added that the drawback of passive systems is that it may require converting the RRH itself. “If you want a passive WDM with coloured optics plugged into the BBU and RRH, some vendors support that and some don’t.”
“So even though active solutions cost more it avoids changing out optics in RRHs that are already deployed. If you are an MNO building our your own network with 1-2 vendors that doesn;t matter, but if you are operating a wholesale model you have to be able to go where the customers go, and that means having a RAN-neutral solution.”
“Adding in a transponder as a demarcation point in the network enables them to have this RAN neutral solution. It also gives you more capabilities in building and managing the network – one of the key things here is getting rid of stuff at the cell site – if you are adding stuff back in again you have to be very focussed on space and power.”
Another technical aspect that Infinera will be providing is delay compensation.
“Remote Radio Heads (RRHs) from different vendors all respond differently to delay. Some RAN vendors can cope well if the latency changes,” Baldry said, “while for others if the latency changes then the network drops.” To get around this, Infinera is introducing delay compensation – effectively making all connections run at the latency of the slowest – so that if a connection on a ring based topology is switched to a longer path, the latency observed by the RRH doesn’t change.
Is C-RAN about to have its moment? The market is still out. HGC has a huge amount of fibre deployed in Hong Kong – both in fibre to the building and some FTTH. That is clearly a key enabler, and it is mirrored in other high profile C-RAN deployments in South Korea and China. But what of other markets?
Baldry said that he has felt that 2015-16 will be the years operators start to evaluate and test, with only early adopters starting to deploy. Largescale deployments will come from 2017 onwards.
ADVA’s Magee said that away from fibre-rich markets, operators are still “gathering evidence” on C-RAN, including looking at CPRI transport. But there is a parallel development in exploring fronthaul over the Ethernet layer, and that is more in the 3-5 year window, he said. That goes along with 5G timescales, which in turn rest on virtualisation of the RAN – which moves the C in C-RAN from Centralised to Cloud.
For Magee, RAN virtualisation will start with cell site gateways (routers), with the next step bringing the physical BBU into the central office where the CSG is now. “The only real de facto standard to do that [now] is CPRI,” he added. But Magee doubted that CPRI can scale for 5G – “We will not have the vBBU until fronthaul can be supported by Ethernet,” he said.
Baldry said that the immediate driver for C-RAN deployments is the tighter sync requirements of LTE-A features such as CoMP and eICIC, with 5G requirements adding to that later.