Cavium’s CEO reckons they’ve got silicon that covers all the architecture options for small cells and Cloud RAN. They’ve even proposed a new middle ground with smarter radio heads. He shares opinions on timescales, architectures and equipment vendors.
Are the huge market forecasts for traffic growth realistic?
The amount of data being consumed wirelessly today is becoming ridiculous, especially in Asia. I’ve even seen people watch entire movies on their mobile phones.
Wi-Fi will address a portion of this demand, but it has no handover or mobility and cannot easily be managed by the service providers. That’s good if you remain sitting in one place and don’t move around.
LTE has extremely good performance, is much better managed and has significantly lower latency and we’ve seen some consumers exhibit a strong preference for it. For example, a major European Service Provider used Wi-Fi hotspots in some locations and when they introduced LTE small cells found that data traffic shifted 85% from Wi-Fi to LTE.
LTE is being rolled out rapidly in advanced countries including US and China. India will start this year. At the same time, we still see a lot of 3G equipment being shipped.
Where is the industry going with regard to Small Cells?
The mobile industry is undergoing a densification of the network to satisfy growing traffic demand. Once macrocell capacity has been put in place, it is inevitable that small cells will follow in some form or other.
Small Cell rollout is probably a year or two behind where it should have been, and I think the latter part of 2015 will have significant uptake. There will be multiple trials this year followed by mass deployment during 2016 and 2017.
There will be a need for both indoor and outdoor small cells, with in-building adoption currently ahead of outdoor.
Which small cell architecture will win?
We propose three architectural alternatives for the densification of the mobile radio access network:
- Cloud RAN with remote radio heads connected using CPRI
- Cloud RAN with smart radio heads
- Fully independent small cells
We can provide silicon for all three alternatives. We expect to see a mixed proportion of all three, depending on regional and other factors.
Our proposal for smart radio heads involves adding a portion of time critical Layer 2 functionality to the normally dumb remote radio head, implementing more of the latency sensitive and high bandwidth functions locally. This allows us to remove most of the Layer 2 functions and virtualise these centrally in the cloud. A key benefit is that this smart radio head does not require CPRI, so doesn’t need fibre backhaul. The backhaul link requires around 200Mbps compared to 2Gbps for CPRI.
A few years ago, our customers [basestation vendors] said they wanted to develop and control their own physical layer (PHY). Today they don’t and just want to buy that in. Instead, they want to focus more on SON and other value added features. We provide a production quality L1 PHY and our own pre-integrated and pre-verified L2-L3 solution allowing us to support these different architectures.
Which equipment vendors will be the winners and losers in long term?
There are some six or seven main RAN equipment vendors today, who are present across all these scenarios. Nokia, Ericsson and Huawei are well established while new vendors such as Samsung and ZTE are growing fast.
Unlike traditional Macro RAN deployments , which are only supplied by OEMs the Small Cell will have a much more diverse supplier base which includes both OEMs and ODMs