Todd Mersch, Director of Software and Solutions at Radisys, is seeing LTE small cell technology becoming deployed on a wider scale. He reports back some of the key learning points from real world field experience, shares which small cell roadmap features are most sought after and believes the number of small cell vendors is likely to continue to increase.
Where are you seeing most activity in new small cell deployments today?
In addition to our normal day-to-day activities, we've been enabling a lot of new LTE deployments. We've talked a lot about both KT and SK Telecom (both South Korean operators), who have been moving forward quickly with their public access small cell deployments. Both are now moving into LTE residential small cell. This has been a major focus for our team, where we use the same software load across all our customers, allowing everyone to share the benefits of real-world operation.
We've had to address the same challenges learned from early 3G femtocell rollouts, such as zero touch configuration, remote management and being completely transparent to the end user.
From a 3G perspective, we see two different markets developing. Some operators are keen to discuss multimode – combined 3G/LTE/Wi-Fi small cells - but these are not necessarily near-term deployments outside NTT DoCoMo of Japan. We'll continue to have a strong 3G flavour in small cells for some time to come, because there is already a large installed customer base with 3G compatible handsets/devices.
We are also seeing a lot of activity in adjacent markets, such as in aircraft and for military use. These take advantage of the considerable existing investment in small cell technology, from chipsets through software and system design, to adapt to their specific needs. This brings considerable advantages in both cost and shorter development cycle time.
What lessons have you learnt from LTE small cell deployments to date?
Practical experience has shown that peak traffic levels, especially for signalling traffic, in LTE networks can be high. Active design work is required to ensure the LTE core network isn't flooded with sudden surges of transactions. There are several ways to deal with this.
One is to use an LTE femtocell gateway, which aggregates the signalling traffic and reduces the number of messages forwarded to the core network. The gateway can also be used to facilitate X.2 based handover between LTE cells, where there is not yet a signalling connection setup between the two basestations.
The other is to ensure there is enough headroom to handle peaks in capacity within the small cells themselves. So for example if a device is specified to handle up to 32 active users, we'd recommend ensuring you can allow more users to connect in a more limited way. We've reduced the memory overhead required to handle excess users and adapted to cope with sudden peak loads in a controlled manner. It can be fairly straightforward to deal with one additional user, say increasing from 32 to 33, when you can simply redirect the last customer to use a macrocell. Achieving this on a larger scale involves much more, and we've thought through and implemented a solution that solves that issue which we also bring to our other customers.
Where is the next hive of small cell activity?
Looking ahead, we see TD-LTE emerging as a more pressing requirement. We've had support for that for some time, and have integrated our software with several chipset vendors. We're seeing demand not only from China, but also from Softbank in Japan and Sprint in the US. The industry has speculated that once Softbank's investment in Sprint USA takes effect there is an opportunity for greater economies of scale through combination of buying power which could make them very influential operators in the TD-LTE market..
TD-LTE is somewhat different from the 3G TDD mode (TD-SCDMA), sharing a common software stack except for the low level Layer 1 and parts of Layer 2. From a deployment perspective, there is at least 80% re-use of our LTE software – we simply turn on the TDD variant within our existing stack. By contrast, 3G TD-SCDMA requires significantly more dedicated software.
China Mobile also continues to want to evolve TD-SCDMA which now has over 100 million active users. Today, Mindspeed is the only silicon vendor with support for both 3G TD-SCDMA and TD-LTE on the same chip and demonstrated this at MWC.
What is the most urgent feature requested for your roadmap?
We have been seeing a lot of requirement for Carrier Aggregation*, both for 3G and LTE. If there is one feature from the LTE Advanced release that is consistently being asked for, it's CA. There are various possible combinations, and we're not seeing this just at the high end. So not just 2x20MHz LTE carriers = 40MHz, but also combining carriers from both low and high frequency bands to achieve a fatter pipe.
We've also witnessed CA being chosen to circumvent interference concerns, delivering high capacity from a small cell using the higher frequency spectrum within the area of a higher capacity macrocell. A practical constraint nowadays is the backhaul fibre connectivity between the base stations themselves – not a problem in China, Korea or Japan but may be an issue elsewhere.
In the end, the race for CA is more about achieving headline data rates and faster speeds rather than system efficiency. It allows operators to combine their spectrum assets more effectively to establish competitive advantage.
A lot of these LTE radio interface innovations have also been added to the 3G specification, although they are constrained by the 5MHz channel bandwidth of 3G compared to 10 or 20MHz bandwidth of LTE. For example, DC-HSPA (Dual Carrier High Speed Packet Access) is a popular 3G feature which is similar to Carrier Aggregation for LTE achieving peak speeds of up to 42Mbps or more.
This extends the life of the 3G network but does still continue to require further investment. In situations where regulators continue to allocate dedicated 3G spectrum or where operators can't refarm their existing spectrum across to LTE, then this provides a good evolution path for those who can't move to LTE as quickly as others.
*[Carrier Aggregation allows two separate carrier frequencies to be used to carry data traffic in parallel, achieving much higher data rates overall. The frequencies don't have to be adjacent or even in the same frequency band – for example it's possible to combine one carrier at 700MHz with another at 2.6GHz].
Have you seen SON (Self Optimising Networks) evolving from existing small cell solutions?
The industry is adopting SON more extensively, and is evolving towards a hybrid SON technique. This combines a centralised SON server with more dynamic/real-time SON features on the small cells themselves.
For example, SK Telecom use their own proprietary SON server which connects through standard TR.069 management interfaces of the equipment vendors. It uploads performance data and downloads configuration parameters, such as setting maximum RF power levels etc. across both macrocell and small cell networks.
Radisys software handles the dynamic, autonomous changes within each small cell based on real-world conditions. The central SON server may set maximum limits or other constraints within which the small cell software will adapt as required.
And finally, are there still any new small cell vendors coming onto the scene, or is the market saturated already?
We are still finding new customers who are looking to build their own small cell products. China is specifically a growing opportunity and will have the same supply chain structure as in Korea, sourced from both foreign and domestic vendors – not just Huawei and ZTE. We've also seen more activity in the larger Tier 2 vendors, the type of company similar to Fujitsu or Hitachi, who want to resell their own box rather than one from someone else.
Cisco's recent acquisition of Ubiquisys brings more credibility to the small cell market, ultimately causing more pull through for our software. Existing 3G or LTE only customers are looking to expand into multimode products. Taiwan also continues to be a very good market for us – there is still a desire from operators that in the long term they can disaggregate the supply chain and take advantage of cost optimisation.