Residential and Enterprise femtocells are normally connected by traditional broadband internet service, using DSL, cable modem or enterprise-grade broadband. Public access small cells, on the other hand, are usually installed and managed by the network operator themselves. This means they have to design and deploy a backhaul network suitable for the task. It must be high enough capacity to meet the demands of multi-mode 3G/LTE small cells, yet cost effective to be viable. As if that wasn't enough, it must not constrain the location of the small cells and be easy to install by non-specialist staff.
The potential lower utilisation of small cells
As a very general statement, there tends to be better utilisation of facilities which are shared across a wider mix of users. In a macrocell, the large coverage area might serve residential, business and entertainment areas which smooth the type of usage throughout the day. As the cell coverage area reduces, it becomes inevitable that usage becomes more specific – catering for a specific venue that may only be used for certain periods of the day, or for certain types of user.
This results in the peak-to-mean ratio being higher – potentially driving the cost of providing peak capacity higher.
Consider a cluster of small cells covering the same area as a macrocell. In the simplest case, if all the users moved from one side of the coverage area to the other, the macrocell could reallocate its full resource to deliver peak capacity to either location. A small cell grid covering the same region would potentially need to offer that peak capacity in both regions by installing more total capacity. Fortunately, the low cost of small cell radio capacity more than offsets that issue, but it increases the importance of locating the small cells precisely where the traffic demand is greatest.
Backhaul is a substantial cost in the end to end solution
Backhaul is a significant cost of the total solution. If the backhaul infrastructure can somehow be shared, effectively reducing the peak-to-mean ratio by dynamically assigning more capacity to those small cells as they need it, then this may improve financial viability.
Future required backhaul data rates are substantial. By contrast, one public Wi-Fi operator quotes average backhaul use of around 2Mbps, with peak of 8Mbps per hotspot – even though the Wi-Fi radio link can operate at higher speeds. 3G and LTE cellular networks are unlikely to want to limit their small cells to that extent – specifying 42 Mbps 3G and 100Mbps+ LTE co-hosted on the same box. It raises the question of whether to cater for peak instantaneous demand, or design for more typical mean traffic levels.
Is fibre an option?
The highest traffic areas tend to be those with the highest population density – the city centres, financial/business districts with tower blocks, stadiums and other venues where crowds congregate.
A few countries do seem to have fibre available almost everywhere and at an economic price.
Having fibre nearby, say within 100 meters, isn't good enough – the benefit of a small cell may be reduced or made worthless if it had to be located that far away.
Many wireline network operators are reported to be backing away from installing fibre directly to the premises, instead investing in Vectored DSL and bonding which runs at up to 100Mbit/s over copper pairs from the neighbourhood cabinet. This still requires physical copper wires to be run to the small cell device.
Cable broadband (DOCSIS) may also be an option in some areas, especially North America.
Wireless transmission for the last mile simplifies installation
Although it depends on the specific situation, a common public access small cell backhaul architecture would use wireless for the last link to a local hub (which might be an existing cellsite). From there, higher capacity point-to-point microwave or fibre would connect back into the core network.
With large numbers of small cells to deploy, the short installation times needed from less skilled staff drive the need for a simple and adaptable solution. The added flexibility to position and locate the small cells precisely where the coverage is required is also a significant benefit to the whole cost equation.
Non-Line-of-Sight (NLoS) vs Line of Sight
Basically, higher RF frequencies don't penetrate obstacles or go round corners. That makes them more suitable for satellite links and point-to-point microwave where there is a direct and unobstructed pathway. Lower frequencies, say less than 5GHz (some prefer 3.5GHz or less) are much more useful because they can be used for NLoS links. As a result, this spectrum is much more in demand and costly.
Some operators have experimented using unlicensed spectrum at 2.4GHz and 5GHz, but there is concern that unpredictable/uncontrolled competing use may dramatically affect capacity and the quality of service delivered.
Many small cells are necessarily being located in areas of poor coverage, street canyons or indoor public areas which macrocells can't easily serve. These locations are less likely to be suitable for point-to-point wireless links, giving NLoS products an advantage. However the limited amount of spectrum will limit the widespread deployment of this approach especially in areas of highest traffic demand.
Where used, NLoS should help reduce overall costs because installation of equipment won't require alignment of the antenna, or revisits for adjustments in later life. This means less time from lower skilled field staff is needed.
Point-to-Point (P-P) vs Point-to-MultiPoint (P-MP)
A point-to-multipoint backhaul system shares capacity from a central hub between a cluster of small cells. This adapts to meet the instantaneous demand, so that if users in one or two small cells require most of the capacity then it can be directed there. If this load changes to other cells, then the backhaul capacity also moves. This can be compared to the footprint of a macrocell, allowing the total capacity to be served and shared across a wider range – offsetting the earlier concern about high peak-to-mean ratio of use made at the beginning of this article. The combination of antennas used in a P-MP system – one sectoral antenna at the ‘hub’ and one directional antenna at each remote terminal - simplifies the installation and ongoing maintenance.
Point-to-Point microwave and millimetre wave solutions also have a part to play. Relative to NLoS frequencies, there is significantly more spectrum available especially above 60GHz.
Large parts of many of today's backhaul networks are designed using a tree architecture, branching out to each cell individually. In some areas (e.g. long trunk roads), a chain of cellsites may be connected in series, but there is a limit to the number of hops that can be used.
Clusters of small cells lend themselves to other topologies, such as:
- Mesh, with several alternate routes. This provides redundancy and load sharing.
- Point-to-Multipoint. As above, shares available backhaul wireless capacity between a cluster
Other factors to consider
Backhaul designers have other concerns and constraints too. These include (but are not limited to):
- Security: Both physical (public access small cells are in unsupervised areas) and interference/jamming/interception
- Synchronisation and timing: Many macrocells today derive timing signals from the wireline backhaul connection.
- Physical design: weight, power, size, appearance etc.
- Multi-mode: Shared backhaul for 3G, LTE and carrier Wi-Fi
- Skill level and numbers/availability of technicians required to install and maintain
- Capacity planning and performance reporting tools to monitor and expand the system to match demand
Last but not least, the cost of the overall system must be viable. Many strategic network planners are calculating the costs of building out these small cell networks, for which the backhaul is a significant element.
This is a complex and costly aspect of future network design for which there is no simple and straightforward answers. It's not a static problem either - innovative wireless backhaul products are being introduced at low cost, spectrum availability and regulation may change, small cell product development continues apace.
The best answer will comprise a toolkit, selecting several of the options mentioned above. It will be important to revisit the assumptions made during the early planning cycle to confirm they remain valid and appropriate as the network evolves in the years ahead.
Small Cell Forum detailed technical white paper describing the benefits and drawbacks of each technology option.
My thanks to Dr John Naylon, CTO of Cambridge Broadband Networks Limited, for explaining the issues and technology available. CBNL has been heavily involved in writing the NGMN Small Cell Backhaul Requirements and you can view his presentation at Small Cell World Summit 2012 here
Also an earlier Small Cell Backhaul presentation from MWC 2012 can be found here