Bill Hogg, AT&T SVP for Network Planning and Engineering, presented at last month's Small Cell Americas conference. We reported some of the highlights in our event report, but thought that this keynote justified more coverage. With AT&T aiming to deploy 40,000 non-residential Small Cells before the end of next year, many of its tried and trusted methods will need to change. He explained where Small Cell solutions do and don't fit as well as some of the radical changes being implemented throughout the country.
We've recapped and summarised his conference presentation below as accurately as we can. Any misrepresentations or errors in our reporting are unintentional.
One size doesn't fit all
Small cells aren't yet appropriate for all sizes of buildings. For example, DAS (Distributed Antenna Systems) are more cost effective for 30 storey buildings whereas a 4-5 storey building or a campus environment would be a better fit for Small Cells. Today's 10-15 storey crossover point between Small Cells and DAS will change and evolve to become 20-25 storey once the ecosystem of Small Cells gains scale and drives across the industry.
A range of Small Cell products are required for different applications. Residential units are adequate if they can handle 4-5 users. Higher capacity metrocells supporting 32 concurrent users today will evolve into multi-standard products supporting up to 64 concurrent users using 3G/LTE and Carrier Wi-Fi.
I like the option of having Wi-Fi built into the Small Cells as a single physical package which could share the same LAN building infrastructure. While not suggesting this would be a good choice in every case, in certain applications it is a good combination to drive both cellular and unlicenced capabilities together. There is a sense of which tools (i.e. radio technologies) can be applied to different situations.
AT&T have already deployed Small Cells in a wide variety of different applications: everywhere from the top of telegraph poles to inside all sorts of buildings. A good example of the challenges we face is a case study from Disney theme parks. We deployed macrocells, 25 DAS systems and 350 Small Cells across several major Disney properties to address their needs.
Identifying where to install
We realised we'd need a whole new set of planning tools to achieve the flexibility and speed of deployment that Small Cells require. This starts with RF analysis, taking different pieces of information and identifying where best to locate Small Cells. One presentation format for this type of data is a heat map of usage that is geo-located.
We also look at fast growing areas of the network and at where RF code/power are being exhausted.
Inventories of light poles and other potential physical locations provide a list of suitable candidate sites. We can then piece this data together to identify good locations including backhaul options.
Can't place an order
Previously, our backhaul transmission systems were designed to connect to large buildings or macrocell sites. We found we couldn't place an order for backhaul to a specific light pole, only to a known US address with a ZIP postal code. We needed to redesign and retool our backhaul ordering process to be able to accommodate that.
We also needed a new workflow engine that takes care of all of the key steps throughout the implementation cycle, from planning through to deployment.
Our planning process also needs to take local factors into account. Rules about zoning are unique to each municipality and it can take anything from days to years to get approval.
We also need a good control plane management – understanding how to manage the Small Cell alongside the Macrocell layer, including handoffs and driving the traffic between the layers. This requires a level of complexity that is still immature today. SON (Self Organising Network) capabilities help achieve this but a there remains a lot of work across the ecosystem to be done.
All of these are elements for tooling are needed to achieve our objective of 10,000s of (non residential) units, meaning all of these aspects have to be addressed in a scalable way. We took a clean sheet of paper approach and didn't try to adapt existing traditional methods.
- Data traffic growth drives spectral efficiency, extra spectrum and densification
- Network densification drives more macrocells, DAS and Small Cells
- More Small Cells drives new tools, new processes and cost effective backhaul
Perhaps smaller operators elsewhere may not need such sophisticated retooling, but those looking to deploy 10,000s rather than a few thousand will need to adopt a similar approach.
Questions from the audience
How accurate are the current geo-location RF tools?
X and Y resolution is getting better – we can usually determine the individual buildings but not always. Z (height) geo-location is non-existent.
Do you use different KPIs for Small Cells vs Macrocells?
We need careful layer management in boundary conditions, especially where the dominant macro competing with Small Cells can lead to dropped calls. But in a lot of areas we can manage that if engineered correctly. Overall, the accessibility improves at the risk of dropped calls on the cell edge.
How much backhaul capacity do you need for a Small Cell?
Typically 10 to 20 Mbps is adequate today. A 50Mbps Ethernet circuit may be overkill for a Small Cell except if it is daisy chained. We use a whole multitude of solutions from NLoS/LoS microwave, DSL, Cable etc. The main challenge is the price point – our internal price points are an aggressive sub $200/month. Managing the QoS and peak transmission rates are a challenge but we have seen quite a bit of success. We try to use wireless backhaul where possible, but where Cable is available they are also aggressive in this space. We've made good progress with commitments for performance and QoS targets.
What is the target cycle time to deploy a non-residential Small Cell?
Overall, our target is 75 days from identifying a candidate site to being fully commissioned. For indoor Small Cells, we can and have achieved 30 days or better. In a few extreme places with heavy zoning (city planning) constraints, you might be looking at anything up to 3 years.
From a customer perspective, I believe they would be pleased with that kind of turnaround [Ed Note: 30 to 75 days, not 3 years!].
This article was based on a public presentation by Bill Hogg at Small Cells Americas December 2013 and is not endorsed or vetted by AT&T. We have made our best effort to faithfully summarise and recap what was said. Any misrepresentation or errors are unintentional.