You may think that Urban Small Cells are only for those ultra-busy downtown city districts, and that green, leafy suburbs would continue to be served by traditional macrocell towers with a few residential/in-building femtocells thrown in.
If so, you need to rethink that perception.
David Orloff, AT&T's Director for Small Cells Platforms, speaking at Small Cell World Summit, explained three use cases where AT&T has deployed groups of small cells in the suburbs. We followed up directly with him with some questions to clarify some of the issues they faced and problems solved.
AT&T is in the vanguard of small cell adoption, visibly chairing the Small Cell Forum, publicly promoting their use at events and even in TV advertisements, but most demonstrably through aggressive deployment. Over 50% of its network densification program will use small cells by 2015.
When will Urban Small Cells become a reality on our street lampposts?
We see that's happening already. There are many outdoor locations that are difficult to get site permission for larger sites, or which don't justify the full cost of a macro. In many cases – including school areas, suburban, residential – we are actively installing urban small cells now.
In the picture on the right, you can see a typical street deployment on a light pole/lamp post. The bottom-most white box is the DC power cabinet including backhaul and surge protection. The box below the street light is the small cell itself. The white box in the middle is not related to the small cell installation.
How do you determine where to deploy small cells?
Our customers would like to have good service everywhere, so we engineer small cells for customer experience. We focus on a set of KPIs which will be familiar to many network engineers: coverage, capacity, voice retainability, packet accessibility, throughput and mobility. I look at these figures every day.
We've developed analysis tools that take several inputs from RF design, social data, transport data and performance data to pinpoint where and when to deploy small cells efficiently. This allows us to prioritise and plan our rollout both strategically and tactically.
Can you describe some typical suburban small cell deployments?
There's no single use case. Here are three different examples and some of the issues we found:
1) A residential neighbourhood at the edge of macrocell coverage
We installed 15 3G metrocells in a residential neighbourhood, located at the edge of multiple macrocells. This configuration dramatically improved coverage throughout the neighbourhood.
There were a limited number of entrance and egress points to which we paid particular attention. We also closely monitored how these edge-point small cells interact with the macrocells, and made some optimisations to deal with the overlap to avoid dropped calls.
2) A mixed suburban environment with a raised suburban road running through the middle, with dense foliage throughout
We deployed 22 small cells in this cluster to great effect. At first, we found that the signal level dropped off fairly quickly between the macro and small cell clusters. The relatively high speed of the traffic on the highway created a challenge when interacting with the small cell layer.
Specifically, the handover success rate from small cell to macro wasn't as good as we expected. When we dug deeper, we found the handover transition wasn't fast enough due to terrain and foliage. So we optimised the handover by adjusting signal thresholds and quality, adding a couple of blind handovers (i.e. adding extra cells into the neighbor lists), and this gave us an immediate improvement in call retainability. However, I would caution that blind handover really only works best when you have very predictable target cells to handover into.
We also minimized the inter-layer transitions, avoiding unnecessary handovers between the macrocells and small cells cluster. Minimizing the transition in this way keeps the traffic on the preferred layer and simplifies the neighbor lists.
3) A residential environment with difficult terrain
We installed 13 metrocells in an area of poor macro coverage and saw an instant improvement in customer experience. This scenario involved a higher number of small cells in the cluster that actively interact with the macrocell layer. It's not about the total number of cell sites, just those that handover to and from the small cells.
Once optimised, we saw significant improvement in call retainability.
How much is SON (Self Organising Networks) being used to simplify the configuration process?
SON is already extremely valuable with small cell implementations during both initial deployment as well as ongoing operation. The Centralized Auto PSC PCI and Distributed (Node Level) Auto PSC PCI functions allow for small cell clusters to initialize and maintain an interference optimized environment.
We also heavily use Auto Neighbor Relation (ANR) functions to maintain interoperability and reciprocal neighbor relation management with the Macro environment.
Any final thoughts?
50% of AT&T densification efforts will use small cells. We are in an aggressive deployment mode right now. If we continue to engineer our network based on customers' experience, then customers will have the service they need and the quality they expect.