The Small Cell World Summit (SCWS) in London is a highlight of this sector’s calendar, and last week it celebrated its tenth annual event. The show has mirrored the history of small cell networks to date, which has been the usual mixture, in the mobile world, of innovation, disruption, inflated expectations and frustrations. But in 2016, the broader mobile platform has caught up with the small cell pioneers – with densification and in-building networks at the heart of many operators’ investment plans, the miniaturized base stations have crossed into the mainstream, with significant implications for operators’ approach to deployment and financing; and for the incumbent wireless vendors.
So how far has this industry segment, and the Small Cell Forum which supports and promotes it, achieved the goals of 2007? In one way, the remit is almost unrecognizable. In the early conferences – which were called Femtocell Summit then – the focus of the vendors and the Forum was on residential access points and getting good quality cellular links into the home WiFi-style, with simple, affordable, plug-and-play units backhauled by a carrier’s broadband line.
In 2008, at the second event, the highlights were CDMA-based deployment plans by Sprint; the first low cost system-on-chip for a femtocell (from picoChip, a pioneer in this sector, now part of Intel); and the start of talk about integration with the home network, and about presence-aware applications which would add value to the basic proposition of coverage and capacity beyond the reach of the macro network.
Some of these themes recur today, though the focus is not just on the residential market, but also on three other broad target sectors as identified by the Forum and its working groups – enterprise (including public-facing locations like venues), urban (public outdoor networks) and rural/remote (which also includes temporary or emergency networks). Applications to harness features inherent in small cells, such as location and presence awareness and personalized services, remain an important area of work, especially in the enterprise. The Forum has worked on a set of open APIs (application programming interfaces) to encourage development of apps which would enrich the value proposition for small cell deployers and encourage uptake.
And integration with other networks, though not just in the home, is a constant theme. Back in 2007, it was hard to avoid questions about how small cells would compete with WiFi, or be made redundant by it. In 2016, that issue is even more pertinent, as WiFi continues to extend the functions it supports, often encroaching on cellular territory with developments like IMS-based Voice over WiFi. However, the debate has also become far more nuanced. Small cells certainly have to take account of WiFi’s incumbency indoors but their supporters are also taking a keen interest in LTE’s move into unlicensed spectrum – at the high frequencies of the 5 GHz band, this is entirely a small cell play because of the power and propagation limitations in that spectrum. And approaches to coexistence with WiFi have evolved from multimode access points (now common in the enterprise) and even chipsets, to a 5G-style movement towards virtualized platforms that support many air interfaces, with seamless movement between the connections and intelligent load balancing and capacity management.
Many of these issues were scarcely envisaged in 2007 – they have emerged as operators have started to define what ‘heterogeneous network’ will really mean in real world deployment, and as the industry starts to work out how that HetNet of 2016 will be able to evolve smoothly into 5G, without the need for a big bang upgrade any time soon.
All that was far over the horizon in 2007, but there were still significant challenges in convincing the industry that the femtocell would enhance the operator business case and become an important part of the mobile network. Now, the residential small cell has achieved the large scale which is essential to the economics of any platform that relies on low cost hardware and standardized interfaces. According to the latest Market Status report from Mobile Experts, almost 13m residential small cells have shipped to date. Now, annual shipments in this sector have almost plateaued, and significant growth lies elsewhere, particularly in the enterprise market, which is nearing the 1bn cumulative installations mark.
However, we should not forget the important foundation stones which were laid by the work on the residential femtocell, most of which are relevant to every target market. In particular, the idea of a cellular base station with standardized interfaces, supporting multivendor interoperability in the same way as WiFi access points, has been highly disruptive to the mobile ecosystem. The definition of the Iuh interface for those early femtocells was
From those starting points, the industry developed and changed rapidly. There have been delays and disappointed expectations – early forecasts of imminent mass-scale deployment underestimated the challenges of finding sites and backhaul, and driving down the cost of the equipment to the point where hundreds of cells could be rolled out cost-effectively. LTE support was too slow in coming, as was multi-operator support (still an issue).
There was resistance from some areas to the idea of a new ecosystem structure, more akin to that of the WiFi or PC/IP industry than cellular – there were echoes of WiMAX’s similar but failed attempt to democratize the platform with open device ecosystems, multivendor testing and plugfests, standard interfaces and commoditized access points.
And as with any new technology, every time one challenge was met, the world had shifted again and new ones had emerged. The more operators moved towards truly dense networks, the more they needed automation, through techniques like SON (self-optimizing networks – see separate item), to make it practical to plan and manage so many cells. The industry has raced to keep up with these changing demands, especially as densification is starting to happen in high capacity locations such as sports grounds and in in-building sites, and will spread into the entire public network as carriers implement the dense HetNet.
So if the topics of discussion in 2008 were applications, interference risks to the macro, and the magical sub-$100 femtocell, by 2009 the hot issues had already changed. In Wireless Watch in that year, the conference write-up highlighted the first public tests of Iuh, and the need for services and interoperability, which emerged as more challenging for operators than interference. A year later, Vodafone stole the limelight with its residential femtocell launch in the UK and this sector was on the brink of scale, while at the 2011 conference, the home market had reached its tipping point, and the key discussions had moved on to metrocells and outdoor public network deployments.
In 2012, the event, and the Forum, replaced Femtocell with Small Cell in their titles, to reflect the widening scope of the industry, both in terms of target markets and services, and the type of base station which might be supported. Increasingly, the definition has expanded from a self-contained, low power access point with baseband, radio and antenna all included, to recognize any form factor which addresses a very small cell of coverage, which might be a distributed radio (as in the virtualized controller-based indoor architectures of CommScope, Spidercloud, Parallel and others); a distributed antenna system (DAS) driven by a small cell; a WiFi/cellular device; or a distributed indoor system driven from the macro, like Ericsson Radio Dot and Huawei LampSite.
At the 2012 event, Rethink revealed the results of an MNO survey, which identified five key questions which operators were asking themselves before deploying small cells beyond the home. These still resonate today:
- Should the indoor market be the priority, delivering better short term revenues and return on investment?
Though there had been high interest in metrocells, many MNOs were recognizing that most mobile data is consumed indoors, and that the greatest challenges to users’ quality of experience were usually in-building. That conclusion, and the rising important of QoE to reduce churn and improve ARPU, have driven the sharp rise in enterprise small cell investments since last year, with the Forum declaring, in 2015, that it had “crossed the chasm” into mass-scale deployment in this market.
- Should operators turn to neutral host or shared multi-operator systems to ease the economic argument for small cells, especially in the enterprise, where customers of all MNO networks would be using the system?
This remains one of the significant brakes on small cell progress. Although multi-operator technologies exist in the shape of 3GPP’s MOCN and MORAN, neither has been deployed to any large extent. Operators remain nervous of “enabling their competitors” by sharing, yet for many enterprises, multi-operator services are a deal-breaker. The next big hope is that virtualized platforms, evolving towards the 5G concept of ‘network slicing’ (in which slices of capacity are allocated dynamically and on-demand to different service providers) will address the business case and technical barriers to this important issue.
- Will SON deliver?
Operators recognized that dense small cell HetNets would only be practical if many of the planning, management and optimization processes were automated, to reduce operating costs and support near-real time response to network issues. SON has certainly delivered in technology terms, with significant advances in its capabilities. Early deployments have mainly focused on automatic neighbour relations (ANR) but many other functions are on operator roadmaps. SON will be the centrepiece of the Small Cell Forum’s next plugfest, at Telecom Italia’s labs in Naples, Italy this fall; and the Forum’s significant new document, outlining an Integrated HetNet Framework Architecture, details how SON fits into that architecture, for current 4G HetNets and future 5G networks.
- Multivendor or end-to-end?
This is another thorny question. One of the reasons that some MNOs were enthusiastic about femtocells and standardized interfaces from the start was the promise of a shake-up of their supply chains and procurement processes, allowing them to mix and match access points, driving down prices and loosening their ties to the traditional macro vendors. In reality, the more complex the HetNet becomes, the more some operators retreat into the perceived safety and integration of an end-to-end solution – though some trends, notably the shift in focus towards indoor networks, have enabled non-traditional vendors to start offering their own end-to-end solutions. Cisco is a notable example, building on its enterprise dominance in wireline and WiFi to become an indoor HetNet provider (helped by its acquisitions of Ubiquisys, Meraki and Intucell).
The multivendor dream is still there, and is supported by the Forum’s continuing work on standardized interfaces, including FAPI; by interoperability testing initiatives; and by the advances of independent vendors which are producing compelling solutions even without macro integration. Spidercloud, Parallel Wireless, IP.access, CommScope/Airvana and Airspan all have large operator deployments of their small cell platforms independent of the macro supplier, though there is a potential new lock-in, in the network controller. However, a few MNOs are deploying HetNets with small cells from more than one vendor, including Sprint and China Mobile.
Virtualization and SDN also have multivendor networks as a key goal for many operators, and these are emerging in parallel with dense small cells, as a foundational ingredient of the true HetNet and of 5G. Some of the vendors highlighted above already use some degree of virtualization, splitting the access points so that control of a whole network of small cells is centralized in a single controller, on-premise or in the cloud. These vRANs will become open and multivendor once interfaces for the split access points are defined, and that is the goal of the Forum’s nFAPI development, which lies at the heart of its HetNet 2020 roadmap and evolving architecture.
- Should intelligence be concentrated at the edge or centralized?
The biggest change in the mobile landscape is that the network is becoming an IT applications platform and MNOs, if they are to avoid being relegated to bitpipes, are turning into cloud service providers. Those services platforms – as well as the key functions of the network itself, as they slowly move towards Cloud-RAN – will logically sit in a centralized cloud platform. But that raises huge issues of latency and overburdening the network, if every network event or user data packet has to travel to the core and the cloud. In 2012, operators were already talking about placing network intelligence, and end user applications, at the edge of the network to improve responsiveness and quality of experience. Even before that, Ubiquisys and Intel had cooperated on an x86-based companion processor for a small cell, which could support data functions such as caching video close to the user. Then Nokia did something similar with its Liquid Apps platform, and Quortus showed how most of the functions of an evolved packet core (EPC) could be virtualized and run on a small cell or edge-based controller. In the enterprise controller architectures mentioned above, the central point for network functions and for applications is localized in the company itself (or its cloud), creating large numbers of mini-Cloud-RANs. Indeed, the move towards virtualizing the RAN will start, for most operators, in the small cell layer, which avoids the challenges of re-architecting the macro network too soon.
Since 2012, many of those issues about edge-based intelligence have gained new prominence as a result of ETSI’s Mobile Edge Computing (MEC) initiative. This is inherently a small cell issue, since the most logical physical place to support MEC services is via a highly distributed network of end points with high capacity connectivity to controllers and the cloud. This trend will greatly enhance the value of small cells to their operators, and many will take on processing and memory capabilities to support content caching; location-aware network optimization and analytics; and end user data applications, increasingly driven by context awareness. These functions will give the operator enhanced ability to monetize the users (and, in the IoT, the connected ‘things’) and the data they produce.
So the 2016 Small Cells World Summit saw this technology acquiring a highly significant role in operators’ network and business plans, as the foundation stone of the HetNet and of important trends like MEC and RAN virtualization. The Forum unveiled the latest instalment in its Release Programme, a large collection of technical and business documents designed as a blueprint for operators, to accelerate deployment by reducing risk and providing best practice. The latest episode, Release 7, concerns HetNet and SON, and the path to 5G – hefty enough topics to merit a two-phase release. The first documents, including the central HetNet architecture framework and new operator and ecosystem surveys, were published last week, and the second phase will follow next month at Mobile World Congress Shanghai.
The Forum’s work program, and that of the wider industry, will be focused on addressing the technical and commercial challenges of turning that HetNet framework into profitable reality in the near term. It has identified the key challenges as:
- Multivendor interoperability.
- SON evolution.
- SON benchmarking.
- Backhaul SON.
- HetNet capacity planning.
- NFV implications.
- Unlicensed spectrum in the HetNet.
- Indoor location.
- Energy savings.
These are far wider in scope, and arguably more challenging, than the issues facing the fledgling ecosystem in 2007. But much of the work done in the femtocell days has enabled operators to think in terms of open, multivendor, IT-enabled HetNets today, and to start rolling out base stations on a huge scale, especially indoors. For instance, Nokia claimed last week that it had installed the world’s largest indoor small cell network, a 3G/LTE FlexiZone system which replaced a DAS. This was for Telefonica Chile in the Costanera shopping center in Chilean capital Santiago. The mall has six floors and 168,000 square meters of retail space, and a reinforced concrete structure, but its 110,000 shoppers per day require deep in-building coverage. When the small cells industry gathers again in 2017, such projects should have become significantly easier to achieve, and more commonplace.
IP.access was particularly prominent in those conversations, and remains an innovator today when many of its competitors have been acquired (Ubiquisys by Cisco, Airvana by CommScope). IP.access was a very early player in small cells, with a well-established GSM product, and has been a regular winner of Small Cell Forum awards, started in 2009 to recognize commercial success and innovation in the sector. In 2016, the company’s new Viper platform won the Judges’ Choice award for a product which has the potential to push the boundaries of small cells in terms of functionality and market impact (and two ip.access executives, co-founder Nick Johnson and Neil Piercy, won special awards for contributions to the Forum).
In the next edition of Wireless Watch, we will cover further aspects of the summit and key small cell trends, especially Mobile Edge Computing. We will also provide key highlights from our latest operator survey on small cell deployment plans to 2020.