To meet mobile data demands in the second half of the decade, operators look to build two layers, with different base station types
By Caroline Gabriel
There is little room left for architectural debates in the wireless industry - in 2014, the answer is clear. The capacity requirements for wireless networks will be so great that there will be no room for either/or. Operators will be deploying combinations of many approaches on their cell sites, and in addition, making other, more fundamental shifts. Software will become as great a contributor to network capacity as additional hardware, via optimization and smart provisioning tools, while virtualization will be an important way to add density when hardware is reaching its limits.
Mobile data usage will rise by an estimated 12.5 times between 2013 and 2019, from a level Cisco calculated at 1.5 exabytes in 2013. These daunting capacity, coverage and data rate challenges mean traditional approaches will no longer suffice, and over 90% of operators have at least one brand new architecture in their five-year plan, while over one-third plan to use all the key new methods being studied in a new report from Maravedis-Rethink - small cell HetNet, virtualized RAN, new-wave distributed antenna and SuperMIMO - by 2019, somewhere in their systems. The report, Towards the hyper-dense network - the shape of the HetNet 2013-2019, surveyed about 150 mobile operators and found that capex in the period will not be driven primarily by conventional equipment, though the rise in volume roll-outs of LTE worldwide will be an important factor. Instead, there will be a significant shift in operator budgets towards new HetNet platforms, driven by advanced software, virtualization and new access points.
Overall, the operators' response to the capacity challenge is to adopt two layers (at least) of cells, macro for coverage and true mobility, and small cell for capacity and indoor services. A third layer may evolve at the end of the decade for the internet of things (IoT). The two layers underpin a typical three-stage 4G deployment, which most operators plan to follow - 1) coverage first, macro layer-only; 2) quick-fix capacity, often via WiFi or opportunistic small cells; 3) greater virtualization in parallel with a move to hyper-density.
In both the main layers, three architectures can be used, and will often be mixed and matched or even integrated, and run in unlicensed as well as licensed spectrum.
The three architectures are:
- Homogeneous - self-contained macro or small cell base stations, with the baseband, antenna and radio in the same box or located at close quarters.
- Distributed radio or virtualized RAN - where multiple radio/antenna units share a virtualized pool of baseband processing, which may be located on a nearby base station hotel or in the cloud. This evolves to Cloud-RAN.
- Distributed antenna or DAS - where a baseband/RF unit feeds a number of antenna nodes distributed around a building or neighbourhood. Traditionally a large venue solution, smaller and more open approaches are being developed to target small cells.
The next wave of macro innovation is focused on virtualization, and antennas, with technologies such as massive MIMO and Active Antenna System (AAS). In addition, the macro layer will be boosted by the introduction of some key features of LTE-Advanced, such as carrier aggregation, eICIC and CoMP, which will make new dense architectures easier to roll out. These add up to a new wave often called 'Super Macro'.
The number of LTE or multimode macro base stations deployed will peak in 2015, about a year later than we had previously envisaged - continued growth will be driven by the rising interest in macro-first enhancements as well as emerging market roll-outs, which will usually focus on coverage first. After 2015, there will be a decline in the number of macro sites built out for LTE, and over the whole period, there will be a compound annual decline of 4% for base stations, though only 1.8% for macro sites overall because of the shift to software upgrades and virtualization.
Traditional homogeneous macro stations will be a tiny fraction of the market by 2019, while remote radio heads will feature in 25% of deployments still, just ahead of C-RAN. The other architecture to have a significant impact by then will be SuperMacro.
Although there will be a steady shift towards SingleRAN strategies (replacing legacy kit with new, flexible, multimode base stations), rather than overlays, there will be limited addition of brand new sites in the macro layer. Upgrading or overlaying existing sites will be the dominant approach to LTE build-out. All this activity will create an installed base of macro sites that will still top 5m in 2019, with the largest bases in Asia-Pacific and Europe, though there will also have been extensive decommissioning of sites because of new architectures, a shift of attention to the small cell under-layer, and RAN sharing.
Meanwhile, large-scale deployments of public access small cells are still in their infancy, but there is already talk of 'hyper-dense' networks to cope with hotspots of intense data usage. In total, the number of public access small cell sites - sites deployed in a separate layer (distinct spectrum from the macro layer and closer to the ground), and less than 200 meters in radius - will reach an installed base total of over 15m by 2019. These sites will be equipped with a variety of technologies including metrocells, WiFi, DAS and virtualized small cells. This is no longer a market which is only about one technology - self-contained metrocells - but about a combination of options to create data density where required. We have reduced our forecast for metrocells somewhat, but see increased addressable market for complementary technologies, driven by new types of operators including 'WiFi-first' players.
One of the significant shifts in the pattern of deployment, which we noted in our 2013 reports, has been intensifying operator focus on indoor deployment, often at the expense of outdoor. Outdoor dense roll-outs, especially of metrocells, have been pushed back by at least two years on average, while indoor build-out plans have remained unchanged or even been accelerated. The tipping point will be in 2016, when for the first time outdoor cells will be more than one-third of total deployments.
In regional terms, the small cells space is dominated by Asia-Pacific throughout the decade and this is one reason for the steadily increasing importance of TDD spectrum, as several Asian players, notably China Mobile, are TDD-led. However, most MNOs will deploy unpaired spectrum as a capacity option from year three or four of their LTE program - and it will often be focused on small cells, for which the short range of the primary TDD bands, 2.3GHz, 2.5/2.6GHz and 3.5GHz are well targeted.
The report models four alternative scenarios with different balances between the four equipment types, based on the range of possibilities which operators are outlining for their future roll-outs, and a series of variables including product delays/accelerations, emergence of standards, and cost patterns.