There are various ways of increasing the peak data rate of a mobile network. A popular feature of LTE Advanced is Carrier Aggregation which bonds two or more separate carriers to double or triple peak transmission speeds. Here, we look at how best these might be implemented, together with various alternatives and associated trade-offs and how this relates to/complements Small Cell deployment.
What is Carrier Aggregation?
Combining parallel transmission paths to achieve a higher total throughput.
This has been done in 3G, where DC-HSPA (Dual Channel High Speed Packet Access) offers peak rates of up to 42Mbps using two separate 5MHz carriers. This is fairly common and widely supported by many smartphones today. In theory, up to four independent carriers could be combined to produce even higher data rates but few operators would have that much spectrum to spare.
3GPP has defined this as an LTE Advanced feature, introduced in their standards Release 10. It's one of the most popular LTE-A features requested by operators, allowing them to quote higher peak rates in the marketing programs and achieve additional network effiency.
Doubling the size of carrier frequency bands
Most LTE networks today use FDD with uplink and downlink each allocated 10MHz of bandwidth. This theoretically can achieve 70Mbps or so of total peak throughput, but this drops to perhaps 20Mbps when loaded and shared between many subscribers. Additional capacity can be added by installing extra transceivers at busy cellsites, and transmitting on separate carrier frequencies, but each smartphone would normally communicate only using one carrier.
Some operators enjoy the luxury of sufficient LTE spectrum that allows them to double peak throughput to each smartphone by allocating wider carrier frequency bands. For example, EE in the UK have deployed 20+20MHz FDD to double their throughput. There are additional efficiencies by simply widening the band, with less co-ordination and control required to manage two separate parallel channels.
This doesn't preclude also using Carrier Aggregation, and indeed EE have been trialling 4x speeds by using CA with 20+20MHz bands to achieve up to 300Mbps.
Spectrum scarcity drives faster adoption of CA
Many operators, such as those in South Korea and US, don't have the luxury of such extensive adjacent spectrum and so have been pushed into using CA more quickly than elsewhere.
It's possible to assemble a faster connection by using more than two different carriers. The current LTE-A standard permits up to 5 carriers which could use 100MHz bandwidth to deliver up to 750Mbps. The bandwidths of each segment don't have to match.
For example, in South Korea SK Telecom has split its 35MHz of 1800MHz spectrum using 20MHz down and 15MHz up to achieve 150Mbps today. It will then combine this with a 10MHz band at 800MHz to achieve speeds of 225Mbps.
SK Telecom also demonstrated 450Mbps rates using 3x 20MHz carriers at Mobile World Congress. Competitor LG Uplus is also reported to plan three band carrier aggregation service this year, split across 800MHz, 1800MHz and 2600MHz and potentially achieving 300Mbps.
In the US, AT&T deployed CA last month but few if any handsets are available that support it. Once more widespread, it will help compete in regions where it lacks the large blocks of contiguous spectrum that Verizon and Sprint enjoy.
One of the potential issues with combining signals at significantly different frequencies is that they may have quite different propagation characteristics. Lower RF frequencies may penetrate inside buildings or achieve longer range, while factors such as inter-cell interference and multi-path affect each each carrier frequency differently. Compare it to traffic all moving at a similar speed along a multi-lane freeway versus each taking diverse routes along local roads.
Delivered from the same large or small Cell
Every operator I've spoken to about their CA plans has told me they intend to introduce this feature with the total traffic handled by the same macrocell site or small cell. This avoids potential latency issues when serving the same data via multiple cellsites as well as making it easier to control and manage.
A more advanced CA technique called Multiflow allows data channels to be shared across multiple cells, either between Small/Large in different layers or between neighbouring ones. The CoMP (Co-ordinated Multi-path) LTE feature also allows that, but does require very tight co-ordination between a group of cellsites – effectively meaning dark fibre to all, and ideally a common co-located point for baseband processing. This requires more of a Cloud RAN based architecture rather than Small Cells, where the baseband processing is contained within the cellsites themselves and places less rigorous demands on the backhaul connection.
Single carrier Small Cells may outperform dual carrier macrocells
But let's not get too carried away with theoretical peak rates. Where Small Cells are deployed to serve a hotspot with relatively few users very close by, there will often be much better RF transmission path and dedicated capacity available to each smartphone. This higher quality RF channel can use higher levels of modulation, deal with lower levels of interference and may be able to deliver higher data rates than a congested macrocell serving larger numbers of users across a cluttered urban environment.
So in my view, it's not a priority that Small Cells implement Carrier Aggregation immediately. Features such as tri-mode (3G/LTE/Wi-Fi) and enhanced SON would seem to me to have more short term priority for product roadmaps.
LTE-U and LTE-DSA
Several vendors have been promoting the use of LTE in the unlicenced (Wi-Fi) frequency bands, such as at 5GHz. The proposal is to retain a "master channel" in the licenced mobile spectrum, adding extra LTE carriers using Wi-Fi spectrum where necessary to achieve higher speeds. This is termed Supplemental Downlink (SDL) and would use Carrier Aggregation to add on extra capacity on demand, where any additional throughput is considered a bonus. There would be no spectrum cost involved, but also no guarantee of protection from interference in that band.
As Frank Rayal points out in this article, Wi-Fi technology uses a "listen before talk" approach (known as Carrier Sense Multiple Access or CSMA) whereas LTE transmissions are masterminded by the network. This technical issue plagued the use of WiMAX in unlicenced bands and would need to be overcome.
He also mentions LTE-DSA (LTE in Dynamic Shared Access spectrum) which could be used at 3.5GHz, for which the FCC is only just beginning to define the rules.
Aggregating with Wi-Fi
Could Carrier Aggregation be used with Wi-Fi? I saw this demonstrated by InterDigital as an Over the Top service some years ago at Mobile World Congress. They used a smartphone App to drive multiple connections using both cellular and Wi-Fi, recombining the data traffic in a server in the Cloud. Their 2012 white paper details the history of combining the two technologies.
Today, the Wi-Fi data path uses a different IP address and different route to connect to the Internet from any cellular data connection. It would be necessary either to recombine those in the mobile network, embedding the capability seamlessly within the smartphone or to install an App which handles this externally. Few apps can cope today with switching seamlessly between cellular and Wi-Fi. Web browsers may be able to (although it would break any secure SSL sessions); streaming audio applications typically can't. Skype is one of few that can cope with either or both.
Looking further ahead – combining LTE FDD and TDD together
Another combination that may easily be overlooked is the ability to combine both FDD and TDD carriers. While today's LTE networks are almost exclusively FDD or TDD (Softbank in Japan is perhaps a good example of one using both), the value of available spectrum is too precious to waste in the long term. In Korea, both KT and SKT have demonstrated TDD-FDD CA using NSN equipment.
Unlike 3G, where TD-SCDMA was radically different from the mainstream FDD mode, LTE has very close alignment between FDD and TDD modes. Both are embedded in many smartphone chipsets and the large potential markets of China, India and Japan will ensure continued availability and investment.
Those operators with both FDD and TDD assets may well choose to adopt CA, perhaps with FDD as the main "master channel" and gaining extra performance by switching on additional TD-LTE carrier when needing higher speeds. This would be an ideal capability to implement within a Small Cell, especially once LTE becomes more established.
Carrier Aggregation (CA) implements channel bonding between two or more different frequencies to achieve higher peak datarates. These can be adjacent spectrum bands but often involve different frequencies, which may result in different performance characteristics for each.
Most operators will look to deploy CA where both carriers are co-located in the same cell. This will initially be in macrocells, and later in Small Cells.
CA could be used to combine LTE operating in unlicenced spectrum (such as Wi-Fi) if the regulator permits it. Today's LTE-U proposals position this as bonus supplementary capacity with the primary connection retained on licenced spectrum.
It would be more difficult though not impossible to combine LTE with Wi-Fi, with each data path taking quite different routes through the network and involving multiple IP addresses.