Your browser is not supported. Please update it.

9 March 2015

The 3.5GHz band: a far better place for small cells than 5GHz

Some WiFi providers may already complain that 5GHz will be congested before we know it, but to cellular operators it looks like a lush piece of high capacity, relatively untouched spectrum. The main way for an MNO to harness that has been to integrate WiFi more tightly into the cellular network so that it can be controlled in the same way as 3G/4G base stations – and its users tracked and monetized accordingly. Seamless hand-off and automatic authentication, using technologies like the WiFi Alliance’s Passpoint, have helped with this, as have tools to converge cellular and WiFi in the small cell, the packet core or the BSS/OSS.

However, the problem is that WiFi, and the 5GHz band, are now being aggressively used by non-mobile operators, notably cable providers and telcos, to launch WiFi-first or WiFi-only wireless services and quad plays. These threaten to reduce the revenues cellcos can get from MVNO deals with the wireline players as well as from end users. One solution, of course, will be to make their own services far more attractive than those of the MSOs, in terms of cost and performance.

That could be by adopting WiFi-first themselves, as Sprint and others have, and balancing usage efficiently and dynamically between cellular and WLAN. Or it could be by making LTE cheaper to use, which would be helped if the economics of unlicensed spectrum could be introduced to 4G – not by adding WiFi to the HetNet but by running LTE in 5GHz itself. That would have the dual attraction, for some operators, of reducing the overall cost of delivering 4G capacity, and of squeezing the spectrum available to WiFi.

Both approaches were on display at Mobile World Congress. The talk has gone well beyond simple WiFi data offload and is now focused on using WiFi more creatively as part of a HetNet, as seen in Alcatel-Lucent’s launch of its Wireless Unified Network (WUN), and Qualcomm’s demonstration of WiFi aggregated to an LTE system.

Meanwhile, there was also considerable buzz around LTE-LAA (Licensed Assisted Access), the technology which allows LTE to run in 5GHz and provide a supplemental downlink carrier for a licensed-band 4G network. Also heavily backed by Qualcomm – as well as the big mobile OEMs and some operators like T-Mobile and China Mobile – LAA will not be standardized until next year’s 3GPP Release 13, but is likely to turn up, in pre-standard form, before that. There were many demonstrations and talks in Barcelona, aiming to calm fears that LTE-LAA would be a difficult neighbour for WiFi.

Qualcomm and others were also talking about LWA (LTE WiFi Access) which supports WiFi link aggregation. An LTE data session is augmented by extra traffic sent over standard WiFi, which can be aggregated in the core. Like LAA, this should be included in the 2016 3GPP Release. Similar cellular/WiFi aggregation approaches could look ahead to 5G by using higher bands and the WiGig technology in 60GHz. This was discussed by SK Telecom, which has trialled a system to augment an LTE data session with another wireless technology running in a wide band at frequencies up to 100GHz.

As the industry rushes to address issues of interference, coexistence and device support for LAA – not to mention the political backlash from Silicon Valley at the prospect of 5GHz being used for a technology that is accessible only to LTE licence holders – there is a far less controversial option to increase capacity for small cells at affordable cost. That option is the 3.5GHz band. Like 5GHz, its high frequency means it is only really suited to small cell architectures, and it could enable a near-global band to support mass market device economics and global roaming.

It is already largely earmarked for wireless broadband (though in many countries mobility needs to be added to the licence terms), and it has the protection that licensed spectrum provides for LTE quality of service and security, while generally being sold at low cost compared to lower frequency bands. Many countries are likely to follow the lead of the US in allowing the plentiful capacity at 3.5GHz to be shared among multiple operators, a move which could make this spectrum a valuable proving ground for ideas of dynamic spectrum access and flexible sharing schemes.

Those who believe the LTE industry should prioritize its efforts on 3.5GHz, not 5GHz – like ip.access CTO Nick Johnson, argue that 3.5GHz could be a specific band for small cells, optimized to deliver a high end LTE user experience. In 5GHz, by contrast, there is the risk that “we go to the lowest common denominator in order to avoid interference” and so dilute some of the advantages that Qualcomm and co are claiming for LAA over WiFi.

Qualcomm itself is open to any band which strengthens LTE’s hand against WiFi. Its CTO Matt Grob noted in an interview that, while 3GPP efforts so far had focused on 5GHz, LTE-LAA was designed to be band-neutral and could be deployed itself in 3.5GHz, allowing the small cells to provide supplemental capacity for a parent macro network. And Verizon referred to LTE-LAA plans both in 5GHz and 3.5GHz.

Momentum is growing behind the latter among non-US regulators too. Last year, Japan’s regulator allocated 120MHz in the band to the three main mobile operators for LTE-Advanced small cells, which are likely to be deployed from 2016. Now one of those, Softbank, has demonstrated IP-CoMP (coordinated multipoint) technology in a technical trial with small cell provider Airspan. The test involved a cluster of 3.5GHz TDD small cells sharing a single PCI and operating in joint transmission mode to improve coverage and capacity. Unlike Cloud-RAN, this approach can use Ethernet backhaul to link the small cell cluster to the macro station, and so does not require fiber.

Meanwhile, Ericsson has completed 3.5GHz TD-LTE trials with Australia’s National Broadband Network company, using devices from NetComm running on Sequans chips. In many parts of the world, there are legacy WiMAX or proprietary BWA networks running in this band, which could be repurposed over time for TD-LTE small cells to add capacity to a parent 4G network.

The leading role has certainly been played by the US, where the FCC wants to redesignate 3.5GHz as the ‘Citizens Broadband Radio Service’ (CBRS). The regulator has proposed a three-tiered spectrum sharing system in this band, with different levels of access priority for federal incumbents, licensed non-federal users and general access users.

The US Navy, which has radar systems in this spectrum, would be the top-tier licensee. The second tier would be for priority access licensees (PALs), probably wireless carriers. PAL users will be able to obtain one-year licences for 10MHz of spectrum as long as they guarantee not to interfere with the Navy users (they do not get protection from government users in return). The third tier is for general authorized access (GAA) users, which could be MNOs, enterprises or unlicensed technology users. They have the same rights and obligations as PAL users in the second tier. Equipment that covers one part of the band has to work in all parts of the band for all users, so carrier equipment would have to work for GAA users.

Here, the LTE-LAA situation in 5GHz could be reversed, with Google seeing opportunities for a new implementation of WiFi here. Overall, this is one of the most interesting of the emerging spectrum options – one where licensed and unlicensed technologies may coexist; traditional and non-traditional service models develop; and where the nascent spectrum sharing ideas, which most agree are vital to avoid future shortages, can be tested in relatively uncluttered frequencies.

Google supports the FCC’s plan to have a centralized spectrum manager, or Spectrum Access System (SAS), to assign channels dynamically and support dynamic aggregation to assign the channels in real time. This kind of model has been championed by the search giant for years, to encourage more flexible spectrum allocation. It argues this would support large numbers of small service providers, even those which need bandwidth only at certain times of day, for instance, and so could not afford a full MVNO deal – and it would also provide better QoS levels for certain sensitive applications than a fully unlicensed free-for-all band. Carriers like AT&T have supported SAS as a way to manage the different tiers of licensees, but oppose the dynamic aspects, which would start to break down the habit of long term, static spectrum allocations on which the power of the cellcos rests.

The big question in both the bands at issue is whether technology neutrality – in modern times enshrined in most regulatory thinking – is really such a good thing, or whether dedicated bands, while less democratic, may be necessary to allow technologies to perform at their best and so deliver the speeds and QoS that new services demand.

In a world where LTE and WiFi stay in their separate bands, convergence at network, device or chip level becomes important, and this was the main thrust of Alcatel-Lucent’s big MWC announcements. In its Wireless Unified Network (WUN), LTE and WiFi radios are both operational at once, with data signals from the two types of access points being blended to leverage the upload and download capabilities of each, simultaneously. The first step is to launch ‘WiFi Boost’, which combines the downlink of WiFi with the uplink of cellular, and ‘WiFi Boost Plus’ which also combines cellular downlink. In future iterations, LTE-LAA will be included in the mix, and ALU has already introduced small cells with this capability, but it sees WiFi Boost as a shorter term option to achieve the best of both worlds.

This is likely to be the commercialization of a technology ALU has already implemented for major US operators, reportedly Comcast being among them. Its system can identify devices from either network and knows what the user’s access rights are on LTE and WiFi.

The goal is to boost speeds by harnessing WiFi capacity, though not, in this case, via LAA. It demonstrated data rates up to 600Mbps at MWC, with some of the data travelling over 4G and some over WiFi. In typical homes Alcatel says this WiFi boost can increase download speeds up to 70% and increase upload speeds by an order of magnitude or more, compared to standalone WiFi at the cell edge.

WiFi offload until now has relied on a policy decision being made for each activity, and then on ANDSF to select the right network for the right activity. This system is still policy-driven, but uses both networks at full speed, rather than offloading to one.

This technology could put ALU in a position to facilitate cellco networks making deals with large owners of WiFi, for instance Comcast in the US and Liberty Global in Europe. We presume that a deal between a cable operator and a cellular operator using this system, could mean that cellular minutes are offered at the WiFi cell edge to speed up downloads, and WiFi could be offered to the cellular network, essentially meaning that it becomes a two-way data exchange. In previous WIFi offload models, all the payment and data use has been one way, with WiFi helping the cellular network and not getting anything in return.

Mike Schabel, VP of small cells at the firm, said: “As an industry we have historically treated cellular and WiFi as unique technologies. For example, the industry debates the merits of voice/data over WiFi versus voice/data over LTE. From a consumer perspective, it should simply be just wireless.”

Alcatel-Lucent will trial the WiFi boost capabilities in the second quarter of 2015 with commercial availability in the second half of 2015.

Cisco is also making a bid to straddle the cellular/WiFi divide, and has come out with an analytics system it calls Mobility IQ, which can provide a cloud-based software as a service, to give performance views across LTE, 3G and WiFi simultaneously.

Mobility IQ is supposed to give service providers real-time visibility into network, user, and business intelligence across Wi-Fi, 3G, and LTE networks. The issue is that data becomes fragmented across technology silos, and network owners have limited visibility into indoor and business networks, where their customers most frequently consume mobile services.

By providing a single, dynamic picture of the health of the entire network and offering APIs with which to take action, Mobility IQ will help optimize performance and identify problem to provide that all important thing a satisfied customer, one of the biggest weapons in a cellco’s arsenal.