WiFi Versus Cellular Carriers – Revolution At Our Doorstep

The questions about spectrum holdings of public carriers may, in fact, be of less importance in the future than they have been in the past.

The big issue for the future of mobile broadband and the mobile cloud used to be whether the carriers would get enough spectrum to handle swelling demand. That issue – and the power structure of the mobile broadband industry – is changing.

This change is occurring largely because of: 1) the priorities of device makers, who are going to design devices to use every bit of spectrum resource available to them; 2) the accelerating development of WiFi, which is now progressing at a rate exceeding advances in licensed spectrum technology – even when we take into account not only 4G, but the gleam in the industry’s eye which is “5G”; 3) the availability of other spectrum alternatives for smart devices, in addition to UMTS licensed spectrum and WiFi.

Today we think about spectrum in three broad categories.

A) UMTS: 3G – 4G (LTE) – LTE Advanced

B) Wi-Fi: primarily 802.11a, b, g, n, 802.11 ac and 802.11ad (WiGig)

C) Other spectrum alternatives: licensed and unlicensed

What we are interested in is how each category will evolve to meet the stunning Deluge of mobile broadband traffic which is overwhelming the industry. While Cisco has lulled the industry with its VNI forecast, as we have written in our “Mobile Traffic Deluge” study, the demand for mobile broadband capacity exceeds that forecast by several multiples.

The wireless links essential to the mobile cloud and the vast majority of the M2M future are undergoing a dramatic series of capability advancements. Not surprisingly, they are interrelated, with developments in silicon. Silicon and RAN are advancing together.

The performance of the RAN is vital for the future of the mobile cloud, since the RAN is the link between the device(s) and the cloud(s). This is the connecting link between the three elements in the Von Neumann computing model – Processing, Storage, Input/Output. The more robust these links, in term of data transfer rate, latency, availability, reliability and ubiquity, the better can the apps perform.

Until recently UMTS has completely dominated the mobile spectrum picture. The story of UMTS is the story of licensed spectrum and its use (application for commercial purposes) and the global progression of standard protocols and architecture – for radio access networks and core networks.

This involved an evolution over approximately 10-year timeframes from one global virtually universal radio access protocol to the next. The first 10 years were essentially analog. Then from 1990 to 2000 we saw the development of 2G and from 2000 to 2010 3G.

While 3G represented an increase from about 50 kbs to over 500 kbs, in downlink speeds, the 4G evolution, beginning with the Verizon launch of LTE, brought commercial broadband data service at an average of 5-to-12 mbs and latencies generally in the range of 50 to 100 ms (milliseconds).

Therefore within each decade UMTS was able to deliver about a 10 times expansion in radio access network throughput with technology that is commercially available and for which there are devices for the masses at reasonable prices.

Early trials of of LTE Advanced, beginning in the latter half of 2013 for commercial availability in 2017, expect data rates 25 to 50 times that of current LTE systems. (A word of caution: Just as LTE literature today can claim data transfer rates of 50 to 100 mbs and LTE Advanced as a gigabit RAN offering, it is important to translate the speeds from the lab to the realistic speeds in the marketplace.)

Despite this evolution, we are at a point where virtually everybody recognizes that the carrier networks are going to be unable to service even the modest demand growth forecast by parties such as Cisco.

In a simpler world, where carriers dominated, the answer would be, well, simple. Carriers would modernize their networks on a schedule that suited their required financial returns, and users would wait.

The world is no longer that simple. As we’ve explained in our studies over recent years, the development of mobile and of the mobile cloud is being driven by a changing “power picture” in which device, apps and alternative infrastructure companies – basically a web-based “gang” – is driving relentless advances in mobile broadband data and in cloud computing. These companies, led by Google, Apple, Facebook, Amazon, Microsoft are gaining momentum and seizing power from the carriers.

Starting with the acceleration in the expansion of WiFi in the past three years, it is necessary to consider how the position of the UMTS carriers versus alternatives is changing. The questions about spectrum holdings of public carriers may, in fact, be of less importance in the future than they have been in the past. This is not a suggestion to slow down the release of more spectrum to carriers but it is a perspective of understanding the issues and solutions before us in permitting the full development of the mobile cloud.

When we look at spectrum both licensed and unlicensed, from the standpoint of various industry players, the following key conclusion emerges.

For some industry players, the spectrum future looks quite promising, while for others there is less of a rosy picture.

From the vantage of a device maker, the unlicensed picture is one of an enormous range of free spectrum assets identified in table 1 (the Wi-Fi perspective) – up to 250 MHz of highly contiguous spectrum assets, in the US. If one includes the assets available for 802.11 ac (which operates at 5 GHz) and 802.11 ad (at 60 GHz) then we would have almost 2,400 MHz of spectrum available for these extremely cost-effective and high throughput 802.11 devices and networks. (Tables are included at the end of this article.)

With respect to UMTS spectrum, the issue is not the gross amount of spectrum – the three major U.S. carriers hold about 360 MHz of total licensed spectrum – but the fact that the spectrum is balkanized, chopped up among licensee awards for different generations of technology, and in many cases with geographical limitations.

This malady is identified in table 2, which lists for the U.S. the spectrum allocations, permitted usage and channel sizes. The total amount offered is over 500 megahertz but the ownership rules, use limitations and noncontiguous nature create a labyrinth like maze for a device or app developer to work their way through. In addition each spectrum sliver represents a part of the total device market and in light of roaming limitations both technical and commercial, greatly limits the devices’ addressable market.

Thus the critical realization, is that LTE Advanced and LTE have spectrum assets in North America which range from, effectively, at most, 80 MHz down to 20 MHz. It is clear that Verizon’s early 2000 purchase, in the 700 MHz range of a 22 MHz nationwide license was crucial to enable its first LTE launch

The forthcoming LTE Advanced standard states that operators should look for 50+50 MHz of paired spectrum and ideally it should be contiguous. (Although the standard does make provision for combinations of adjacent spectrums to reach as broad a set of megahertz assets as possible). In many cases licensed spectrum operators must deal with LTE implementations with 20 MHz or less. Therefore the shortage of spectrum assets, especially contiguous ones, would lead any carrier globally to press strongly for more spectrum.

This is true even though these spectrum assets may be aided by MiMo, QAM (quadrature amplitude modulation) advances and spectrum bonding, (common to all spectrum bands); these techniques are expensive to implement.

In other words as you look at these two tables, the players, i.e., device makers or network operators, can look at this as a feast or famine and these two perspectives are coexisting at the same time.

Clearly for those who believe that licensed spectrum is critical to their business model, the path to more business from the mobile cloud is to get more spectrum and therefore more productive capacity. In the case of a device maker or a cloud/platform provider you will really want to optimize your ability to use both licensed and unlicensed and maximizing your utilization of unlicensed spectrum is a no-brainer.

No discussion of a device maker’s perspective would be complete, without recognizing that from a device standpoint not only is the availability of a UMTS network a simple and on the chip solution as is WiFi, but the device also has Bluetooth – and with 802.11 ah (very short range) and NFC, additional 2.4 spectrum bands are available.

WiFi and carrier network spectrum capacity are not by any means an apples-to-apples comparison, especially when we consider overall coverage. In the WiFi area, however, over 400,000 hot spots have arisen, unrestrained by carrier capex limitations, or zoning or permitting issues. These hot spots do carry a lot of traffic and play an ever-increasing role in device-to-cloud and all data traffic movement.

The recent announcements of the Wi-Fi Alliance and the cable industry members led by Comcast and Time Warner of programs for hotspot 2.0 implementation would indicate that there will be at least several million additional hotspots in service with both public and private SSIDs (service set identifications) by 2015. In addition these players have high quality, well-engineered backhaul capability.

In addition to the growing WiFi component is the startling rise of the “personal hotspot”. We estimate that there are more than 54 million smartphones with the capability of acting as a personal hotspot in the United States as of now. Zero to 54 million in less than 8 years.

It is easy to project that there will be over 100 million personal hotspots, essentially of the iOS and Android nature, available in the United States by 2017. This is clearly a different topological picture; as a matter fact this is a different set of network issues then the standard cellular topological model of UMTS. This growth is a classic pandemic model.

In addition to creating a forecaster’s nightmare, all of these maxi and mini RANS need some type of backhaul and will tend to increase the efficient use of all backhaul assets from terrestrial to spectrum utilization. All links will be more heavily loaded than before.

Therefore it is a distinct possibility that the focus we’ve all had on the UMTS and the 3GPP (3rd Generation Partnership Project) architectures may be of lesser importance concerning what happens to the mobile cloud and data transport dependencies. The UMTS networks may very well be the backup, rural connectors, highways, between centers, rather then being the most important places of RF access or perhaps even aggregation.

Once the hotspot at a personal level is incorporated into the national topology, the advances in mesh networking may have a major effect on several aspects of the mobile cloud.

The cloud, in many situations, may consist of just a device level locus of computing assets, which includes other devices, in a mesh network formation. An interesting example was offered in the recent patent filing by Facebook. It is clear that device to device communications, based on user profile selections, is possible and when coupled with a mesh network management system, one can:

1. Optimize use of spectrum resources by: mapping spectrum assets currently available to an application’s requirements. This mapping is likely to be best done in a cloud controller type of architecture. The cloud would have the best overview of the RAN ecosystem.

2. Select the best protocols for distance (proximity) and service, etc.

3. Enable applicable security layers

4. Synchronize as appropriate

5. Invoke collaboration tools as applicable

6. Use intelligent assistants and their cloud resources

The above list is quite extensible. Clearly the best area of residence for such a network management system is in the cloud, incorporating the device.

This then results in the situation where the over the top concept is really a question of how your personal cloud can communicate to someone else’s mobile personal cloud, in many circumstances without using any other intermediary transport organization like a UMTS carrier or even a Wi-Fi service provider.

The implications for the mobile cloud is not just in faster downloads and better apps processing, but the world of big data processing and sophisticated assistance is at an individual’s disposal.

All in all, a revolution in the RAN area is at our doorstep.

Table 1. WiFi Bands, U.S.

Frequency Range (MHz) Bandwidth (MHz) Protocol Speed* (Mbps) Comment
2400-2500 100 802.11b,g,n 54-150 channel size up to 22MHz
5725-5875 150 802.11 n/ac up to 600 channel size up to 40MHz
470-710 white spaces 802.11af <10 white space – SDR1 needed (cognitive)
60 GHz 2,160 802.11ad (WiGig) >1000 short range gigabit RAN
* Note: MiMo configuration can increase these rates by factors of up to 6X
2402-2480 79 spread spectrum, freq’cy hopping – e.g. Bluetooth 4-Jan low power

1 SDR – software defined radio

Table 2. UMTS Bands, U.S.

Frequency Range (MHz) Bandwidth (MHz) Band Designation Channel Scheme Notes
698-806 108 700 3G/4G Blocks A-E; upper C block – 22MHz
800 LTE 18 + 18 2 Blocks A, B
850 3G 25 + 25
PCS 3G/4G 60 + 60 6 Blocks
45 AWS 3G/4G A, B, C, Blocks
2496-2690 194 BRS 4G A-F (10 + 30 MHz)

Notes to Table 2: Total carrier spectrum – about 553 MHz
1. BRS: Clearwire (Sprint) has about 120 MHz nationwide
2. AWS. Being used by VZ for 4G expansion and T-Mobile for LTE launch
3. The 553 MHz is largely controlled by an oligopoly

Company Approximate national holdings (MHz)
Sprint/Clearwire 184
Verizon 100
AT&T  74
Total 3 Companies 361