page image

CHAPTER VI - Setting the Predicates: Communications Architecture Today and Tomorrow

The Roundtable began with a look at the characteristics of network architecture, both wired and wireless, that are relevant to the broader policy inquiry. This initial inquiry also addressed the goals of a robust communications system and, in particular, the ability and limits of wireless services to substitute for and/or complement wired in achieving those goals.

Contemporary wireless broadband infrastructure consists of licensed spectrum on 3G and 4G networks and unlicensed spectrum on Wi-Fi networks. In addition to the macrocells—wide-area, high-power base stations—that comprise the foundation of modern cellular networks, femtocells, picocells, Wi-Fi routers and other hardware expand and extend basic wireless connectivity. Additionally, wireless can include other spectrum such as TV whitespace if it is designated for such use.

These wireless broadband networks compete in many ways with wired broadband technologies. Contemporary wired broadband technology consists of Digital Subscriber Lines (DSL), hybrid fiber optic and coaxial networks using cable television and cable modems, and fiber optic to the home networks. With the advent of new satellite technology like ViaSat, satellite broadband is increasingly able to offer fixed wireless broadband connectivity to the home, as well.

Within wired infrastructure, there are significant differences that make different technologies more attractive than others along certain dimensions. Among other things, for instance:

  • Although DSL continues to advance and squeeze additional bitrates from this technology, it will likely always lag to some degree behind cable and fiber because of the physical limitations of the copper wire that transmits DSL signals.
  • While the last mile of the cable network is shared access (the cable connection comes to the neighborhood and provides access to all homes in a neighborhood), DSL offers dedicated bandwidth in the last mile. This can have implications for the effects of congestion.
  • Fiber optic cable offers greater bandwidth, but it is more costly initially.

Differences Among Wired and Wireless Options

There are fundamental differences between wired and wireless related to capacity, topology, reliability, mobility and economic investment. The table below, presented to the group by Jon Peha, summarizes these differences:


Wired

Wireless

Capacity

Abundant

Limited by Spectrum

Topology

Point-to-point

Point-to-multipoint

Reliability

Reliable

Less Reliable

Mobility

Fixed

Mobile

Economics

Fiber is the Highest Capital Investment

High Capital Investment

Figure 1: Fundamental Differences Between Wired and Wireless

Importantly, wireless is capacity constrained by spectrum availability and the fact that signals decay faster through the air than they decay through copper wire. In the long term, while there may be advances in wireless technology, capacity is likely to be more abundant in wired than in wireless.

In wired networks, the communication is point-to-point, determined by where the wires go. In wireless networks, there is some point-to-point, but it is easy to communicate from point to multipoint. This capability makes mobility and portability possible. This is useful if the goal is to broadcast the same information to everyone on a network, or on the move, but it also means that the network is always sharing capacity, creating a greater potential for congestion.

Reliability is a key attribute of communications networks, and discussion of reliability (and its implications for substitution) was vigorous. Although reliability can have any number of meanings the participants generally referred to reliability as “the ability to transmit all the bits through the medium.” In the wireless realm, reliability also refers to ability to transmit over a given geographic space.

Economic Considerations

The economic differences between wired and wireless networks are significant. There are large upfront costs to install physical wires as opposed to the relatively lower upfront costs associated with emitting a wireless signal, which creates advantages to building wireless networks. Additionally, it is extremely costly to upgrade wired from copper to fiber, for example, but upgrading within a given medium is appreciably less expensive. For example, upgrading fiber from 100 Mbps to 1 Gbps is not nearly as expensive as upgrading from coaxial cable to fiber optic cable.

For wireless, the building of towers requires the most economic investment, and it is the number (and placement) of towers that determines a wireless network’s ability to overcome geographic, interference and capacity constraints.

Jon Peha, Professor at Carnegie Mellon University, pointed out that he thought it was a mistake to group wireless in rural areas with wireless in urban areas; the constraints are different in kind. He described the difference as “coverage limited” (rural) or “capacity limited” (urban). As he further explained:

A cellular provider that is coverage limited means a cellular system in which, if you deploy the smallest number of towers you need to cover the region, the capacity will be just fine. You don't worry about capacity if you can meet your coverage needs, your capacity will be ok. This is true in lots of rural areas, and it’s true in the majority of the country by area, not by population.

But you also have capacity limited regions. That is, the problem of an insufficient number of towers would be congestion, not geographic reach. To ensure reliable coverage you need more than the limited number of towers. For these networks engineers say you don't need to worry about coverage; if you can meet the capacity requirements, coverage will take care of itself. It’s exactly the opposite design problem from an engineering perspective. And it turns the economics on its head.

An additional economic factor that was discussed was backhaul expense, which is the cost associated with connecting the edges of a wireless network to the wired backbone infrastructure. This is an important factor when considering the costs a wireless carrier faces when building out a wireless network.

The chart below, also from Jon Peha, summarizes the technological and basic economic differences between the various wired and wireless broadband options:

DSL
Cable
Wireless
Fiber/home

Edge
(Physical
Layer)

Twisted Pair
Coaxial
Cable
Radio
Frequency
Fiber Drop
Aggregation
DSLAM
Fiber Note
Backhaul
Optical Splitter
Strengths
Uses Existing Telco
Last Mile
Passes
95% Of
US Homes
Mobile
Future Proof
Concerns
Upper Limit on Bitrate
Shared Access Network
Innsufficient Available Bandwith
Higher Expense
Recent
Technology
Innovations
Vectoring
DOCSIS 3.1
Advance RF Technology
Fiberhoods
Figure 2: Technological and Basic Economic Differences Between Wired and Wireless Options

Capacity- and Coverage-Limited Areas

Given these economic factors, if a wireless carrier wants to increase capacity, the carrier must improve network efficiency or else either build a new tower or license additional spectrum—or likely both. Both of these options are expensive, so if the carrier is capacity constrained, it may choose to implement business model fixes, like usage-based pricing during peak hours, to keep usage from maxing out existing capacity. For a wireline provider, meanwhile, once it has already gone through the process of laying fiber, the process of expanding capacity is less costly.

On the other hand, if a wireless carrier wants to increase coverage, the economics are somewhat different. Business model fixes like usage-based pricing won’t solve coverage issues, nor will increased network efficiency. Instead, some combination of more infrastructure (e.g., towers) and longer transmission distances (e.g., through the use of lower frequency spectrum) will be required, both of which are, again, expensive. Thus a firm in a coverage-limited area will maximize coverage using as few towers as possible. The focus shifts from quality to coverage, so there will likely be uneven coverage across the region due to topography and population density, but there may be no congestion. Similarly for a wireline provider, expanded geographic coverage can’t be achieved without expensive infrastructure construction to reach previously unserved areas.

In those capacity-constrained situations where business model solutions are feasible and employed, they affect usage patterns and incentives in ways that expanded infrastructure solutions tend not to. One of the identified challenges of the usage-based pricing system, for example, is how it alters incentives among users (and thus carriers and edge providers). The imposition of a usage-based pricing regime will reduce consumers’ incentive to purchase and use relatively low-value-per-bit applications. Thus, for example, demand for an application like remote computer backup that uses a lot of data but provides relatively little value to the consumer will fall under such a regime. In other words, consumers will be “priced out” of lower-value-per-bit applications. Demand for high-value-per-bit applications like VoIP and telemedicine, on the other hand, may not be affected, or may increase.

In short, the incentives of the carrier change depending on the type of infrastructure and the type of constraint, and the resulting decisions can affect consumer demand and the overall mix of uses to which the network is put.

Jonathan Adelstein, President and CEO of PCIA, also identified technological innovations as another solution to expanding coverage (and as an alternative to building towers and licensing spectrum). As technology improves bits per second and per hertz, capacity can increase at lower cost than would otherwise be required. He also suggested that antennas on buildings, rooftops, and other tall structures could be an alternative to building towers. Antennas could expand capacity and coverage effectively because capacity and coverage are usually only necessary where there is a population center.

Kathleen Ham, Vice President of Federal Regulatory Affairs at T-Mobile, also identified roaming costs as an additional hurdle to expanding the coverage footprint for a carrier like T-Mobile. This concern met resistance, however, because roaming is not a cost that all carriers face and ultimately is a distribution issue—which carrier gets the revenue from the infrastructure. Whether a carrier decides to pay another for roaming is a business decision; either it is more profitable to pay the roaming fees or to build the requisite infrastructure.

In the end, however, it is important to note that very few networks are or will remain for long either all wireless or all wired. As Jon Peha noted:

Mobile data traffic is growing at about 75 percent per year, which is astounding. There is just no way, no matter what else we do—and there are lots of other things that we should be doing—that you can meet that kind of growth rate without having large numbers of devices with short-range lengths.

Now that doesn't tell me what the policy is or what the business model is because short-range lengths could be devices owned by the carrier or devices owned by the end-user or devices owned by third-parties like an airport or a Starbucks. Those short-range lengths may be unlicensed on Wi-Fi, they may be licensed on femtocells, etc. But I do know there have to be lots and lots of short lengths. And if there are lots and lots of short lengths, all of which need high data rate back to the Internet, probably on a wire, it means there has to be lots of wired infrastructure. So we can't entirely be talking about these things as if they are separate.

Rural Infrastructure Challenges

While wireless networks have the potential to offer less-costly and more efficient service to sparsely populated rural areas, it is important to consider the wired network infrastructure needed to support it. In order to build a wireless network to provide service in rural and underserved areas there must be some degree of wired network infrastructure; without it, wireless networks would be unable to broadcast a signal powerful enough to provide service.

Several participants took issue with some assessments of wireless network costs that seem to assume that the requisite wired network already existed. In reality, given the need for wired infrastructure to support rural build-out, the cost (or some of the cost) of wired infrastructure must be included in considering the cost of wireless networks.

There was also debate about whether simply adding spectrum, especially lower band spectrum that has more desirable propagation characteristics, would help coverage in rural areas.

In the first place, merely having spectrum isn’t the end of the story—it also requires towers, backhaul and other infrastructure, and in rural areas especially, the return on invested capital (ROIC) on such investments can be quite low, making even wireless build-out in rural areas an expensive proposition. And as Public Knowledge Senior Vice President Harold Feld noted, these architecture issues can even lead to congestion in relatively unpopulated areas. Moreover, where the spectrum available for use is high-frequency spectrum, more towers are required because of the propagation characteristics of higher-frequency spectrum. Jon Peha noted that: “If you double my frequency it turns out you roughly double the number of towers I need. If you can only get high frequencies, your costs are going to be a lot greater.”

Because infrastructure is so important—again, more than might be expected in rural areas—regulatory impediments to infrastructure construction become even more significant. Kathleen Ham pointed out that:

There are a lot of instances when we want to build out and we can't because of zoning and other obstacles. On Long Island we have a lot of people who want to use T-Mobile service, but we're having a difficult time getting towers sited out on Long Island because of the zoning difficulties there.

Jonathan Adelstein argued that spectrum is a long-term solution because it takes years for acquired spectrum to reach consumers. First, the carrier has to upgrade infrastructure by installing new antennas and transmitters on towers to handle the new frequencies. Second, the consumers’ products also need to be upgraded to use the new frequencies. “It isn’t as if the carrier can acquire more spectrum, flip a switch and begin offloading to the new spectrum band.”

To the extent that the short-term fix, then, is infrastructure, the need to speed up the regulatory processes is significant.

Conclusion

This session focused on the key differences between wired and wireless broadband technologies, specifically the economic factors associated with each technology, as well as the unique challenges facing each network in dealing with capacity and coverage constraints. It was clear from the discussion that bright line distinctions (and thus bright line regulatory prescriptions) are difficult to come by. In reality, most networks are hybrids, comprising both wired and wireless elements along with other infrastructure elements like switches and routers. And the different types of constraints can both have different economic consequences and create different regulatory issues. Nevertheless, regardless of the precise nature of each network, however, the removal of outdated or unnecessary regulatory impediments to infrastructure construction seems unambiguously good, facilitating the relief of both capacity and coverage constraints.

Share On