Broadband Fixed Wireless Access as a Key Component of the Future Integrated Communications Environment

Copyright 2001 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

This article was published in the September 2001 issue of
IEEE Communications.

ABSTRACT
It is an exciting time for broadband fixed wireless with key developments in frequency bands from 1 to 60 GHz and a range of new technologies being developed. While working on these new technologies, it is easy to forget that fixed wireless access will form part of an integrated communications environment of the future where users will have one communications device working in the home, at the office, and outdoors. This article predicts the communications environment of the next 20 years and looks at the role of fixed access within that environment. This involves assessing how fixed access systems will interface and integrate with in-home wireless networks, how their architecture will enable multiservice operators to utilize the same core network across a range of different access technologies, and how they will act as a channel to carry mobile traffic originating within the building. Based on the requirements this vision and architecture implies, this article critically assesses the different fixed wireless technologies available to date and compares their capabilities to provide future-proof broadband fixed wireless platforms.

Broadband Fixed Wireless Access as a Key Component of the Future Integrated Communications Environment

William Webb, Motorola

Today's Communications

Communications today is a mixed and rather disorganized environment. The typical office worker in a developed country currently has a wide range of ways to communicate, including:

The office telephone, used mostly for voice communications complete with mailbox system
The office fax machine, now being used less as e-mail takes over
The office LAN, providing high-data-rate communications such as e-mail and file transfer
Dialup networking when out of the office, providing the same capabilities as the LAN but at a much slower rate
Mobile telephones providing voice communications, a mailbox, and in some cases low-speed data access
A pager providing one or two-way messaging
A home telephone providing voice communications and dialup access along with a home answering machine
A computer at home linked to a different e-mail system, perhaps using high-speed connections such as asynchronous digital subscriber line (ADSL) or cable modems

Managing all these different communication devices is complex and time-consuming. The worker who has all of these (and many do) will have five phone numbers, three voicemail systems, and two e-mail addresses. There is no interconnection between any of these devices, so all the different mailboxes have to be checked separately, using different protocols and passwords. Contacting such an individual is problematic due to the choice of numbers to call, and many default to calling the mobile number as the one most likely to be answered. Although many are working on systems such as unified messaging, designed to allow all types of communications (voice, fax, e-mail) to be sent to one number, we are still some way from the ideal situation where individuals only have one "address" and all communications is unified. Effectively, there is little convergence, at least as far as the user is concerned, between all these different fixed and mobile systems. How this will change, and more detail on what the future will look like, especially for fixed wireless, is the subject of this article.

How We Will Communicate in the Next 10 to 20 Years

Based on an understanding of possible types of communications and the shortcomings of current communications systems, we might predict the following advances over the next 10 to 20 years; these are discussed in more detail in [1]:

Video communications wherever possible: When talking from the home or office, all communications should have the option of video links and hands-free talking to make communications as natural as possible. This may not always be appropriate, especially when mobile, but the option should be available.
Complete unification of all messaging: Each individual should have a single "address" which will typically be of the form "john.smith@my-isp.com" (some further detail may be required to overcome the problem of multiple John Smiths) to which all communications will be directed.
Intelligent filtering and redirection: Upon receiving a message the network, based on preferences and past actions, will determine what to do with it, knowing the current status of John Smith (whether mobile, at home, etc.). Work calls might only be forwarded during the weekend if they are from certain individuals, otherwise stored and replayed on return to the office, and so on.
Freedom to communicate anywhere: It should be possible to have almost any type of communications anywhere, although the higher the bandwidth and the more "difficult" the environment, the higher the cost.
Simplicity: For example, upon walking into a hotel room, communications devices should automatically network with the hotel communications system, determine whether the tariff charged by the hotel is within bounds set by the user, and automatically start downloading information, presenting it to the user in accordance with his preferences.
Context-sensitive information: As well as being able to get information from the Internet on request, the Internet should provide the user with the information he needs depending on his location, plans, and circumstances.

Technically, all this is relatively straightforward. No fundamental breakthroughs in communication theory, device design, or computing power are necessary to realize this vision. The key issues preventing realization of this vision today are:

Lack of bandwidth: Most homes and mobile phones do not have access to sufficient bandwidth to realize video transmissions of good or high quality.
Multiplicity of disparate systems: As discussed earlier, there are many different communication systems which, to date, are rarely linked in an intelligent fashion, partly because they utilize different protocols, technologies, and paradigms.
Multiplicity of different operators: Different systems are often run by different operators who do not always perceive commercial justification for tightly integrating with other systems which may be run by competitors, particularly since many operators are now involved in a complex web of partnerships.
Economics: Provision of some systems, such as a radio transmission node in each hotel room, is generally not economically viable today and must await lower cost realizations.
Lack of standardization: For a user to enter a hotel and his/her laptop to automatically download e-mails, there must be an agreed on radio standard and infrastructure in place so that the hotel and the laptop can communicate. In many areas, standards are being developed but are far from ubiquitous.

The following sections look at the developments underway today which might form the basis of realizing the vision and extrapolates these forward. Key for the fixed wireless arena is the requirement for ubiquitous and high-speed wireless access to the home. We are still some way from realizing this vision. In the rest of this article we consider some of the constraints and technologies that might be adopted.

The Future Architecture: A Truly Converged Communications Environment

A summary of the network of the future that would deliver the requirements discussed above is shown in Fig. 1. Much is missing from this figure, and much has been simplified in order to show all the key elements in one picture. This figure demonstrates just how fixed and mobile systems will converge:

Both will be linked back to the same postmaster ¹ using common protocols and possibly a common core network (when both are owned by the same operator).
Mobile devices will also utilize in-home and office radio networks connected back through fixed networks into the postmaster, which coordinates their use.

In summary, the key elements of realizing the network of the future are:

Ubiquitous broadband access to the home delivered using a range of different technologies, including fixed wireless, based on technologies discussed in later sections.
Standardized in-home networks consisting of simple radio devices in each room connected to a home LAN. It is likely these will be enhanced developments of standards like BlueTooth.
Standardized radio devices in most home and office appliances using the same short-range radio standard.
The provision of an "intelligent postmaster" function, probably provided by third-party entities.
A standard protocol for all networks to communicate with the postmaster and each other, probably using IP as the underlying transport mechanism and building on the protocols developed for the third generation (3G).
Widespread cellular architecture using a single 3G standard, or alternatively multimode phones operating over a multiplicity of standards with high-speed access delivered by wireless LAN (W-LAN) solutions in certain important areas.
A standard approach for office wireless networks, most likely based on wireless LANs, common to all offices.
Communicator devices able to work on the cellular, home, and office networks in a seamless manner.
An environment that enables the development of innovative services by third parties, probably delivered through the Internet, and can be downloaded and run by all communicator devices using languages such as Java.

We now consider the ability of broadband fixed wireless to play its envisaged role in this future vision.

Technical Constraints on Broadband Fixed Wireless Systems

      If fixed wireless is to play a key role in this network of the future, it must be able to deliver high data rates to most homes. Being more specific about high data rates is difficult because it depends on the user, but we might note that high-definition video transmission requires around 8 Mb/s, and if we allow for multiple simultaneous transmissions to or from the home, data rates in excess of 10 Mb/s will be needed. To date, fixed wireless has been unable to deliver data rates in excess of 10 Mb/s to a high percentage of homes in a given area cost-effectively. In this section we examine the theoretical and economic constraints on fixed wireless to assess whether this might change in the future.
      It is possible, by making some assumptions, to calculate the theoretical capacity that can be provided by fixed wireless solutions. This approach is described in detail in [2] and summarized below. The approach starts with Shannon's law, setting out the maximum information that can be transmitted per second per hertz of spectrum, and adds equations developed by Lee [3] and Webb [4, 5] to model operation in a clustered cellular environment. Key to modeling capacity in a fixed wireless environment is understanding that propagation conditions are different to those in a mobile environment. Because many systems use directional antennas with line of sight (LOS) or near-LOS paths, the path loss exponent is often closer to that of free space, namely 2, than the mobile case, where it tends to fall between 3.5 and 4. However, for interfering signals there is often no LOS, and there may be isolation created by the directional antenna. As a result, the interfering path loss may be closer to that for mobile.
      The mathematical analysis shows that a solution can be derived [2] indicating that the capacity is inversely proportional to the cell radius and the bit rate required per user. One of the key parameters is the modulation scheme that is adopted. Figure 2 shows the variation of capacity, M, with the signal-to-interference ratio (SIR). The curve clearly shows that the highest efficiencies can be obtained at the lowest SIRs. This result is in line with earlier work conducted by Webb [4, 5] where it was reported that the best results were obtained with single-level modulation as opposed to multilevel modulation. Hence, we assume the use of quadrature phase shift keying (QPSK) modulation.

Analysis

With a range of assumptions, detailed in [2], it is possible to determine the viability of broadband fixed wireless systems which technically approach the Shannon limit and economically fall in line with current revenue expectations. The end result is shown in Fig. 3.
Figure 3 shows that below a spectrum allocation in megahertz of five times the user data rate in megabits per second (e.g., if the user data rate were 10 Mb/s, the spectrum allocation would be 50 MHz), profitable operation seems unlikely. Spectrum allocations above around 10 times the user data rate result in little extra increase in profitability; hence, a 10 times allocation is probably most appropriate, minimizing use of the scarce spectrum resource.² (Note that the assignment would need to be twice this size in practice to allow duplex communications; i.e., an uplink and a downlink assignment, both equal to 10 times the user rate would be required.) Hence, theoretically, with a 2

100 MHz spectrum assignment, a 10 Mb/s duplex service can be profitably offered to residential users.
Another way of looking at this result is that it would appear to be profitable to operate a broadband fixed wireless system in the region where the maximum data rate is around 10 percent of the spectrum assignment. The data rate per subscriber is then dependent only on the spectrum assigned by the regulator. For typical assignments in the frequency bands of 10 GHz and above, and some assignments at 2 GHz,³ bandwidths of 10 Mb/s per subscriber using fixed wireless would seem both technically and economically viable.
In the next section, specific technologies that might meet or exceed this performance are described.

Technologies For Broadband Fixed Access

There are a wide range of different technologies proposed for fixed access. Here, the key technologies we might consider in future systems are listed and briefly evaluated. For a more detailed description of these technologies than space allows here, see [2].

At the System Level

Here there appear to be two basic concepts available:

A conventional point-to-multipoint (PMP) solution where each subscriber unit communicates directly with a base station
A mesh approach where subscriber units communicate with nearest neighbors and information is passed back through the mesh in a manner analogous to Internet traffic

Examples of these two concepts are shown in Fig. 4. The status quo is represented by the first option. Here we consider the merit of the mesh approach when compared with the conventional PMP structure.
The mesh approach effectively changes the "rules" used for capacity and link budget calculations by turning each link into a point-to-point link. Arguably this has a similar effect to adaptive base station antennas in a conventional system which can provide a narrow beam to each subscriber unit. The potential advantages of mesh solutions are:

An increase in capacity as a result of frequencies being reused on a very localized level. Effectively, this is the equivalent of a microcellular approach on a conventional design, although these capacity gains could be offset by the need for each node to relay traffic.
An improvement in quality as a result of each link being short and hence having a high link budget.
A possible cost reduction in the subscriber unit as a result of the less demanding link budget. However, this may be offset by the additional complexity required to provide the repeater element needed within the mesh architecture.
An ability to replan the system without repointing subscriber antennas (e.g., in cases where subscriber numbers grow more quickly than anticipated).
A potentially nearly "infrastructureless" deployment.

Against this needs to be balanced the potential disadvantages:

Highly complex algorithms are required to manage the system and avoid "hot spots" which may be unstable and result in poor availability.
Different and novel medium access control (MAC) mechanisms may be required which will need development and add to the complexity.
The initial investment is relatively high since "seed nodes" have to be placed so that the mesh can form as soon as the first subscriber is brought onto the system.
Marketing issues may be problematic in that customers may not want to rely on nodes not in their control and not on their premises for their connectivity, and may not want their equipment to be relaying messages for others.

It is difficult to draw definitive conclusions at this point since many of the above variables are unknown. If the complexity and risk can be overcome, it seems highly likely that mesh systems will provide greater capacity than conventional systems for a given cost.

Layer One/Two

There are a number of discrete technologies here, which are mostly independent, so each can be considered separately. The issues are:

Time-division duplex (TDD) vs. frequency-division duplex (FDD)
Adaptive vs. fixed rate modulation
Orthogonal frequency-division multiplexed (OFDM) vs. single-carrier
Code-division multiple access (CDMA) vs. time-division multiple access (TDMA)
Adaptive vs. conventional antennas

TDD vs. FDD -- FDD represents the status quo. The question is whether TDD brings substantial benefits to the operator. Key to this question is:

Whether the spectrum is simplex or duplex. If it is simplex, TDD overcomes the need for a guard band, which can be wasteful of spectrum; however, TDD needs a guard time, which may be as large percentage-wise.
Whether the data is both asymmetric and the asymmetry is time-variable. If the asymmetry is known beforehand, unbalanced FDD assignments can be used; if not, TDD can bring some efficiency gains.

Determining the gains is then an issue of understanding the variability of the asymmetry and the simplex or duplex nature of the spectrum. It seems likely that if the asymmetry is highly time-variable, TDD will bring definite advantages, in principle up to a maximum of a 100 percent capacity gain (where traffic only flows in one direction -- 100 percent asymmetry). It also seems certain that in simplex bands TDD will bring advantages unless complex frequency assignment procedures are adopted for FDD whereby the guard band is different in different cells, requiring complex planning and possibly greater expense for the subscriber unit. Hence, assuming that the cost of implementing TDD is not great, it is likely to bring significant benefits.
In conclusion, forward predictions of asymmetry time variance are uncertain, so TDD gains cannot be definitively quantified, but it seems possible that there may be some worthwhile gains in certain situations.

Adaptive vs. Fixed Rate Modulation -- With adaptive modulation, instantaneous carrier-to-interference (C/I) and signal-to-noise ratio (S/N) measurements are made, and the number of modulation levels are modified dynamically. Hence, if the subscriber is experiencing relatively good S/N, perhaps because he is not in a fade, more modulation levels can be used without a greater power requirement, thus not adding additional interference to the system [6]. Adaptive modulation is the technique proposed for EDGE, an enhancement to existing cellular systems. Adaptive modulation provides the greatest gains in fading channels where the number of modulation levels can be instantaneously matched to the channel conditions. In Gaussian channels they bring no advantages. Fixed wireless channels tend to fall somewhere between these two, so the advantages are likely to be less than in the mobile case, but they are still likely to bring some gains.

OFDM vs. Single-Carrier -- Here single-carrier is assumed to mean existing schemes (sometimes erroneously considered quadrature amplitude modulation, QAM), while OFDM is considered in the broadest sense to include vector OFDM (VOFDM) and other variants. OFDM brings two benefits:

It effectively removes the need for an equalizer by turning a wideband signal into a multitude of narrowband signals.
It can overcome some specific types of narrowband interference more simply than other schemes.

However, it equally brings some disadvantages:

It requires an overhead of around 12 percent for training sequences and cyclical redundancy; however, equalizers in non-OFDM solutions also require training sequences, so depending on the size of the cyclical redundancy, this may not be an issue.
Because it transforms intersymbol interference (ISI) into narrowband Rayleigh fading, it foregoes the opportunity to make use of the effective diversity in multiple paths: an equalizer will actually increase the performance by combining the multiple signals, whereas within OFDM these signals are nonresolvable and appear as Rayleigh fading.
The peak/average ratio of OFDM is perhaps 3–5 dB higher than, say, QPSK, putting more stress on power amplifier design. This is an issue especially for subscriber units.

      As a result, it is clear that, compared to a single-carrier solution with an equalizer able to accommodate the channel ISI, OFDM will result in inferior performance. However, equally, in the case where the equalizer is unable to accommodate the channel ISI, the single-carrier solution will typically fail, whereas the OFDM solution will mostly continue to work. The key unknown is to what extent the channel will exhibit ISI beyond the range of a commercially viable equalizer or whether there will be narrowband interference. It is generally agreed that, to date, there is insufficient information about the channel to be able to definitively answer this problem. Given the lack of information, OFDM represents the "more conservative" solution, guaranteeing operation in most environments while not necessarily maximizing performance.
      CDMA vs. TDMA -- CDMA is generally agreed to be the most efficient multiple access scheme for mobile applications. However, there are different constraints within the fixed access environment related to the desire of operators to be able to instantaneously give all, or a substantial part of, the available bandwidth to an individual subscriber. Because of the manner in which CDMA is configured, within a sector in a clustered environment typically only around 200 kb/s of throughput are available per megahertz of spectrum.⁴ This compares with a QPSK TDMA solution where up to around 1.8 Mb/s might be available. Of course, CDMA allows single frequency reuse, so overall efficiency is high, but this example shows that in order to provide a subscriber with, say, 10 Mb/s of data, CDMA would require either a carrier of 50 MHz bandwidth or a multicarrier receiver, while TDMA would only require a carrier of about 6 MHz.
      So, in summary, it seems clear that for constant bit rate narrowband services, CDMA is the most spectrally efficient solution by some distance. For broadband applications above circa 2 Mb/s/user, CDMA solutions will probably not be economically viable. Below 2 Mb/s, subject to there being sufficient spectrum and the cost of the CDMA system being competitive, CDMA is probably the optimal multiple access scheme.

Adaptive vs. Conventional Antennas -- The conventional approach is to use sectored antennas at the base station, possibly with diversity, and a directional antenna at the subscriber unit, again possibly with diversity. Adaptive antennas bring potential gains as follows:

At the base station they can result in a narrow beam to an individual subscriber, limiting interference to other sectors and thus increasing capacity.
At the base station they can be used to null interferers, enhancing the C/I of the received signal.
At the subscriber unit they can be used to null interference, again increasing the C/I.

It is not clear how great these gains will be. Because of the directionality already present on fixed wireless links, it is likely that the gains would be less than those for the mobile case. However, deployment is also simpler than the mobile case since there is little need to track subscribers. In the mobile case, adaptive antennas tend to enhance the uplink rather than the downlink; for fixed wireless the most constrained link is typically the downlink because of the asymmetry of usage, so different techniques will need to be used. Another point to note is that adaptive antennas will be simpler for TDMA transmission where one array can be steered to each subscriber, rather than CDMA transmissions where multiple arrays would be required to steer the different codes constituting a single carrier to different subscribers.
As a summary to adaptive antennas, it would appear that probably the most useful deployment will be of antennas that can illuminate a subscriber using a narrow beam, but substantial work is required to understand whether the capacity gains this would bring would be offset by the additional cost of the solution.

The Future Direction Of Broadband Fixed Wireless

This article demonstrates that fixed wireless has a significant role to play in the future of broadband communications, being used in areas where the copper or cable infrastructure is not appropriate or by new operators who do not have access to these legacy resources. It also demonstrates that operators can economically and technically offer broadband services to users of 10 Mb/s or more provided that they have a spectrum allocation of 100 MHz or more. Finally, it demonstrates that there are a plethora of technical options that can be used to provide fixed wireless solutions, including a number of promising new ideas that could overcome problems of poor coverage at higher frequencies.

Acknowledgments

This article is based on previously published material from WAS 2000 organized by DELSON GROUP.

References
[1] W. Webb, The Future of Wireless Communications, Artech House, 2001.
[2] W. Webb, Introduction to Wireless Local Loop: Narrowband and Broadband Solutions, Artech House, 2000.
[3] W. C. Y. Lee, "Spectrum Efficiency in Cellular," IEEE Trans. Vehic. Tech., vol. 38, no 2, May 1989, pp. 69–75.
[4] W. Webb, "Modulation Methods for PCNs," IEEE Commun. Mag., vol. 30, no. 12, Dec. 1992, pp. 90–95.
[5] W. Webb, "Spectrum Efficiency of Multilevel Modulation Shemes in Mobile Radio Communications," IEEE Trans. Commun., vol. 43, no 8, Aug. 1995, pp. 2344–49.
[6] L. Hanzo, W. Webb, and T. Keller, Single and Multi Carrier Quadrature Amplitude Modulation, Wiley, 2000.

Biography
William Webb [SM] graduated in electronic engineering from Southampton University, United Kingdom, with a first class honors degree and all top year prizes in 1989. Since then he has been awarded a Ph.D. and an M.B.A. From 1989 to 1997 he worked for a range of communications consultancies in the United Kingdom in the fields of hardware design, computer simulation, propagation modeling, spectrum management, and GSM standardization. In 1998 he moved to Motorola in the United States, where he is responsible for strategic management across Motorola's entire communications portfolio. He has published over 50 papers, holds four patents and a number of awards, is a fellow of the IEE, and is the author of seven books, including The Complete Wireless Communications Professional and The Future of Wireless Communications.