Using Advanced Modulation Techniques in a WLAN Environment
Allan R. Scott
For some time now, a quiet debate has been ongoing within the wireless community about what is necessary to enable wireless local area network (WLAN) systems to achieve widespread use. One view has held that the horizontal user is concerned about interoperability and that the adoption of an industry standard is a prerequisite for mainstream users to move forward with investment in the technology. Another view has been that, in the absence of an introduction of technology able to deliver performance sufficient to support bandwidth-intensive applications, industry agreement on a standard would have little impact on the market.
As the work of the 802.11 committee draws to a close, the chief question potential users of WLAN technology should be asking themselves is what does it mean? They may be surprised by the answer. In their effort to reach a consensus on a standard and salvage more than five years of work, the committee members elected to adopt an approach that will limit the performance and future improvements in performance for WLAN technology. In turn, these limitations stand to impede the delivery of bandwidth-intensive solutions, such as concurrent multi-application support that could grow the market beyond its heretofore limited base, the exact opposite of what the committee had set out to do.
In the near term, users will continue to see a wide range of noncompatible systems marketed until compliant products become available, a process that will take a year or so to complete. In the long term, most WLAN manufacturers will probably modify their products to make them standard compliant and either offer separate compliant and noncompliant product lines, or offer user-configurable dual-mode systems that operate in either compliant or noncompliant modes, depending on whether the user feels performance or interoperability is more important.
Another point the user may want to consider is that even with the adoption of a standard, direct sequence (DS) and frequency hopping (FH) WLAN systems will not be able to interoperate. Given the large installed base of DS systems, the inherent advantage of FH in a picocellular environment and the industry's move from DS to FH as the preferred technology for heavy traffic, mobile WLAN systems, the confusion in the market is far from over.
Despite the hoopla that the industry will create following the announcement of the adoption of a standard, most industry watchers are certain to agree on two points regarding WLAN technology: the technology that eventually becomes the standard will be dictated by market requirements, not by committee, and for the market to grow beyond its traditionally limited base, the products developed based on that technology will have to perform a lot better than they do today. This article relates specifically to modulation techniques used in a spread spectrum system. It is interesting to note that two of the reasons behind the committee's decision not to consider modulation techniques enabled by recent developments in technology, such as quadrature amplitude modulation (QAM), was because it would be inherently more difficult (that is, costly) for members to implement (given their already high levels of investment without a return) and would further delay the publication of an industry standard despite the fact that advanced modulation techniques, such as QAM, enable substantially higher performance.
Although wireless spread spectrum technology has been in use by the military for decades, only within the last 10 years has the basic technology been applied to the development of systems for commercial use. Expectations for the industry ran high initially, but the market for WLAN systems was slow to develop, encumbered by factors including slow performance, high prices, bulky form factors and incompatibility between competing systems.
The limitations of the technology focused the deployment of WLAN technology to those markets where the amount of data being transmitted was small (referred to as bursty data) and where immediate access to data by mobile workers over a large but well-defined area (for example, warehouse, distribution and retail environments) was crucial. Not surprisingly, the first companies to exploit the capability for portable WLAN computing were major system solution providers to those vertical markets.
Because at the time the technology first became available for commercial use suitable spread spectrum systems did not exist, these vertical market suppliers developed their own RF capabilities. Today, the largest suppliers to the vertical retail, warehousing and distribution markets still account for approximately 80 percent of all WLAN system shipments.1
Several of the early WLAN systems developed for those vertical markets relied on DS techniques and, consequently, were able to deliver raw data rates as high as 2 Mbps per channel. These DS-based systems were ideal for delivering wireless bridge functionality in a point-to-point configuration, or cells structured in a point-to-multipoint configuration with limited overlapping. However, DS suffered when used in a microcellular environment with multiple overlapping cells because DS tends to block itself. Although FH spread spectrum radio technology did not suffer from the self-blocking tendency of DS, the raw data rates delivered were substantially lower than DS.
Both DS and FH spread spectrum WLAN systems suffered from their inefficient use of the available bandwidth. A typical DS system with a raw data rate of 2 Mbps can deliver a throughput of 500 kbps. A typical FH system with a data rate of 1.6 Mbps is lucky to deliver throughput of 400 kbps. The balance of the available bandwidth goes toward overhead. Since throughput is what the user has available to support his or her application and since Windows®-based applications require a maximum of several megabits per second generally, neither DS nor FH WLAN systems were suitable for portable network computing in horizontal markets other than for simple e-mail access.
Another key limitation of spread spectrum WLAN systems was that they were just too big. Most of these systems were two-piece devices, which used an externally mounted antenna system that had to be affixed to the back of the PC. The bulkiness of the configuration in and of itself was an impediment to deployment, particularly given the trend in portable computing for smaller computing devices.
Infrared (IR) WLAN technology, both focused and diffused, was also popular early on. IR delivered high rates, used a small form factor and offered a low cost basis because of its widespread use in devices such as remote controls for televisions. However, IR was limited in relation to portable network computing because of its inability to penetrate walls. Nevertheless, the ability to transfer data inexpensively and rapidly from one computing device to another, such as a printer, has resulted in the incorporation of IR technology into almost all portable computing devices shipped today.
QPSK & QAM Digital Modulation2
Digital modulation transforms input digital signals into waveforms that are compatible with the nature of the communications channel. Examples of various modulation schemes are shown in Figure 1 . One of the two major categories of digital modulation uses a constant amplitude carrier and carries the information in phase or frequency variations known as phase-shift keying (PSK) or frequency-shift keying (FSK). The majority of FH WLAN systems today employ simple FSK modulation schemes. The other category of digital modulation, amplitude-shift keying (ASK), conveys the information in carrier amplitude variations.
Fig. 1: Modulation techniques; (a) FSK, (b) QPSK, (c) 16 QAM and (d) 64 QAM.
More advanced modulation techniques convey multiple bits of information simultaneously by providing multiple states in each symbol of transmitted information. Quadrature PSK (QPSK) conveys two bits per symbol and is prevalent in satellite communication.
QAM systems combines PSK and ASK to increase the number of states per symbol. Each state is defined as a specific amplitude and phase, meaning that the generation and detection of symbols are more complex than a simple phase detection or amplitude detection device.
Each time the number of states per symbol is increased, the bandwidth efficiency also increases. This bandwidth efficiency is measured in bits per second per hertz (bps/Hz). The displayed modulation schemes occupy the same bandwidth (after comparable filtering), but have varying efficiencies of (in theory at least) 1 through 6 bps/Hz.
Channel throughput and range are inversely related (as range increases, throughput decreases). Accordingly, use of QPSK and QAM modulation means that at any given range, the system will deliver a higher raw data rate than a system using FSK modulation.
Adaptive Equalization and Retransmission Avoidance2
Since more states produce a higher channel throughput, it would seem that just increasing the number of states would provide a simple way to increase channel capacity. Unfortunately, multipath interference and noise prevent the system from taking advantage of the increased states.
One way to deal with noise and other interference such as multipath is to employ error detection and correction algorithms. Using this technique, the receiver identifies corrupted signals and asks the transmitter to resend. The downside of this approach is that retransmissions affect user data throughput adversely. Satellite systems are able to implement QPSK and QAM modulation because they employ some of the most powerful error-correction techniques known. The systems can afford to retransmit without noticeable impact on user data throughput because of the tremendous amounts of bandwidth they have at their disposal. However, channel bandwidth available for unlicensed use is limited and every retransmission lowers user data throughput.
One solution to the problem of improving bandwidth utilization and minimizing the number of retransmissions is through implementation of an adaptive equalization technique. The technique allows the WLAN system to reconstruct the original signal dynamically. Accordingly, and assuming the same operating environment, a system incorporating adaptive equalization will deliver higher user data throughput than a system relying solely on retransmission.
Without adaptive equalization the receiver is unable to distinguish the various states from the noise. Without adaptive equalization, effective user data throughput actually can deteriorate with the implementation of advanced modulation schemes. Figure 2 shows the impact of adaptive equalization on a 16-QAM system.
Fig. 2: A 16 QAM system (a) without and (b) using adaptive equalization.
Spatial Diversity in the Antenna Systems
Another factor affecting bandwidth utilization is signal strength. Assuming a constant noise level environment, signal strength decreases as the distance between a transmitter and receive increases, both in absolute terms and as a ratio of signal to noise. As the signal-to-noise ratio decreases, the number of over-the-air errors increases, requiring more and more retransmissions.
One way to combat this effect is simply to increase the number of antenna elements. Although such a solution is fairly simple to implement in an external access point, such a solution is impractical for the small interface device that links to the portable computing device. In almost all WLAN systems being marketed today, the interface device uses an externally mounted antenna system, as shown in Figure 3 . The addition of a second antenna would make the system even bulkier.
Fig. 3: A typical 1.6 Mbps WLAN system.
This problem has been addressed by providing an antenna system that delivers spatial diversity at both the network interface card (NIC) and access point. To avoid a bulky external antenna system with two antennas affixed to the portable computing device, a unique antenna design was developed that delivers spatial diversity and yet is small enough to be embedded fully into the NIC, as shown in Figure 4 . Unlike other PC card NIC devices that require a physical cable connection to an external antenna system mounted on the back of the PC, a single-piece PC card type II design can be easily plugged into a portable computing device no larger than a calculator.
Fig. 4: A 3.2 Mbps WLAN system with spatial diversity; (a) the PC card adapter and (b) the access point.
All of the advanced software subsystems that manage such a WLAN system require a good deal of horsepower. Rather than rely on the user's portable computing device for processor power, a 32-bit reduced instruction set computing processor was designed into the silicon of the basic chipset. The incorporation of a powerful processor engine yields several ancillary benefits.
One benefit is that the need for an external central processing unit (CPU) within the access point is eliminated. Most WLAN systems today require that a WLAN PC card be married with an external CPU or PC to form an access point or the equivalent. The requirement for an external CPU increases the cost basis for both the system and the network infrastructure substantially. A chief impediment to the broader deployment of WLAN systems has been the high cost of access point infrastructure. The cost of installing the necessary access point infrastructure can be onerous, particularly in a cost-cutting business environment where the chief benefit may be difficult to measure.
Another benefit of the processor engine is that the WLAN system itself becomes a computing platform able to support user-selected software-based functionality. For example, although a manufacturer will provide some level of roaming capability within the WLAN system typically, a user or systems integrator may prefer to use his or her own customized software to enable the network to retain its own custom look and feel. This type of flexibility is of particular importance to systems integrators and original equipment manufacturer suppliers who have invested substantial sums into the development of proprietary software systems, and who want to continue to supply a unique system to their customers.
The result is a small, economical system capable of operating efficiently in a microcell architecture, able to communicate through walls and able to deliver user data throughput adequate enough to support Windows and other applications concurrently, in effect blending the performance of IR with the benefits of RF. Depending on the implementation of the technology, the same basic design is extendable to 64-QAM modulation, which enables the delivery of a raw data rate of 4.8 Mbps with a throughput of 3 Mbps.
The issue of throughput is important for reasons beyond convenience and responsiveness; it impacts system economics as well. High throughput enables concurrent support of multiple applications using the same network infrastructure.
WaveAccess' JaguarTM WLAN products were introduced in April 1996.
- Yankee Group statistics.
- Portions based on an article appearing in Communications Systems Design magazine; used with permission.