Across the industry, designers have pitched a variety of standards as potential solutions for home networking. In the emerging home networking space, the design community has focused most of its attention on the Bluetooth, HomePNA, HomeRF, and HomePlug specifications. However, designers have lately begun to devote more focus to wireless LAN (WLAN) technology as a viable home networking solution. Specifically, equipment developers are starting to build systems based on the IEEE's 802.11a WLAN specification.

The IEEE 802.11a specification, which provides for 5-GHz operation and 54-Mbps data rates, brings many new benefits to the developers of home networking equipment designers. From higher data rates to improved spectral efficiency, 802.11a delivers improved performance in the home networking environment.

The 2.4-GHz traffic jam

Currently, designers of home networking equipment are trying to build systems that operate in the Federal Communications Commission's (FCC's) 2.4- to 2.4835-GHz unlicensed industrial, scientific, and medical (ISM) band. The challenge with developing products in this band, however, is interference.

Over the past few years, the 2.4- to 2.4835-GHz band has become overpopulated with wireless networking products such as Bluetooth systems, HomeRF systems, IEEE 802.11b WLAN devices, and cordless phones, among others. This increased use has created congestion bottlenecks and increased interference in the home-networking environment.

The interference in the 2.4-GHz band results from a myriad of incompatible data transmission techniques. Devices operating in this band can either be classified as direct sequence spread spectrum (DSSS) or frequency hopping spread spectrum (FHSS) systems.

The DSSS data transmission scheme is used primarily by IEEE 802.11b systems and can deliver data rates up to 11 Mbps.

How it works

In IEEE 802.11b products, the DSSS mechanism works by first multiplying the narrowband message by a larger bandwidth signal, which is usually a pseudorandom noise (PN) code, before being sent by the transmitter (see Figure 1). This broadens the spectrum. The amount of broadening is commonly referred to as the process gain and is defined as the ratio of the transmitted bandwidth after the data is spread (BWTransmitted) to the bandwidth of the actual information (BWInfo). Usually the processing gain (GP) is much larger than unity.

This broadening technique can be used to allow the development of multi-access systems. In such systems, each user's transmission would be spread with a different PN code and sent within the same frequency band. When the desired spread signal is received, it is multiplied by the appropriate PN code again. This process causes the desired received signal to be despread back to the original transmitted data while all other signals and noise, which are uncorrelated to the PN spreading code used, become more spread. The desired signal is then filtered to remove the remaining wide spread interference and noise signals. In this manner, each user's signal can be independently sent and received in the presence of other users' signals within the same frequency band.

Unfortunately, the IEEE 802.11b DSSS standard, as defined, cannot employ DSSS for multi-access. In short, the required processing gain and bit energy to noise ratio (Eb/N0) do not allow another signal of the same strength (but with different PN code) to be transmitted at the same time. The IEEE 802.11b DS systems can only tolerate interference up to -3 dB relative to the desired signal, which is much less than the interference that can be presented to one user by another on the same channel.

Thus, DSSS technology cannot be used to allow simultaneous transmissions in the same frequency band. It is for this reason that the 802.11b standard segregates simultaneous transmissions onto three different frequencies, each of 22-MHz bandwidth, in order to allow them to transmit at the same time in the allotted 2.4-GHz band.

FHSS systems, such as Bluetooth and HomeRF devices, differ from DSSS systems in the manner they try to avoid interference. FHSS systems avoid interference with other transmission signals in the same band by hopping over many different frequency channels. During any one hop, an FHSS signal appears to be in a narrowband signal. In the case of Bluetooth, for example, this narrowband signal has a 1-MHz bandwidth and hops 1,600 times per second over 79 channels. Therefore, an FHSS signal is agile and does not spend much time in any one frequency.

Performance challenge

One of the main challenges with FHSS systems is data rate performance. In their current states, FHSS systems can employ 1-MHz narrowband signals that deliver data rates up to 2 Mbps. With the passage of a new rule by the FCC, system designers can increase signal bandwidth to 5 MHz, allowing systems to achieve data rates of up to 10 Mbps. But getting beyond 10 Mbps is not an easy task for FHSS systems, which could limit their performance capabilities in the home-networking environment.

Another key problem with both FHSS and DSSS is interference. Now that FHSS radios can operate at wider bandwidths, there is a clear concern in the industry that more interference may occur between DSSS and FHSS products.

To evaluate the simultaneous interference between DSSS systems such as IEEE 802.11b and FHSS systems such as Bluetooth, it is best to separately examine the effect of each system on the other.

In terms of the interference of narrowband FHSS systems on DSSS transmission, the interference level is severe - especially in the presence of a fast moving signal such as Bluetooth. In terms of the interference of broadband DSSS systems on FHSS transmissions, the effect is still significant. When FHSS systems hit a frequency that has interference, the signal can hop to another channel that is outside the bandwidth of the DSSS system. However, regardless of the scenario, the throughput is almost always reduced.

The 5-GHz autobahn

By contrast, the 5-GHz frequency spectrum employed by IEEE 802.11a is devoid of interference issues. In the United States, the U-NII band was originally aimed at inexpensive computer networking in public schools. Since no other devices currently operate in the U-NII bands (5.15 to 5.35 GHz and 5.725 to 5.825 GHz), the ability to provision up to 54 Mbps in 20-MHz channels increments is not a problem. Thus, the IEEE 802.11a standard has defined eight such channels in the lower 200-MHz region and four channels in the upper 100-MHz region of this spectrum.

These high data rates are more than sufficient to scale to the need of present and future home networking traffic such as multiple streaming MP3 audio and voice-over-IP (VoIP) telephony, Internet access, digital TV, MPEG-2 DVD streams, and even MPEG-4 video on demand (VoD) services. The data rates of Bluetooth, 802.11b, and HomeRF (with even the recent extensions as allowed by the FCC) offer no greater than 11-Mbps system capacity, which is not enough to carry even a single channel of HDTV.

The OFDM advantage

Besides being high-speed and immune to interference from competing standards in the same frequency spectrum, 5-GHz systems implemented with international standards such as IEEE 802.11a also provide, by design, more robust transmissions. This is largely because the standard uses orthogonal frequency division multiplexing (OFDM) as its modulation technique in overcoming multipath effects common in home environments.

Multipath effects occur when the transmitted signal is reflected from objects such as walls, furniture, and other indoor objects. Under such circumstances, the transmitted signal may not have a single direct path to the receiver. Rather, there can be a number of different paths, or multipaths, each of which has a different distance to travel from the transmitter to the receiver, and thus each experiences a different delay. As a result, the signal can have multiple copies of itself, all of which arrive at the receiver at different moments in time. Thus, from the receiver's point of view, it receives multiple copies of the same signal with many different signal strengths or powers.

For example, suppose that the maximum difference between the first and last multipath signal at the receiver is tmax (known as the delay spread). Furthermore, assume that for every T time interval the transmitter transmits a discrete block of digital information (a symbol). Under such conditions, a given received symbol can be potentially corrupted by up to tmax/T previous symbols (shown in Figure 2). This effect is commonly known as inter-symbol interference (ISI). Correcting for a large tmax/T ratio is an expensive proposition for the receiver.

Since delay spread is a function of the environment and cannot be altered by the radio, the only recourse is to increase the transmission interval time T between the transmitted symbols. But this would slow the transmission and run contrary to the mantra of high speed.

OFDM satisfies both requirements: instead of transmitting the information using one frequency, or carrier, at an interval of T, OFDM divides the transmission among N different subcarriers, each with a transmission interval time lengthened by N (or with 1/N as much data). Thus, despite the fact that the data rate for each individual subcarrier has been reduced by a factor of N, the paralleling of N different transmissions means that the overall transmission rate of the system will remain the same. In addition, the tmax/T ratio, with respect to each subcarrier, has been decreased to tmax/(T * N). This means that each subcarrier is now N times more multipath-tolerant and ISI-tolerant. For IEEE 802.11a systems, the number of subcarriers is equal to 52.

OFDM is also one of the most spectrally efficient data transmission techniques available. This means it can transmit a large amount of data in a given frequency bandwidth.

Instead of separating each of the 52 subcarriers with a guard interval, OFDM actually overlaps them. If this is done incorrectly, it could lead to an effect known as inter-carrier interference (ICI), where the data from one subcarrier cannot be distinguished unambiguously from its adjacent subcarriers.

OFDM avoids this problem by making sure that the subcarriers are orthogonal to each other. Orthogonality is best illustrated in Figure 3. Here, three subcarriers are depicted, with data modulated on each. Together, these subcarriers make up one OFDM symbol that is then sent out into the channel. In an actual OFDM system, these waveforms would all have different phases or different amplitudes and phases, due to the use of either phase-shift keying (PSK) or quadrature amplitude modulation (QAM), respectively. For the sake of illustrative simplicity, all are shown here with the same amplitude and phase offsets.

Orthogonality

The orthogonality of OFDM comes from the precise relationship between the subcarriers that make up one OFDM symbol. In an OFDM system, each subcarrier has exactly an integer number of cycles in a given T time interval. In other words, the number of cycles between any two adjacent subcarriers differs exactly by one. This implies that each subcarrier frequency is an integer multiple of a base frequency (that is, f1 = f0, f2 = 2 * f0, f3 = 3 * f0, and so on). These properties allow each subcarrier to be individually and independently demodulated from any other adjacent subcarriers.

An equivalent frequency domain representation of Figure 3 is seen in Figure 4. Here, each subcarrier's frequency spectrum is represented by a sinc function, one of whose properties is to peak at its center frequency and go to zero at all integer multiplies of this frequency. This property allows the OFDM receiver to effectively demodulate each subcarrier because, at the peaks of each of these sinc functions, the contributions from other subcarrier sinc functions are zero. It is the orthogonality that allows the subcarriers to be packed tightly against one another and to efficiently use the frequency spectrum, thus delivering unprecedented 54-Mbps WLAN connections.

Built using CMOS

One of the misconceptions about the commercial viability of 5 GHz is its perceived high cost. The high operating frequency and low noise requirements in the front-end RF transceiver would seem to demand the use of silicon germanium (SiGe), gallium arsenide (GaAs), or other more exotic and expensive technologies.

Happily, this is not the case. CMOS has proven to be a remarkably resilient technology, one that has been improved at every successive technology generation. As a result of the continuous efforts in CMOS scaling, it is now possible to use the technology to build high frequency-integrated circuits.

A recognized metric for evaluating the speed of each technology is the ft or current gain cut-off frequency. This frequency corresponds to the point where a transistor's measured current gain is equal to one (when the output equals its input current value for, effectively, no gain). This sets the highest possible clock rate with a given technology. In almost all cases, the continuous shrinkage of the minimum gate length of a CMOS device leads to ever-increasing cutoff frequencies.

Figure 5 shows a comparison between CMOS ft values and the other less mainstream technologies. For CMOS technologies with 0.25-micron gate lengths, the ft value is between 20 to 30 GHz and for 0.18-micron this value increases to 30 to 60 GHz. Although GaAs and SiGe technologies can produce at least double these values in the 70- to 100-GHz ranges, such speeds might be overkill. Current, mainstream CMOS have more than enough horsepower to support the demands of current 5-GHz WLAN products.

In addition to speed, the manufacturability of a technology is also an important consideration. In this respect, CMOS is not simply adequate, but, again, surpasses other technologies. CMOS reaps the benefits of almost 35 years of continuous silicon processing advancements. Critical dimensions can now be precisely controlled for CMOS transistor channel lengths and gate-oxide thicknesses with the latter regulated with fidelity approaching one or two atomic layer thicknesses.

Material advantage

Also, the CMOS manufacturing process benefits significantly from the good material qualities of silicon, which can be self-protecting. Its ability to form a silicon dioxide layer often protects it from unwanted particle contamination and chemical reactions. The same cannot be said for SiGe or GaAs.

Additionally, CMOS manufacturability is further improved by the noticeable absence of high-energy processing steps that can cause defects. As one example, in advanced SiGe processes, molecular beam epitaxy (MBE) is one of the techniques used to achieve precise dimension control. Unfortunately, this technique introduces wafer defects on the order of 100/cm2 - two to three orders of magnitude (100 to 1,000 times) higher than that seen on a mass volume, 200-mm CMOS silicon wafers.

But the manufacturability comparisons between CMOS and all other comers might not be fair. The key observation here is that non-silicon substrates, the starting materials on which circuits are fabricated, are not as mechanically robust and chemically resilient as silicon substrates used in CMOS. As a result, GaAs wafers are usually limited to a mere 75 mm in diameter because they are unable to sustain the wafer-level bending forces and stresses common to 200- and 300-mm silicon wafers now being used in mass production. As such, products made with these technologies cannot leverage the economies of scale afforded by CMOS products and are inherently more expensive.

Linking the home

A high-performance, cost-effective wireless solution that can support the data rates required for rich mobile multimedia applications can only be addressed by 5-GHz systems. The "cleaner air" at 5 GHz along with the fact that these new WLAN technologies can be completely implemented in every-day, high-volume CMOS technologies allow 5-GHz WLANs to be very cost-effective. Combine this with the higher 54-Mbps transmission speeds, and it is easy to understand why 5-GHz WLAN technology is a critical link in fulfilling the promise of delivering a high-speed Internet experience to the home.

About the Author

James C. Chen is a product manager with Atheros Communications. He holds a Ph.D. in electrical engineering from the University of California at Berkeley. He can be reached at jamesc@atheros.com.