Advertisement
Product Releases
Advertisement

Smart Antenns Benefit WLANs

Mon, 10/31/2005 - 8:55am
Smart antennas with fixed beams and adaptive antenna arrays can provide WLANs with range extension, multipath diversity, interference suppression and increased capacity.
By Dr. Jack Winters

Glossary of Acronyms

A/D — Analog-to-Digital
FER — Frame Error Rate
ISI — Intersymbol Interference
M — The number of antenna elements or beams in a smart antenna
MIMO — Multiple-Input-Multiple-Output
MRC — Maximal Ratio Combining
OFDM — Orthogonal Frequency Division Multiplexing
PCMCIA — Personal Computer Memory Card Industry Association
QoS — Quality of Service
SNR — Signal-to-Noise Ratio
TDD — Time Division Duplexing
Wi-Fi— Wireless Fidelity
WLAN — Wireless Local Area Network

WLANs have become an integral part of both the home and the enterprise with the proliferation of inexpensive, high performance 11 Mb/s products based on the IEEE 802.11b (Wi-Fi) standard. Everywhere you turn, there's an access point to connect to — at home, in the office, at airports, on trains and in retail outlets, just to mention a few.


The WLAN Explosion

The explosion of WLAN technology stems from the easy-to-use systems that are now available. And manufacturers are adding to that growth by building Wi-Fi technology directly into portable computers and communicators as standard equipment. Numerous reports detail the growth of WLANs and Wi-Fi, and all signs point to a continued push toward mobility and wireless connectivity.


However, in spite of this looming ubiquity, realistically, wireless communications systems are limited in performance and capacity by numerous factors, including: limited spectrum, delay spread, co-channel interference and multipath fading. These factors result in the end-user experiencing the gamut in QoS, from non-existent to full speed ahead. End-users can be too far away from an access point, behind a wall, in a "dead spot" or working off a laptop. In fact, network managers often find that WLANs fall short of expected range when actually deployed.


Far too often, marketing materials stem from a vendor's specifications in ideal surroundings. For example, a manufacturer may indicate that the wireless system has a range of 300 feet; however, as is common knowledge, obstacles such as walls, desks and filing cabinets can significantly decrease the range and coverage uniformity in some directions.


Smart Antennas to the Rescue

To address this, a technique to overcome shortcomings in legacy products is to implement a new, cost-effective smart antenna technology. Smart antennas can suppress interfering signals, combat signal fading and increase signal range, thereby increasing the performance and capacity of wireless systems. Although smart antennas improve WLAN performance, various types of spatial processing techniques have different advantages and disadvantages in each type of system, so it is important to fully understand each technique.


A smart antenna is generally defined as a multibeam antenna or adaptive array that tracks users as they move through an environment, and also tracks environmental changes. The two basic types of smart antennas are shown in Figure 1.


Figure 1. Multibeam and adaptive array smart antennas. Click here to enlarge.
Click here to enlarge.

Multibeam antenna— The multibeam (phased array) antenna forms several narrow beams and uses a beam selector to choose the beam for reception that has the highest signal power.


Adaptive array— The signal is received by several antenna elements, each with similar antenna patterns, and the received signals are weighted and combined to form the output signal.


The multibeam antenna is simpler to implement because the beamformer is fixed. This allows beam selection to only be needed every few seconds to account for user movement, whereas the adaptive array must calculate the complex beamforming weights at least an order of magnitude faster than the fading rate, which can translate into several Hertz for ordinary users.


Both of these smart antennas can significantly improve the performance of wireless systems by providing a higher antenna array gain (a gain of M is achieved with M beams or an array of M antennas). A key difference between the two techniques is that the multibeam antenna provides the full gain of M only in a line-of-sight system. In non-line-of-sight systems, such as is typical in an indoor environment, or outdoors with significant multipath, the received signal can arrive via many paths and angles, so the signal energy may not be concentrated in one beam; consequently, the gain will be reduced.


However, the adaptive array enables the gain of M to be achieved regardless of the environment (line-of-sight or multipath). The adaptive array also provides an M-fold diversity gain (this will be detailed further on in the article) to mitigate the effects of multipath fading. Multibeam antennas cannot provide this benefit. Lastly, by using M adaptive antennas at both the transmitter and receiver, MIMO techniques can also be used to provide an M-fold increase in data rate. This is what is being proposed for the next generation of WLANs with the IEEE standard 802.11n, where two or more transmit and receive antennas will most likely be used with MIMO to provide data rates in excess of 100 Mb/s (some 802.11n proposals consider data rates in excess of 500 Mb/s).


Adaptive Array Benefits

For the adaptive array to provide the full diversity gain against the multipath fading, the fading at each antenna element should be nearly independent. However, the correlation of the fading among the antennas can be as high as 0.7 before the degradation exceeds 1 dB. In practice, the correlation can easily be kept below this level through the use of the following diversity types:

Spatial diversity— the spatial separation of the antennas. In a severe multipath environment, such as indoors or around a handset, only about a λ/4 spacing is needed to obtain low fading correlation.

Polarization diversity— the use of dual polarization (horizontal and vertical) allows one physical antenna to be used for two input signals (using different feeds for each polarization).

Pattern diversity— The use of antenna elements with different patterns.


The combination of these three diversity types permits the use of a large number of antennas even in a small form factor, such as a PCMCIA card or handset, with nearly ideal performance.


There are many combining techniques for the adaptive array. The simplest and most common (used in most 802.11b systems and many 802.11a/g systems) is selection diversity. Here, the antenna with the highest received signal power is selected for the output signal. This technique is currently used on many WLAN receivers. However, this technique does not use all the received signal power and therefore is limited in its decibel gain improvement over a single antenna.


Figure 2. Smart antenna appliqué with an 802.11 transceiver. Click here to enlarge.

The optimum technique for range increase and coverage uniformity is MRC, whereby each antenna signal is weighted and combined to maximize the output SNR. The beamforming weights are then the complex conjugate of the channel transfer characteristic, i.e., the received signals are co-phased and their gain adjusted based on their received signal strength. This technique provides a gain of M along with a diversity gain of M in a Rayleigh fading environment. Note that one method for generating the beamformer weights is simply to correlate the output signal with the received signal on each antenna. Such a method is referred to as a blind technique, as the adaptive array will basically maximize the SNR of any received signal. The advantage is that the same beamformer can then be used with any type of wireless signal, e.g., 802.11b/g/a, and demodulation of the received signal is not required.


In a multipath environment, when the difference in the propagation delays of the various paths between the transmitter and receiver (the delay spread) becomes comparable to the symbol period, frequency-selective fading results in ISI, which degrades performance. To overcome this impairment, typically temporal equalization or OFDM at the receiver is used. Both of these techniques are already implemented in the different flavors of 802.11. In this case, spatial processing (i.e., the adaptive array previously described) followed by temporal equalization or OFDM demodulation is not optimum, and joint spatial-temporal processing is needed for optimal performance, although close to optimum performance with the former technique can often be achieved if the delay spread is small.


Analog or Digital Processing

The weighting and combining for smart antennas, as well as the weight generation, can be done with either analog or digital processing. For digital processing, the analog signal received by each antenna is downconverted to baseband and then converted to digital. The digitized signals are then used to calculate the weights, and the digital signals are combined to produce the digital output signal for signal demodulation.


With analog processing, the received RF signals are weighted (i.e., phase-shifted and gain-adjusted) and then combined at RF to produce an RF output signal. In this case the weights can be calculated by correlating the downconverted-to-baseband received and output signals. This correlation can be done either in analog or with A/Ds and digital circuitry. Note that digital signal processing for weighting and combining is easier to implement than analog processing, but requires a costly number of M A/D converters. Digital signal processing for the weight calculation is also easier to implement, but has a problematic greater latency in the weight calculation.


Approaches for Incorporation of Smart Antennas

The three current physical layer standards for 802.11 are 802.11b, 802.11g and 802.11a. All three techniques use TDD, where the same frequency is used for transmission as for reception.


To incorporate smart antennas into 802.11 systems, one method is for a complete redesign of the 802.11 transceiver with smart antenna processing as a integral part. This method allows for the optimal performance improvement with M antennas, as well design optimization. Conversely, it also has the following substantial disadvantages:

Substantial redesign required— Every chip manufacturer would need to undergo major redesigns without the desired gains being completely verified until the chip is completed.

Broad cost increases— A redesign would lead to a broad increase in the cost of the 802.11 chipset for smart antenna gains that may not be valued by all users. In a price-sensitive market, any undervalued increase in cost will lead to lost sales by consumers who may not fully understand or grasp whether they need the gains of smart antennas, and who solely purchase on price. Chip vendors can avoid this risk by creating two versions of their chips: one with and one without smart antenna functionality. However, if the chip manufacturers make two versions of their chips without at least a doubling of volume, they risk losing the price advantage that comes through volume, a key to the successful economics of the semiconductor business.

Wireless system replacement for consumers— Consumers who purchased WLANs without smart antennas, and then found that they needed them, would have to completely replace their access points, client cards or both.


The benefit of smart antennas for improving WLAN performance is clear, so it is essential to find a cost-effective method to incorporate them into WLANs. A better method of incorporating them into 802.11 systems uses an appliqué approach. The appliqué provides a seamless enhancement of smart antennas within the standards. It provides a better RF signal into existing transceiver chips with minimal or no modification of the transceiver. An example of this appliqué (with four antennas) is shown in Figure 2. The appliqué takes the four signals from the receive antennas and combines them to generate an input signal to the transceiver. The appliqué then attaches to the existing transceiver at the antenna port to provide a better received signal and transmit the signal more effectively.


As discussed previously, the smart antenna should be an adaptive array rather than a multibeam antenna so that an array gain (6 dB with four antennas) can be achieved in all environments, along with a diversity gain in multipath environments. Any type of diversity antennas can be used with such a system, e.g., the four antennas can be two dual-polarized antennas in the same form factor as existing PCMCIA cards (as they typically use two antennas currently). The four physical antennas can be of any type and performance, because smart antennas' benefits are additive to the performance of the individual antennas themselves.


Figure 3 shows a block diagram of the receiver for the appliqué operating at RF. Variable gain amplifiers and phase-shifters, whose values are determined by the antenna weight generation circuitry, adjust the gain and phase, respectively, of the signals received by each antenna. The combined signal is then output to the transceiver.


Figure 3. Appliqué block diagram for the receiver. Click here to enlarge.

Also, as an appliqué, the output signal needs to be an RF signal. Although digital processing would be easier to implement than analog signal processing, as an appliqué this would require baseband processing with four A/Ds and remodulation, which is a much costlier solution. In addition, digital processing introduces latency, which cannot be tolerated in an appliqué that should have no effect other than to increase the receive SNR. Analog processing has no such latency issues.


One technique for the appliqué is blind MRC, which will work for 802.11b, 802.11g and 802.11a, as long as the weights can be calculated in this algorithm within 2 µs after the signal is received, because this is the time allotted for antenna selection in 802.11a/g. This short weight acquisition time requires the use of analog processing.


It is important to note that the use of analog processing in the appliqué means that only spatial processing is practical. This will lead to some degradation in performance over digital spatial-temporal processing with delay spread, but this degradation is generally only a few decibels even in indoor environments with severe delay spread, and the use of less costly analog processing means that more antennas (with even better performance) can potentially be used.


Bi-directional Performance Issues

Because 802.11 uses TDD, if the same antennas are used for transmission as for reception, with the same weights, the same performance can be obtained in both directions with the appliqué on just one device. The only requirements are that the channel hasn't changed between reception and transmission, and the relative delays and gains of each of the receiver chains, as well as the transmitter chains, must be similar. The latter requirement is easily met in a chip implementation. However, because the transmit power amplifiers have peak power limitations, under this constraint, the best performance is achieved by using just the phases of the receive weights for transmission. This type of appliqué is fully standards-compliant, and implementation at either the client or access point gives a performance improvement in any 802.11 system.


The appliqué operates in the receive mode, with the weights adapting until a received signal is detected (at which time the weights are frozen to minimize any effect of weight fluctuation on the transceiver). The weights resume adaptation when the end of the packet is detected. The receiver is turned off and the transmitter is turned on when a transmitted signal is detected from the transceiver.


Although the transmit weights used should be the last receive weights calculated with a single client card and access point, this is not always the case with multiple clients. In this case, MAC information is needed to identify who sent the signal for which the weights have been calculated, and for whom (i.e., what weights are to be used) the transmitted signal is destined. This requires a weight table, along with control signals from the transceiver. An alternative at the access point (where battery life may not be an issue) is to use a bypass power amplifier with a single antenna transmitting enough power to compensate for the receive gain of the smart antenna (with the lower client transmit power). In this case the same range can be provided on the uplink and downlink (although without multipath mitigation and with higher interference from the access point).


Smart Antenna vs. Single Antenna

Figure 4 compares the performance of the four-element smart antenna (baseline) with a single antenna system in 802.11b. Computer simulation results are shown for the FER vs. the SNR for the four data rates of 1, 2, 5.5 and 11 Mb/s. Independent flat Rayleigh fading at each antenna is assumed. At the 802.11 specification of an 8% FER, the required SNR is reduced by 13 dB for all four data rates.


Note that many current 802.11b systems use selection diversity with two antennas. Ideal selection diversity in flat Rayleigh fading achieves a 6.1 dB gain over a single antenna. However, most current 802.11b systems do not see this much gain with selection diversity. They generally do not exceed 3 dB in additional gain.


Figure 4. Comparison of a four-element smart antenna to a single antenna system in 802.11b. Click here to enlarge.

With the four-element smart antenna at both the client and the access point, computer simulation [results generated by Motia (see www.motia.com)] shows a gain of 18 dB over a single antenna system. The gain is less than 26 dB (i.e., twice the 13 dB gain on one side) because the diversity gain is less on the receive side since transmit beamforming provides most of the diversity gain possible. Also, because the weights at the transmitter and the receiver are calculated independently, the gain is less than it would be if they were calculated jointly. With ideal transmit and receive weights, the gain would be 22 dB. Note that we are only 4 dB from ideal, and this also includes implementation loss (i.e., weight estimation error).


With delay spreads typical of indoor environments, the 13 dB is reduced by about 1 dB. For 802.11a/g, the gains are similar, with about a 12 dB gain, although this can be reduced to 10 dB with the delays spreads typical of indoor environments.


The SNR gain with the smart antenna results in a range increase (e.g., a doubling of range for a 12 dB gain with a fourth law power propagation exponent typical for indoor environments). However, this gain can also translate into a data rate increase in 802.11 systems because higher SNRs allow for higher data rates.


Conclusions

In closing, the advantages that smart antenna technology brings to WLANs can be quickly infused into existing and future products using an appliqué approach. With an appliqué, smart antenna investments can be assured to be fully 802.11 compliant, work with any transceiver chipset, be virtually "plug and play" and be applicable to access points and client devices. Once in place, smart antenna technology can enable range extension up to 4X, throughput increases of 100% and more, eliminate dead spots, and reduce power drain during transmission up to 90%.


Together, combining smart antenna technology with an appliqué approach will enable you to be much more cost effective, help speed your products to market faster, and deliver to your customers the performance and range they expect from WLAN technology.


About the Author

Prior to Motia, Dr. Jack Winters was division manager of the wireless systems research department at AT&T Labs-Research. While there, he led research on signal processing, modulation and coding techniques for communication systems, specifically in the use of adaptive antennas in cellular and indoor wireless systems. He is an IEEE Fellow and an IEEE distinguished lecturer in both the communications and vehicular technology societies. He has more than 20 patents. He has published more than 40 journal and 70 conference papers and has made more than 20 tutorial/seminar presentations. Winters was named a New Jersey Inventor of the Year for 2001. He received a PhD in electrical engineering from Ohio State University.

Advertisement

Share this Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading