Wireless-LAN technology is not keeping up with the demands of home video networks, interactive gaming, voice-over-Internet Protocol, transportation system access and last-mile data distribution. Products attempting to use WLAN for video distribution have fallen well short of consumer expectations for link range and picture quality, so adoption has faltered. And IT managers report that capacity in enterprise networks is poor compared with Ethernet.

The IEEE 802.11n Task Group was formed to increase WLAN throughput above 100 Mbits/second while maintaining backward compatibility with existing 802.11a/b/g devices, raising spectral efficiency to at least 3 bits/s per hertz and maximizing range. Because throughput is specified at the media-access control (MAC) layer, the radio physical-layer data rate must be near 150 Mbits/s. These lofty goals for the next generation of WLAN chip sets are now achievable with proven technology.

Multiple-input, multiple-output antenna technology is a compelling answer to 802.11n's requirements. MIMO essentially multiplies data throughput, with a simultaneous increase in range and reliability, without consuming any extra frequency spectrum. And it can be used in a fashion that maintains backward compatibility to the installed base of WLAN products. The 802.11n standard undoubtedly will include the benefits of MIMO technology in conjunction with other coding and modulation advances.

The conventional approach to increasing spectral efficiency is to decrease coding redundancy, increase coding gain or increase modulation density. A simple theoretical capacity analysis shows that these techniques alone are not enough to satisfy the needs of 802.11n.

Cost-effective, highly integrated WLAN radio transceiver chip sets can provide approximately 25 dB of signal-to-noise ratio (SNR) at the range limit of the present 54-Mbit/s WLAN modes. Market feedback indicates range performance for existing 54-Mbit/s modes is insufficient, so to improve range a next-generation system must require less than 25 dB SNR. The Shannon capacity limit for conventional systems in a perfect 25-dB SNR channel is 8.3 bits/s/Hz. Using the 802.11a/g orthogonal frequency-division multiplex (OFDM) signal bandwidth of 16 MHz (20-MHz channels, 2-MHz guardbands), the theoretical data rate limit is 133 Mbits/s. Even this practically unachievable theoretical limit does not meet .11n throughput requirements after MAC overhead is considered.

The 'more spectrum' drug
The most brute-force approach to increasing wireless data rate is to use more frequency channels to increase modulation rate. This "channel bonding" ap-proach will not meet the needs of WLAN consumers, for many reasons.

First, while channel bonding increases data rate, it decreases range for the same transmit power. Second, channel bonding robs channels from other systems that operate nearby. To avoid interference, an access point in the vicinity of another AP must occupy different frequency spectrum; otherwise, the performance of both systems degrades. With three nonoverlapping 20-MHz channels, the 2.4-GHz band (802.11b/g) will not support even two noninterfering channel-bonded links.

Such interference is a serious detriment in usage scenarios like apartment dwellings, moderately dense suburban neighborhoods and multiaccess-point business or service-provider networks. Channel-bonding proponents argue that consuming more frequency channels is viable because there are "plenty available" in the 5-GHz bands. This claim neglects projected WLAN growth rates and applications. Virtually all channels in the 5-GHz bands will soon become used, making spectrum efficiency critical in densely populated settings where most WLAN consumers live and work. Responsible design of modulation standards that conserve spectrum is crucial to ensuring a positive WLAN consumer experience.

The third problem with channel bonding is that it becomes very difficult and inefficient to support backward compatibility for the massive installed base of 802.11a/b/g devices. To do so will require inefficient signaling. Finally, channel bonding violates government regulations in Japan and some European nations.

Enter MIMO
MIMO answers the question of how to achieve higher data rates with longer range, backward compatibility, global regulatory compliance, all without using more frequency spectrum. Eco-nomical MIMO WLAN systems became a reality in 2003, setting a new wireless course.

MIMO systems use multiple transmit and receive antennas. A high-rate data stream is divided into multiple lower-rate streams, each of which is modulated and transmitted through a different antenna at the same time using the same frequency channel. Because of multipath reflections, each receive antenna output is a linear combination of the multiple transmitted data streams. The data streams are separated at the receiver using algorithms that rely on estimates of all channels between each transmitter and each receiver. In addition to multiplying throughput, range is increased because of an antenna diversity advantage, since each receive antenna has a measurement of each transmitted data stream.

Wireless capacity theory derived in the mid-1990s (some of which was contributed by the authors) extended Shannon's limit to the case of MIMO systems transmitting in multipath channels. This theoretical result proved that the data rate capacity and range of MIMO wireless systems can be increased virtually indefinitely-without using more frequency spectrum-by increasing the number of transmit and receive antennas to exploit multipath. Many experimental measurements that corroborate this theoretical work have since been widely reported.

The first low-cost, mass-market radio chip set implementation of MIMO OFDM technology was produced in 2003. It provides a highly integrated WLAN radio that works at 72, 96 or 108 Mbits/s in either the 2.4-GHz band or any of the 5-GHz bands available worldwide. Since all MIMO data packets are transmitted on one WLAN channel and in compliance with the 802.11 MAC, a mix of MIMO and conventional WLAN devices are seamlessly supported packet by packet. When working with conventional devices in 802.11a/b/g modes, the system uses single-input, multiple-output multiantenna signal processing to extend link range and reliability for all the standard data rates.

MIMO theory
A rate range plot of standard 802.11a products vs. a recently announced MIMO product in a multipath environment has shown that both the MIMO system and the conventional system transmit the same aggregate RF power. From these results, it is clear that MIMO allows for higher data rates and longer ranges than single-antenna systems.

MIMO's detractors cite higher cost and dc power consumption as barriers to overcome for mass adoption. In reality, with currently available RF IC integration density and wafer costs, the very first implementations of 802.11a/b/g-compatible MIMO have been introduced to the market at cost points that are less than what the original 802.11b-only and 802.11a-only chip set costs were when they were introduced. It is also a safe bet that MIMO will become even more cost-competitive in WLANs, since further generations invariably decrease in price.

As for power consumption, the most critical design consideration in WLAN RF systems is the power draw of the transmit power amplifiers. A MIMO system does not rely on more power being transmitted for its performance advantage. Therefore, a MIMO system can be designed to transmit the same total power as a conventional single-transmitter system, except the total output power is supplied by multiple lower-power transmitter chains. Since each lower-power amp has the same (or better) power-added efficiency as a single, larger power amp, properly designed MIMO chip sets' dc power consumption is close to that of single-antenna systems.

V.K. Jones is vice president of ASIC and DSP, Greg Raleigh is president and CEO, and Richard van Nee is director of research at Airgo Networks Inc. (Palo Alto, Calif.).

See related chart