Wireless LAN networks operating at speeds traditionally seen as the preserve of fixed technologies are now on the horizon. Patrick Molin of Agere Systems explains how the goal posts are being shifted by using MIMO technology.
The proliferation of wireless networking in the office, at home, and in public spaces has generated considerable interest in the technology’s potential for driving seemingly ever-higher transmission speeds. The current 802.11b standard delivers 11Mbits/s and emerging standards such as 802.11g and 802.11a are promising to increase throughput to 54Mbits/s. But, the need for speed will continue to grow, and many applications, such as the delivery of HDTV signals for entertainment and hospital communications for medical imaging would benefit from throughput rates beyond the 54 Mbits/s mark. Taking wireless LAN technology beyond the limitations of time and frequency division multiplexing is a new concept under development, one that promises to provide a quantum leap in speed and capacity, bringing wireless capabilities near to those of wired networks.
Late last year, our researchers announced a breakthrough in WLAN technology, demonstrating transmission speeds of 162Mbits/s, roughly 15 times faster than today’s 802.11b wireless data networks. This innovation was achieved not by augmenting the time and the frequency dimensions, which have become increasingly congested, but by adding a third degree of freedom — space. The ability to take existing channels — in this case, three 54Mbit/s channels — and operate them in parallel using multiple transmit and receive antennas, allowed the researchers to demonstrate increased speed and to show that the improvement in data throughput rates increases proportionally to the number of transmit antennas.
Multiple Input/Output
Given the bandwidth restrictions imposed by frequency allocations, the growing congestion from numerous types of devices, and the signal-to-noise ratios in today’s home and office communications environments, our researchers began exploring avenues to provide dramatic improvements to wireless data communications. These engineers and scientists developed the principle of adding a third dimension to the traditional frequency-time continuum: space.
The basic concept of this is to transmit multiple streams of data on multiple transmit antennas, at the same frequency. Typically, multiple receive antennas are used as well since this configuration dramatically improves the overall performance of the system. This principle is called MIMO — Multiple Input Multiple Output.
The use of the MIMO principle for WLAN was first demonstrated in a public forum by Agere at Eindhoven University of Technology in the Netherlands using a platform developed by Agere which combines MIMO with OFDM.
The demonstration, based on 802.11a technology in the 5GHz band, paired three transmit and three receive antennas to achieve 162Mbits/s over a wireless connection. This high data rate was achieved through the MIMO/ODFM transmission combina-tion and even with this enhancement, the technology remains compliant with the IEEE 802.11 standards specifica-tions. Similar rates are expected to be achieved for 802.11g-based transmis-sions using a dual-band radio solution (supporting both 2.4GHz and 5GHz).
MIMO Processing
The demonstration at Eindhoven University provided a proof point and verified that a data stream could be time-division multiplexed across multiple spatial pipes, with each one operating at the same frequency. While the three transmissions do interfere, the three streams can indeed be separated through the use of MIMO processing on the receiving side.
While the most basic configuration for MIMO utilizes two transmit antennas and one receive antenna, the signal-to-noise ratio post-MIMO processing is low in this situation, resulting in inferior performance and a sub-optimal radio architecture. The research team therefore decided to use paired antennas: three transmit and three receive to boost overall system throughput, achieve an acceptable signal-to-noise ratio, and validate the signal processing algorithms. Theoretically, MIMO does not require the number of receiving antennas to equal the number of transmitting antennas, but the MIMO signal processing algorithms simplify when the variables are matched. The MIMO principle is illustrated where three transmitters S1, S2, and S3 are transmitting across a fading wireless medium with transmission properties Hxx to three receivers R1, R2, R3.
Hence, an essential aspect of MIMO processing is the determination of the medium transmission matrix Hxx. In the IEEE 802.11a standard, a preamble training sequence is sent by the transmitter, which aids the receiver in estimating the channel matrix Hxx. During payload reception, the matrix Hxx is refined using known pilot tones, that are also sent by the transmitter.
The MIMO concept suggests that bandwidth can be increased linearly as the number of transmitting antennas increases. If three transmitters can provide 162Mbits/s bandwidth, then six transmitters should be able to provide 324Mbits/s. Indeed, the un-mixing process to separate out signals is acheived usning simple linear matrix algebra well within the capabilities of today’s processors.
In theory, MIMO could work with a single receive antenna, but in principle, the bit-error rate improves dramatically when the number of receive antennas is increased. And in practice, the algorithms are less complex when the number of receivers equals the number of transmitters. When the number of receivers is greater than the number of transmitters, the robustness of the algorithm is further improved at the expense of additional hardware and computational complexity.
Independent of the number of receive antennas, deciding on the number of transmitters is one of the key parameters in any practical application. One key factor is the number of ‘reflectors’ or ‘scatters’ in the medium. The medium must be dispersive for MIMO to be effective; MIMO is less efficient when only direct line-of-sight transmission exists. For example, in a three transmitter/three receiver configuration, it is permissible to have one line-of-sight transmission as long as there are two independent reflections that allow the transmission matrix Hxx to maintain ‘full rank.’ This constraint of independent reflections represents an effective limit to the throughput enhancements possible with MIMO — a ten-by-ten configuration would require nine independent reflections; a scenario with limited probability of occurrence.
Our researchers believe that an 8×8 array is the effective upper limit to practicality; and a 3×3 configuration will work in nearly all home, office, and industrial environments and ultimately provide a cost-effective solution that meets both market and application bandwidth needs.
Another practical limit to MIMO deployment is that the inherent need for multiple antennas and receiver/ transmitters will further tax the power consumption, heat dissipation, size, battery life, and cost constraints for portable devices. Therefore a balance must be struck to optimise the overall price/performance metrics of the system to yield optimum user experiences.
Independent force
The MIMO principle is independent of frequency bands and modulation schemes. In fact, for 802.11g (2.4GHz) where bandwidth is limited and congestion is higher, the ability to add the third dimension of space to time and frequency (and thereby improve the system throughput) may prove to be even more valuable than at 5GHz. The demonstration at Eindhoven selected OFDM at 5GHz since 802.11a is already a ratified IEEE standard. MIMO, however, could also be applicable to GSM/GPRS systems, although the current power requirements make it less attractive to apply multiple antennas in portable cell phones, since the market attention for mobile phones is focused on decreasing size and weight and increasing battery life.
MIMO futures
While mass deployment of MIMO-based systems for the consumer and enterprise segments is probably several years away, it is expected that improvement and refinement of the systems will continue to take place in the interim. A MIMO-OFDM system would require IC chips designed to perform multiple frequency down-conversions, OFDM demodulation, and MIMO interference cancellation. Such processing is possible with today’s ASICs, but the power consumption may yet be too high for many applications in the near term. However, MIMO does demonstrate the future potential for WLAN technology to deliver bandwidths typical of today’s wireline implementations (100 Mbits/s). Such a performance breakthrough would have implications for 3G wireless hotspots, public wireless data access, in-home delivery of HDTV signals, and many other applications in consumer, medical, industrial, and enterprise data communications.