EETimes.com - Dual Rx boosts WLAN OFDM

Antenna diversity can substantially improve the quality of an orthogonal frequency division multiplexing (OFDM) wireless link in multipath environments. To achieve a diversity gain, multiple antennas can be used at either or both ends of the link, but the use of multiple transmit antennas requires appropriate receiver side signal processing geared to the transmit diversity scheme employed.

On the other hand, with receiver antenna diversity a wireless device is always guaranteed an enhanced reception, as the performance gain comes only from the receiver side. This article discusses the performance advantages of using dual antenna paths and appropriate signal processing at the receiver. While the receive antennas and the receive radio paths must be duplicated in principle to realize receive diversity, implementation strategies exist that allow a sharing of the majority of building blocks and circuits. Thus, the added cost of realizing dual receive antenna paths can be made relatively small. The additional power consumption requirement can also be substantially mitigated by employing a dynamic diversity strategy that turns on the power intensive diversity features only as the link quality drops below a satisfactory level.

In addition to the classical diversity gain that arises from optimally combining the dual reads of a given transmit signal, the dual front end performance gains are discussed in conjunction with various building blocks of an OFDM modem front end. When applied to OFDM systems compliant with the IEEE 802.11a standard, the dual receive antenna scheme yields a substantial overall performance improvement over the conventional single receive path technology in terms of the packet error rate, the achievable range, and the speed and link robustness.

Fig. 1 shows typical frequency responses of a wireless channel subject to the multipath phenomenon. As the rms delay spread increases, the frequency selectivity of the channel becomes more pronounced that is, the amount of fading fluctuates more widely from one frequency zone to another.

To obtain a full receive diversity gain in OFDM, the received signals captured off the dual radio paths are combined for each subcarrier or frequency bin separately. The basic idea is to combine the two signals, before demodulation is performed, in such a way that a larger weight is placed on the signal that experiences less fading. The combining operation is nonlinear in general, with an optimal scheme known as maximal ratio combining. The combining is done on a subcarrier by subcarrier basis. Because of the complexity issues, it is critical to use an efficient algorithm that provides near optimal performance with a considerable reduction in hardware.

Large SNR gain

Fig. 2 shows the packet error rate (PER) plotted against the available average signal to noise ratio (SNR) in the channel, with and without the dual receiver antenna diversity. An rms delay spread of 100 ns is chosen for this experiment. The packet size is 1,000 bytes and the data rate is 54 Mbits/second. The packets are generated based on the IEEE 802.11a standard.

As seen, the gain is large with the dual reception technology and a proper diversity technique. Comparing the curves, the dual Rx technology yields an SNR gain of 8 dB at 1 percent PER. This can also be interpreted as the dual Rx technology being able to reach the same PER as the conventional single Rx scheme with an average SNR of 8 dB less.

It is easy to see that this type of SNR advantage leads to an improved range, since there exists a more or less monotonic relationship between the distance and the signal power. Invoking the well known indoor signal propagation loss model recommended by the International Telecommunication Union with the parametric loss exponent fixed at 3.1, as suggested for 5 GHz indoor environments the 8 dB corresponds to a range improvement of about 80 percent. The range improvement can be considerably larger in more severe multipath environments.

The results of Fig. 2 reflect diversity gain with ideal front end digital signal processing. In reality, the overall PER is a highly sensitive function of the digital front end operation, which includes signal detection, carrier frequency and phase synchronization, as well as frame synchronization. We shall consider next the effect of using dual reception on the performance of these front end processing blocks.

Front end signal processing

Fig. 3 shows an OFDM digital front end consisting of various blocks for signal detect, carrier frequency offset (CFO) estimation/correction, frame synchronization and phase locked loop. The signal detect block is responsible for detecting the arrival of a packet. Minimizing both the misdetection and false alarm probabilities is important in maintaining a satisfactory level of throughput. The operation of the signal detector is based on sensing the known repetitive signal pattern in the preamble.

The presence of random noise as well as interference potentially causes the misdetection/false alarm and degrades the network throughput. With dual reception, the signal can be detected more reliably based on two observations of the same signal that are subject to independent noise.

Fig. 4 shows the detection failure and false alarm probabilities with and without the dual Rx capability. The probabilities are plotted against the threshold level at the preamble pattern correlator output. Improved signal detect performance is evident with the dual Rx technology. Margin analysis on frame synchronization also reveals that the dual Rx technology tolerates a much larger range of fast Fourier transform window positioning errors than conventional single Rx schemes.

It is well known that OFDM performance is highly sensitive to CFO and phase errors. The dual Rx technology provides a notable performance improvement in the presence of CFO and phase noise. Fig. 5 shows PER performance of the conventional single path OFDM and the OFDM based on Bermai's dual Rx technology at different delay spread characteristics. The simulation parameters are a 54 Mbit/s data rate, 1,000 byte packets, a channel SNR of 40 dB, an initial CFO of 40 ppm and a phase noise spectrum at 85 dBc/Hz with cutoff at 100 kHz. Moreover, a relative immunity to phase noise is evident with the dual Rx technology. With the phase noise at 85 dBc/Hz, the single Rx scheme cannot achieve the 1 percent PER. At a 2 percent PER, conventional OFDM can tolerate less than 100 ns rms delay spread, whereas Bermai's dual Rx can withstand up to 180 ns rms delay spread, representing a large improvement in multipath/phase noise capability.

In Fig. 6, PER vs. the amount of phase noise plots is shown for the single Rx and dual Rx schemes. The simulation parameters are the same as in Fig. 5 except that the rms delay spread is now fixed at 50 ns and the spectral height of phase noise varies along the x axis.

The dual Rx technology is seen to be much less sensitive to phase noise than the conventional single path Rx.

These experiments demonstrate that Bermai's dual Rx solution is remarkably robust in the presence of large phase noise.

Editor's Note: Antenna diversity has been around in various forms for many years before finding its way into WLANs, with much of the research being done in the context of cellular basestations. Multiple input, multiple output diversity with an OFDM waveform has become the implementation of choice, giving the most impressive performance in a multipath environment. Recently, Airgo Networks announced a MIMO OFDM imple menation for WLANs that took much of the industry by surprise (see Aug. 18, page 1). Airgo CEO Greg Raleigh will give a workshop on diversity in all its forms, with a special emphasis on MIMO OFDM, at the next Communications Design Conference, scheduled for March 29 to April 2, 2004.

Jaekyun Moon is chief scientist and a co founder of Bermai Inc. (Palo Alto, Calif.) and Younggyun Kim is the company's senior architect.

See related chart