Solving Multipath Problems in Indoor WLAN Designs
Engineers must find the appropriate channel model to determine performance benchmarks and to properly analyze the link.
By: John Sakatselis, Senior Manager of Applications Engineering, Harris Semiconductor
Contents
•Delay spread
•Multipath modeling
•Benchmark proposals
•Multipath solutions
•More complex solutions
Since the acceptance of the IEEE 802.11 standard, companies throughout the world quickly have been installing wireless local area network (WLAN) systems in office and other indoor environments. During this installation, however, many questions arise for these engineers. The most common deals with multipath interference.
Multipath is created by the returned reflections of the original signal that are combined at the receiver. These reflections occur when the signal bounces from the various surfaces within the indoor environment. These reflected signals arrive at the receiver with some time delay, which depends on the reflection path that they have followed. As a result, the received signal is corrupt by the delayed versions of the original signal as they randomly arrive and combine at the receiver.
The analysis of multipath involves primarily modeling of the geometry of the indoor configuration, the propagation path loss, and the time delay spread of the reflected signals. Delay spread statistics, however, are the dominating parameters when it comes to analyze and to design mitigation to multipath.
Delay spread
In WLAN applications, delay spread is proportional to the size of the particular indoor area. For example, the office and home environments have a smaller delay spread value than a warehouse or retail space.
Figures 1 and 2 provide a visual look at the effects of multipath fading in a WLAN environment by representing the impulse response of a particular channel. Figure 1 shows the impulse response for a 100 ns root mean square (RMS) delay spread example while figure 2 the response of a 300 ns delay spread example.
Note: in both examples, the time axis is normalized but a representative data rate can set to 11 Mb/s to meet the IEEE 802.11 specifications for high speed operation in the 2.4 GHz band.
Multipath modeling
Many questions arise during the design of a WLAN physical layer. The most common question, however, deals with determining the appropriate channel model for performance benchmarks and link analysis.
There are many academic papers describing channel models for indoor WLAN systems. In addition, many standards bodies have adopted models for these indoor wireless devices.
In general, most of the models developed for indoor WLAN environments assume that the average received multipath power is an exponentially decaying function of the time delay. In addition, these models presuppose that the amplitudes of the individual multipath components themselves are Rayleigh distributed about the average value. Thus, these models assume that the multipath components are equidistant on the time delay axis and that the phase of these multipath components is uniformly distributed.
Similar to other standards bodies, the IEEE 802.11 committee has defined models for indoor WLAN operation. These models, which are designed for 2.4 and 5 GHz waveforms, follow the same basic technical baseline as many other models. The primary difference between the IEEE model and various other models revolves mainly around the instantaneous power delay profiles.
The IEEE 802.11 model uses more paths, which can result in optimistic results when the link is evaluated for antenna diversity performance. When it's time to evaluate the delay spread tolerance for the link, the multiple paths used in the IEEE 802.11 model do not show much different results relative to other available channel models. Thus, the delay spread in 802.11 systems is independent of the number of paths used. This delay spread is more coupled to the actual power level statistics of the components as well as the time delay itself.
Benchmark proposals
Using these channel assumptions above, the IEEE 802.11 committee has benchmarked various physical layer proposals. A major decision point for the direction of the group had to do with understanding the multipath tolerance requirements that the physical layer must exhibit. A number of studies were presented showing results of the delay spread requirements for indoor areas of interest to IEEE 802.11 WLAN target applications.
Table 1 shows the requirements as summarized in one of the discussions. These results were simulated using the channel model baseline of the previous discussion.
In the above table, two delay spread values are given. One is the median delay spread value, which is the 50% value. This implies that 50% of the paths have a delay spread value less than the median.
The other column lists the maximum delay spread value expected for each indoor area. This maximum value is helpful in understanding the worse case requirement that the system needs to tolerate.
Multipath solutions
After understanding the multipath tolerance requirements for the various applications, the next step is to define the appropriate implementations that can meet them. The IEEE 802.11 committee spent several sessions evaluating the feasibility of potential implementations that address the multipath problem.
Determining multipath solution was not an easy task for the IEEE 802.11 committee. Today's wireless market calls for low cost and low power implementations. Therefore, IEEE faced the challenge of finding sophisticated equalization schemes that would not lead to power and cost inefficiencies.
The IEEE 802.11 committee evaluated several implementation steps to mitigate the multipath issue starting from no equalization at all, to using more complex schemes. This line of thinking helped the committee present inexpensive alternatives when appropriate.
The lowest complexity solution is one that does not utilize any equalization but attempts to use antenna diversity as a measure to defend against multipath. This implementation does not require an automatic gain control (AGC) and, thus, can translate into an implementation that saves cost in both the RF/IF as well as the baseband processing design areas due to the lack of the AGC and the equalizer.
More complex solutions
Although the above implementation is sufficient for some WLAN applications, it is not the best approach for all WLAN set ups. In fact, the low complexity solution will suffer in environments of high delay spread.
In more complex setups, engineers can turn to a Rake receiver to defend against multipath problems. A Rake receiver is simply a matched filter at the front end. This filter is matched to the channel and it is compensating the waveform from channel corruption.
Engineers can also employ a sliding decision feedback equalizer (DFE) in complex designs. This equalizer consists of feed forward (FF) and feedback (FB) taps designed to subtract the channel corruption from the received signal.
In highly complex designs, engineers can complement the DFE and/or Rake with intersymbol interference (ISI) and interchip interference (ICI) improvement mechanisms to realize equalization schemes that can tolerate the higher requirements.
Table 2 lists the various combinations of implementations with the corresponding expected performance. The data on table 2 is a sample of typical performance expectations (simulated) as discussed during the IEEE 802.11 standard development process for the 2.4 GHz 11 Mb/s physical layer.
Note that the table lists the RMS delay spread for packet error rates (PER) of 10%. This number was chosen by IEEE 802.11 to keep consistency for comparison reasons. Furthermore the 10% PER is considered as an acceptable rule of thump for the IEEE 802.11 protocol for a good number of applications. This does not imply that the 10% PER number is suitable for all potential IEEE802.11 deployments.
Table 2 also shows the Eb/No value required so that enough white noise is added to cause a 20% PER. The Eb/No is between 7 to 10 dB for the 11 Mb/s waveform at a 10% PER when no multipath is applied.
Engineers can use the data from this table as a guideline to WLAN physical layer design for indoor applications.
About the author…
John Sakatselis is the Senior Manager of Applications Engineering at Harris Semiconductor.
References
1. N. Chayat, "Tentative Criteria for Comparison of Modulation Methods", IEEE P802.11-97/96.
2. Richard Van Nee, "Delay Spread Requirements For Wireless Networks in the 2.4 Ghz and 5 Ghz Bands." , IEEE P802.11-97/125, Nov. 1997.
3. C. Andren , J. Boer, R. Van Nee, M. Webster, "Harris/Lucent TGb Compromise CCK (11MBPS) Proposal", IEEE P802.11- 98/246a, July 1998.