Performance Considerations
When WLANs first became available in the early 1990s, primary applications were wireless bar code solutions for needs like inventory control and retail price marking. Data transfers for these types of applications do not demand very high performance. In fact, 1 Mb/s data rates are generally sufficient to handle the transfer of relatively small bar codes for a limited number of users.
Today, enterprises are deploying WLANs for larger numbers of users with needs for corporate applications that involve large file transfers and wireless telephony. In addition, there are often needs for supporting a high density of users in a smaller area, such as a conference room. The need for higher data rates and techniques to improve performance of WLANs is becoming crucial to support these types of applications. To get that extra performance when designing a WLAN, consider the following elements:
- Throughput versus data rate
- Radio frequency bands
- Transmit power settings
- Transmission channel settings
- Data rate settings
- Antennas
- Amplifiers
- Obstacles
- Radio signal interference
- Channel width settings
- Signal coverage
- Fragmentation settings
- Request-to-send / clear-to-send (RTS/CTS) settings
- Multicasting mechanisms
- Microcell deployment strategies
Throughput Versus Data Rate
The data rate of a signal is based on the time it takes to send information and overhead data bits when transmitting. Therefore, the aggregate data rate (throughput) is actually much lower because of delays between transmissions. Data rate mostly affects the delay performance of a WLAN. The higher the data rate, the lower the delay when sending data from one point to another. As a result, higher data rates can increase the capability to support a larger number of users. Data rate alone is not a good measure for the performance of the system, though. You also need to consider the effect of overhead bits, waiting times to access the medium, and the format of the data being sent.
Actual throughput is a much better indicator of the performance of a WLAN because it provides an indication of the time it takes to send information. Throughput is the flow of information over time. It is important to not confuse throughput with 802.11 data rate, which is the speed that the data bits in individual 802.11 data frames are sent. Because there is idle time between 802.11 data frames and retransmission due to noise, the throughput is always less than the data rate. Throughput, however, provides a more accurate representation of the delays that users experience because they are concerned with how fast information is sent, not 802.11 frames. It is possible to have a very high 802.11 data rate and still have throughput that is relatively low. This can occur, for example, when the user is close to an access point but there are many other users actively accessing the WLAN from the same access point or if there is substantial radio signal interference present. In these cases, the WLAN carries the information at a much slower rate. As a result, be sure to focus on specifying throughput, not 802.11 data rates as the basis for performance.
Radio Frequency Bands
As mentioned earlier in this chapter, you can choose to use 2.4-GHz or 5-GHz (or both) 802.11n bands. In addition to the impact this selection has on range, the choice of frequency band also affects performance. The 5-GHz band includes much more spectrum (and corresponding channels) as compared to the 2.4-GHz band. There are many more overlapping channels in the 5-GHz band as compared to the 2.4-GHz band. In addition, the 5-GHz band is usually relatively free from RF interference sources. As a result, the 5-GHz band offers much greater capacity. Keep in mind, however, that the 5-GHz band, as explained earlier, might provide less range as 2.4-GHz deployments.
Transmit Power Settings
For a constant range, increasing the transmit power of an 802.11 radio increases performance in the outward direction. As the transmit power increases, communications at a particular location will be possible at greater data rates. The basis for this is that increasing of transmit power improves the SNR at a particular location, which allows the receiving radio to decode signals at higher data rates.
For example, as shown in Figure 11-8, increasing the transmit power of the access point by 6 dB causes a 6 dB increase in the signal strength and corresponding SNR throughout the coverage area. In the case shown in Figure 11-8, the client device associated with the access point set to higher transmit power has a SNR of 21 dB, which is significantly higher than it would be if the transmit power of the access point were set to a lower level. The higher signal level (and SNR) allows the client device to receive 802.11 signals at higher data rates.
)
Figure 11-8 Transmit Power Increases Provide Higher Data Rates at Specific Points Due to Corresponding Increases in SNR
This increase in data rate applies to the communications in only one direction, which is the outward path relative to the radio with increased transmit power. The increase of transmit power of an access point, for example, only improves the data rate of the 802.11 data frames being sent from the access point to the client radios. To improve the overall communications of 802.11 signals, you may need to increase the data rate settings on the access points and the client radios. As discussed previously in this chapter, client radios may have considerably less transmit power as the access points. Therefore, you will likely need to increase the transmit power on the client radios to see any improvement in data rates.
Transmission Channel Settings
As with range improvements, transmission channel settings can impact performance as well if it is possible to select a channel that avoids radio signal interference. Figure 11-9 illustrates an example of this concept, where testing has shown that the noise level for channel 1 is 5 dB lower than the noise level for channel 11. Because of the lower radio frequency interference corresponding to channel 1, the resulting SNR for channels at the client device is 15 dB, which enables the applicable client radio to process 802.11 signals at a higher data rate.
)
Figure 11-9 Transmission Channel Changes Can Provide Greater Performance by Lowering Noise Levels
Data Rate Settings
If you need to deploy a high-performance WLAN, consider configuring data rate settings to higher fixed values. This forces operation at a higher specific data rate and avoids transmissions at lower data rates, which would negatively impact overall performance. Just keep in mind, however, that using higher data rate settings will significantly reduce range. Also, as explained earlier in this chapter, data rate settings impact communications in only one direction. For example, setting the access point to 54 Mb/s causes the access point to transmit all data frames at 54 Mb/s, but the client radios may be transmit at different data rates depending on their data rate settings. As a result, you must configure the data rate settings in client radios (which might not be feasible) to realize benefits of using higher fixed data rate settings to improve performance.
Antennas
The use of higher-gain antennas increases communications performance in both directions between access points and client radios. For example, as Figure 11-10 illustrates, if you replace a standard 2 dBi antenna with one having 6 dBi, the signal level and corresponding SNR at the client radio at a specific location will increase. The higher signal level and SNR allows the radio in the access point and client radio to decode signals at higher data rates. A higher degree of antennas diversity has a similar affect on performance. Similar to using higher-gain antennas or diversity to improve range, be sure to take into account different antenna gain and diversity with actual propagation testing in the target operating environment to determine the lowest overall cost of deploying the network.
)
Figure 11-10 Higher-Gain Antennas Boost Performance Throughout the Coverage Area
Amplifiers
In addition to improving range, an amplifier can also increase performance. Because an amplifier increases the signal strength and corresponding SNR throughout the coverage area, a client radio at a specific point is able to decode the 802.11 signals at a higher data rate. For example, as shown in Figure 11-11, without an amplifier at the access point, the signal strength and SNR at the client device is -75 dBm and 15 dB, respectively. After installing an amplifier on the access point, the signal strength and SNR at the client device increase to –69 dBm and 21 dB, respectively. As a result, in this example, the client radio is able to support reception of data frames at a higher rate than without the amplifier because of the higher signal level and SNR. Also for similar reasons, because the amplifier's receive gain, the access point will also be able to receive 802.11 signals from the client radio at a higher data rate.
)
Figure 11-11 Amplifiers Improve Performance Throughout the Coverage Area
Radio Signal Interference
By reducing radio signal interference, it is possible to increase performance at specific points within the coverage area. For example (see Figure 11-12), a reduction of noise levels by 6 dB (possibly by eliminating the operation of other radio equipment) will raise the SNR at points throughout the coverage area by 6 dB. This enables the access points and client radios to successfully decode 802.11 signals at higher data rates.
)
Figure 11-12 Lowering Radio Signal Interference Increases Performance at Specific Locations
Channel Width Settings
The 802.11n standard provides two different channel width settings, 20 MHz and 40 MHz. The wider 40-MHz channels support much higher performance as compared to 20-MHz channels. For 2.4-GHz implementation, however, it is not wise to configure 40-MHz channels. As discussed in other parts of this book, there is not enough spectrum in the 2.4-GHz band to support multiple 40-MHz channels without significant inter–access point interference. As a result, if you choose to implement the 2.4-GHz band, the only practical option is 20-MHz channels.
The 5-GHz band, however, has a much greater amount of spectrum, with plenty of room for 40-MHz channels. Therefore, seriously consider configuring the network for 40-MHz channels. To accommodate 40-MHz channels, the client radios must be 802.11n-compliant. As a result, it might not be possible to implement 40-MHz channels if the network must accommodate legacy (802.11b and 802.11g) client devices. A way of still taking advantage of 40-MHZ channels in this case, as expressed in other parts of this book, is to configure legacy radios to connect to the WLAN over 2.4 GHz (with 20-MHz channels) and have only 802.11n devices connect via 5 GHz (with 40-MHz channels).
Signal Coverage
As the basis for providing good performance, it is important to have adequate signal coverage throughout the required coverage areas. In areas that have weak signal coverage, the signal level and corresponding SNR will be relatively low. The 802.11 radios might still be able to decode the signals and successfully communicate, but the data rate may be fairly low. To ensure that signal coverage is good enough, perform a proper wireless site survey as described in Chapter 15.
Fragmentation Settings
The use of 802.11 fragmentation can increase the reliability of 802.11 data frame transmissions in the presence of radio frequency interference, which improves the throughput of the network. Because of sending smaller frames, corrupted bits in the frame due to radio frequency interference are much less likely to occur. If a frame does receive corrupted bits, the source station can retransmit the frame quickly.
The fragment size value can typically be set manually on access points and client radios between 256 and 2048 bytes. This value is user controllable. In fact, you activate fragmentation by setting a particular frame size threshold (in bytes). If the frame that the access point is transmitting is larger than the threshold, it will trigger the fragmentation function. If the packet size is equal to or less than the threshold, the access point will not use fragmentation. Of course, setting the threshold to the largest value effectively disables fragmentation.
A good method to find out whether you should activate fragmentation is to monitor the WLAN for retransmissions. If very few retransmissions are occurring, do not bother implementing fragmentation. The additional headers applied to each fragment will likely dramatically increase the overhead on the network, which will actually reduce throughput. If you find a relatively high percentage of retransmissions (greater than 5 percent) and the presence of radio signal interference or weak signal levels is likely causing the retransmissions, try using fragmentation. This can improve throughput if the fragmentation threshold is set to the optimum value, which is where the throughput is maximum.
If the retransmission rate is relatively high, start by setting the fragmentation threshold to around 1000 bytes, and then tweak it until you find the best results. After invoking fragmentation, follow up with testing to determine whether the number of collisions is less and that the resulting throughput is better. You should try a different setting or discontinue using it altogether if the throughput drops (even if you have fewer retransmissions).
The issues with using fragmentation to improve performance is that interference driving the retransmission rate may change over time or as users roam throughout different portions of the signal coverage area. On one day, for example, the presence of significant radio signal interference may cause the retransmission rate to be around 20 percent. In this case, you may find that a fragmentation threshold of around 800 bytes maximizes throughput. On another day, the interference may go away, resulting in a drop in the retransmission rate to only 1 percent or 2 percent. Without changing the fragmentation threshold, throughput may actually drop because the cost of additional overhead resulting from the fragmentation mechanism is not providing much benefit of reducing the retransmission rate.
RTS/CTS Settings
The use of 802.11 RTS/CTS can increase the reliability of 802.11 data frame transmissions in the presence of hidden nodes, which improves the throughput of the network. Similar to analyzing the need for fragmentation, a way to gauge whether RTS/CTS will help throughput is to monitor the WLAN for retransmissions. If the retransmission rate is low (under 5 percent), do not implement RTS/CTS. The additional frame transmissions need to implement RTS/CTS will likely dramatically increase the overhead on the network, which will actually reduce throughput.
If the retransmission rate is high, and you find a large number of collisions with users that are relatively far apart and likely out of range of each other (that is, hidden nodes are present), try enabling RTS/CTS on the applicable client radios. After activating RTS/CTS, test to determine whether the number of retransmissions is less and the resulting throughput is better. Because RTS/CTS introduces overhead, disable it if you find a drop in throughput, even if you have fewer collisions. After all, the goal is to improve performance.
Of course, keep in mind that user mobility can change the results. A highly mobile user may be hidden for a short period of time, perhaps when you perform the testing, then be closer to other stations most of the time. If retransmissions are occurring between users within range of each other, the problem may be the result of radio signal interference. In that case, fragmentation might help.
In most cases, initiating RTS/CTS in the access point is fruitless because the hidden station problem does not exist from the perspective of the access point. All stations having valid associations are within range and not hidden from the access point. Forcing the access point to implement the RTS/CTS handshake will significantly increase the overhead and reduce throughput. Focus on using RTS/CTS in the client radios to improve performance.
Bandwidth Control Mechanisms
To provide consistent performance for all users, it might be necessary to implement bandwidth control mechanisms, which divides the total capacity of the network into smaller sizes made available to each user. For example, as shown in Figure 11-13 (case without bandwidth control), a single user (client device A) may download a very large file that requires a few minutes (or even hours), consuming nearly all the capacity of the network. As a result other users, such as client B and C may have very little if no throughput. This uncontrolled use of the network can aggravate users and significantly reduce the effectiveness of the network. A solution to this is to use bandwidth control and configure the access points (or other applicable components) to provide each user a throughput limit, such as 250 kbps each. This level of performance, based on the total users and utilization of the network, forces users to share the total capacity of the WLAN in a manner that ensures sufficient performance for everyone.
)
Figure 11-13 Bandwidth Control Provides Consistent Performance for All Users
Microcell Deployment Strategies
If WLAN requirements call for extremely high performance, consider using a high density of access points. This calls for turning down the transmit power of all access points and client radios, which forces access points to be much closer together (and avoid interfering with each other). This "microcell" architecture makes the physical area collision domains smaller than with conventional access point density. The microcell architecture leads to much higher performance because there are fewer client devices connecting to each access point. This allows each client device radio to consume a greater percentage of the access point's capacity and avoid collisions with other radios. Therefore, performance can be much greater.
An issue with using higher access point density to improve performance is that it leads to a great number of access points. Of course this means that the deployment will cost more, possibly considerably more. To reduce the need for increasing the density of access points, first consider all other less-expensive options covered previously in this chapter.