July 1, 2006

Essential Wi-Fi

Proper Use of RF Amplifiers An amplifier increases the strength of an RF signal, potentially allowing for greater range and higher data rates. But amplifiers are not appropriate for use in all RF installations. In some cases, an amplifier will not increase range or throughput much at all, while in other cases, the amplifier may actually make things worse. This month, we’ll teach you how to decide whether an RF amplifier should be used in your wireless network and give you some things to consider when deciding which amplifier to buy. The most important parameter of an RF amplifier is its gain. Gain is the amount of signal strength that the amplifier will add to the signal. An amplifier may or may not support automatic gain control (AGC). AGC means that the amplifier takes whatever signal strength is input and pulls it up to a constant output level. For example, imagine that the amplifier was set to output a constant 500 mW. If a 100 mW signal was input, the amplifier would add 400 mW; if a 300 mW signal was input, the amplifier would add 200 mW; and so forth. An amplifier without AGC will add a constant amount of power to the signal. For example, it might be rated at 10 dB of gain, which means that whatever signal goes into the amplifier will come out the other side 10 dB stronger (which is equivalent to 10 times stronger). Amplifiers with AGC are often preferred over fixed-gain amplifiers because it is easy to ensure that the system will have a constant output power, regardless of what output power the radio is set to. This makes it easy to perform link budget calculations and to ensure that the system does not violate FCC output power limitations, but amplifiers with AGC cannot be used with OFDM systems like 802.11g and 802.11a. Fidelity Comtech 2.4 GHz indoor RF amplifier. This is a variable-gain unit with AGC. Note the indicator lights that show which output power (100 mW 250 mW, 500 mW, 1 W) is selected.

Amplifiers can be specified as either bidirectional or unidirectional. Bidirectional amplifiers amplify the signal both when it is transmitted and when it is received. Unidirectional amplifiers typically only amplify the signal when it is transmitted (although a receive-only amplifier is certainly also possible). Transmit gain and receive gain will typically not be the same amount, and some popular amplifiers actually mix variable-gain and fixed-gain—for example, an amplifier might be variable-gain on the transmit side and fixed-gain on the receive side. When considering an amplifier, one must first decide whether to use an amplifier at all. If link budget calculations reveal that there is insufficient signal strength to achieve the desired range and data rates, an amplifier can be considered. The amplifier chosen should have enough gain to make the incoming signal strong and robust, but not too much gain. We have seen cases where excessive signal strength actually decreased throughput, to the point where the link actually dropped when the radios were at close range and the amplifiers were inserted. Our tests suggest that this was due to packet corruption caused by overdriving of the radio’s receive circuitry. The lesson here is that the signal strength at the receiving radio should be strong enough to achieve the data rates that you want to achieve, but not too much stronger than that. As a guideline, we suggest that the RSSI should be no more than about -40 dBm for best performance. Amplifiers are often not the best choice for increasing signal strength at the receiver. First, amplifiers—even bidirectional amplifiers—only operate on one side of the link, which can create the Unbalanced Power Effect (UPE). Put simply, UPE occurs when one transmitter has much more signal strength than the other, meaning that packets can get from point A to point B, but point B doesn’t have enough signal strength talk back to point A. Because 802.11 communication is always bi-directional, the link is non-viable even though one side has enough signal strength to get packets through. One way of avoiding UPE is to put an amplifier at both ends of the link such that the output power of both sides is equal. Another way is to use a higher gain antenna or antennas (which work equally on both transmit and receive) instead of amplifiers to achieve the higher signal strength that is desired. This article provides more information on the difference between higher output power and higher antenna gain. Amplifiers can be a powerful tool for increasing the range and throughput of a wireless network, but they must be deployed correctly, and only in places where they will actually improve performance. If your network would benefit from amplifiers, ensure that the amplifier that you buy has appropriate gain for your link budget, that it is specified for the frequency range that you use, and that it either has or does not have AGC, as appropriate for the 802.11 technology that you plan to use. Connect802 sells a variety of amplifiers. We will be happy to evaluate your needs and suggest appropriate equipment.

 Technology and Engineering

Optimum 802.11 Channel Allocation

Common wisdom holds that 802.11b/g channels 1, 6, and 11 are “non-overlapping” in the North American regulatory domain, and that these three channels should be preferred in 802.11 wireless network designs. This month, we’ll explore what “non-overlapping” really means, the effect of using “overlapping” channels, and times when channels 1, 6, and 11 really aren’t so “non-overlapping” after all.

802.11b and 802.11g divide the 2.4 GHz ISM frequency band into 14 channels. Each channel is specified based on a center frequency and a spectral mask. The center frequency is the frequency at which a transmitter on the channel’s energy will be the highest. The spectral mask describes the degree to which the transmitter’s signal energy must be attenuated as one measures further and further up and down from the center frequency. For example, 802.11b/g break the 2.4 GHz ISM band into 14 possible channels, whose center frequencies are 5 MHz apart. The spectral mask for an 802.11b or 802.11g channel requires that the signal energy be attenuated by at least 30 dB from its peak energy at +/- 11 MHz from the center frequency, and attenuated by at least 50 dB from its peak energy at +/- 22 MHz from the center frequency. Outside of +/- 22 MHz from the center frequency, no additional attenuation is required.

Since the spectral mask only defines power output restrictions up to +/- 22 MHz from the center frequency, it is often assumed that the energy of the channel extends no further than that. This assumption is the basis of the statement that 802.11b/g channels are 22 MHz wide. It is also the basis of the statement that channels 1, 6, and 11 do not overlap. But 802.11b/g channels do not simply stop at +/- 22 MHz from the center frequency. The spectral mask simply requires that the signal’s energy be attenuated by at least 50 dB at the +/- 22 MHz points. If the transmitter is sufficiently powerful, the signal can be quite strong even beyond the ±22 MHz point, and channels 1, 6, and 11 can overlap. It is more correct to say that, given the separation between channels 1, 6, and 11, the signal on any of these channels should be sufficiently attenuated at the +/- 22 MHz boundary to minimally interfere with a transmitter on any of the other two channels. However, this is not universally true; for example, a powerful transmitter on channel 1 may output enough signal energy that, even after the required 50 dB of attenuation, the transmitter interferes with a device on another “non-overlapping” channel.

The implications of this insight should be taken in the context of their real-world applicability. The 1, 6, 11 guideline is a good one, and in most circumstances, the separation between channels 1, 6, and 11 is sufficient that they can be considered to be non-overlapping. Nevertheless, it is important to recognize the exceptional circumstances that can cause channels 1, 6, and 11 to become effectively non-overlapping. This can occur when: two devices on different channels have higher-than-normal power output; two devices on different channels are unusually close together physically; or some combination of the two. Higher output might be defined as an output power of roughly 100 mW or greater. Unusually close together might be defined as closer than roughly five feet. As the output power of the two devices increases, and as their physical proximity increases, the likelihood of them interfering with each other increases, even if they are on technically “non-overlapping” channels.

A simple lab test can be performed to verify these results. Place two APs approximately three feet apart on a table. If the APs are capable of variable power output, set them to their highest power output. Configure one AP on channel 1 and the other AP on channel 6. Start a file transfer or other throughput test on channel 1 and measure throughput with channel 6 idle. Then start a file transfer or throughput test on both channels at the same time and compare the throughput. For additional data, repeat the test on channels 1 and 11 instead of 1 and 6, repeat the test with the APs at reduced output power, or repeat the test with the APs located further from each other.

Typical results of this test might show that a file transfer on channel 1 achieves about 19 Mbps of throughput (using 802.11g). When the throughput test on channel 6 is started, the throughput on both channels is reduced to approximately 17 Mbps. Under the described conditions, the throughput on both channels is reduced slightly when both channels are active, even though these channels are supposedly non-overlapping, and so transmissions on one channel should not affect transmissions on the other channel at all. When a throughput test is run on channels 1 and 11, both channels typically achieve the same throughput as when only a single channel is active. This is because channels 1 and 11 are further apart than channels 1 and 6.

The worst possible conditions occur when 802.11b/g devices are spaced closer than five channels apart—for example, channels 1 and 5 (four channels of separation) or channels 1 and 3 (two channels of separation). Under these conditions, there is enough overlap between the channels that the transmissions on one channel corrupt the transmissions on the other, but not enough overlap for the stations to coordinate their transmissions and avoid corruption, as would be the case if they were on the same channel. We at Connect802 occasionally see a network that has been designed around sub-five-channel separation—using channels 1, 4, 8, and 11, for example. In our experience, the packet corruption caused by the additional channel overlap more than negates any potential throughput gain that might be achieved by spacing the APs more tightly. In a case where all three “non-overlapping” channels are used, it is better to place an additional AP on the same channel as another AP than to place the additional AP on an “overlapping” channel. For example, if three APs are on channels 1, 6, and 11, and it is desired to add a fourth AP, it would be better to place that AP on either channel 1, 6, or 11, rather than place the AP on channel 3 or channel 8.

Problems with channel assignment can occur when administrators intentionally choose non-overlapping channels, but they can also occur when “auto-channel-selection” algorithms in access points make poor choices. We have seen at least one case where a certain vendor’s auto-channel-selection algorithm ended up placing ten APs on channels 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10. Performance was, as might be expected, unacceptable. In other cases, auto-channel-selection may do a good job. If your network design uses auto-channel-selection, then we recommend using a site survey tool to verify that good choices are being made. If you see a bunch of APs on adjacent, overlapping channels, you should question whether the algorithm is, in fact, making good choices. Run throughput tests on several APs simultaneously to verify that throughput is as high as you expect it to be and to verify that transmissions on one AP do not excessively reduce throughput on other, nearby APs.

Until now, this article has dealt exclusively with 802.11b and 802.11g. 802.11a does not suffer from problems with channel overlap because all 802.11a channels have sufficient separation between them to avoid interference under most conditions. Therefore, in an 802.11a network, all channels can be used simultaneously without adversely affecting performance.

Ask the Expert

To Infinity... and Beyond!

Web Searching: The Connect802 Web Presence

At Connect802 we're your PAGE ONE resource for wireless networking!

Connect802 has the experience, expertise, and resources to help you with your wireless network system. Use your favorite search engine and see what Connect802 is doing. Each month we give you some suggested search terms for you to explore. Here's this month's list. As you look down the search engine results you'll find Connect802 either at the top, or on the first page (true for Google and Excite, unknown for the rest).

• White Papers from Connect802
• Virtual Site Survey
• Turnkey Wi-Fi Network System
• Create Wi-Fi Plans with RF CAD Software
• The Best Wi-Fi System Design Consulting
• Antenna System Designer On-Line Tool
• Strix Systems Network Design Consulting
• Tropos Networks Network Design Consulting
• FireTide Network Design Consulting
• ValuePoint Network Design Consulting
• Symbol Wireless Network Design Consulting
• Professional Consulting Services for Wireless LAN Support