Scalable voip mobility intedration and deployment- P17 pdf

160 Chapter 5 www.newnespress.com 0°180° 90° -90° 16-QAM quadrature axis (Q) x x x xx x x x xx x x x x x x 0°180° 90° -90° 64-QAM quadrature axis (Q) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x xx xx x x x x x x x x x x x x x x x x x x x x x xxx x x x x in-phase axis (I) in-phase axis (I) Figure 5.21: 16-QAM and 64-QAM Introduction to Wi-Fi 161 www.newnespress.com source bits and record whether that sum is even or odd as one additional bit. If one bit is lost, the sum of the surviving bits can be compared in evenness to the sum bit, therefore recovering the original data. This is called a parity check code, and conceptually introduces the concept of using arithmetic on some or all of the source bits to produce the extra bits. 802.11ag uses a convolutional encoder to expand the source bits, in this case producing twice the number of bits but, unlike the doubler, doing so more intelligently to avoid some loss patterns of equal bit losses being worse than others. This expansion still produces only one data rate, at a 1 2 -coding rate per modulation. To produce the other coding rates, 802.11ag just uses the property that the error-correcting code can tolerate loss, and goes ahead and starts tossing bits to get to a higher coding rate. This process, called puncturing, sounds inefficient by reducing redundancy that was just added (why not just not add as many extra bits in the first place?), but ends up saving on the complexity of the coding hardware, the radio. The overall picture, then, for 802.11ag encoding is for the data bits to be scrambled (as with 802.11b), then expanded with the error-correcting code, then split among the subcarriers, and then modulated. Because of the larger number of things going on in the signal, there is more risk of losing data if the receiver’s timing goes off a bit from the sender’s. To compensate, timing is maintained by the addition of four pilot subcarriers. These carry a known signal pattern— like the preamble—but do so for the length of the transmission. If the receiver’s clock speeds up or slows down relative to the sender’s, then the constellation would essentially rotate, and as the constellation points are now closer together than with 802.11b, the rotation would cause the receiver’s bits to jumble. The pilot subcarriers’ known pattern lets the receiver adjust as needed. 5.5.2.1 Preambles, Slots, and Optimizatoins The preambles of 802.11ag are at 6Mbps, the lowest 802.11ag data rate. They are also significantly shorter than the short 802.11b preambles: 40 microseconds for 802.11ag. Improvements in signal processing technology since 802.11b came along allowed the designers to not need to provide as much synchronization time. Furthermore, the radios can stop receiving and start transmitting more quickly for 802.11ag than they were expected to be able to do for 802.11b. This allows 802.11ag to use shorter slots than 802.11b, allowing for less wasted time. For 802.11g, short slots are an option, and are determined based on the presence of 802.11b clients, to prevent 802.11g clients from using a different contention scheme. 802.11a always assumes the faster slots. 5.5.2.2 802.11b Protection 802.11a has no legacy clients to deal with. However, 802.11g is in the 2.4GHz band, and has to avoid destroying 802.11b performance when the two devices are present together. 162 Chapter 5 www.newnespress.com This destruction would occur because 802.11b radios use carrier detection (see Section 5.4.3) to determine whether the channel is clear before transmitting, and that means that the 802.11b devices are looking for 802.11b transmissions. 802.11g transmissions, however, look nothing like 802.11b, and so 802.11b radios would end up only seeing 802.11g as some sort of foreign interference. To prevent this from disrupting any 802.11b device’s traffic, 802.11g introduces the notion of CTS-to-self protection. Because 802.11b clients can only see 802.11b traffic, a way to stop them from transmitting when an 802.11g transmission will start is for the 802.11g device to send an 802.11b (legacy) CTS message first. This CTS message, sent not as a part of an RTS/CTS transaction (see Section 5.4.6) but from the 802.11g sender to, nominally, itself, sets the virtual carrier sense for all devices that can hear it. The CTS frame has a Duration field—or the length of time to quiet the other stations—long enough to let it finish the 802.11g transmission that will follow. CTS-to-self protection is automatically turned on for any AP that has 802.11b legacy clients associated to it, or for access points who overhear neighboring access points that have 802.11b clients assigned to them. This CTS-to-self message can be incredibly inefficient, and has the potential to disrupt voice mobility networks, as will be mentioned later in this chapter. 802.11g data rates, even though they are identical to 802.11a data rates, may go by the additional term ERP in product literature. The term, short for Extended Rate PHY, is typical of the language used in the 802.11 standard (worse abuses will come up in the next section), but is good to know for when it occasionally slips into product documents for users. When you see ERP, think 802.11g, and when you see non-ERP, think 802.11b-only legacy devices. Neither term is correct for 802.11a devices, which are just known as 802.11a. 5.5.3 802.11n 54Mbps seemed like a lot at the time, but enterprise wireline networks operate at 100Mbps or more. To allow for even higher data rates, IEEE has embarked upon the 802.11n standard. 802.11n revolutionizes Wi-Fi by adding another radio breakthrough, as well as a long list of additional enhancements and optimizations. The most important addition 802.11n brings is the use of a technology called multiple-in, multiple-out (MIMO). MIMO does something that seems counterintuitive—almost magical—to those used to thinking about how two radios transmitting at once cause collisions and destroy wireless networks. MIMO transmits multiple signals at once, on the same channel, at the same time, and at the same power levels. However, MIMO is not magic, just math, and is able to greatly increase the speed of the network. Introduction to Wi-Fi 163 www.newnespress.com MIMO works by requiring each device to have multiple antennas. These antennas are not terribly far apart—a few inches at the most—but they need to be present. MIMO then splits the data across those antennas, sending out multiple spatial streams. These streams go over the air at the same time, and the receiver uses multiple antennas to pick up this transmission, applies some math to separate back out the streams, and then recombines the data. We will go through how this multiple simultaneous transmissions work in a moment. However, because MIMO has some general rules that the products using them need to follow, let us start with those rules, and the state of the technology. 802.11n is a very new standard, and, for 2008 and most of 2009, was not yet finished. However, every major Wi-Fi vendor was selling 802.11n-based products. How can this happen? As it turns out, major parts of 802.11n were complete enough for vendors to build products to from 802.11n Draft 2.0. The features that were complete enough, and were also interesting enough to encourage users to purchase products based on them, were written down by the Wi-Fi Alliance into an industry certification program, also known as 802.11n Draft 2.0. This program specifies a rigorous set of interoperability tests, to ensure that vendors that pass it have built their devices to the same specification (as in, they didn’t make major errors). The existence of the certification program should bring you comfort in knowing that 802.11n devices will work together. For 802.11n Draft 2.0 products, the Wi-Fi Alliance, which uses its role to ensure that devices interoperate, happened to do their work a bit earlier than IEEE. As it has turned out, however, the Draft 2.0 feature set and program are essentially the same as those for the final standard. This makes sense, because the vendors that figure prominently in IEEE and the Wi-Fi Alliance had a tremendous incentive to ensure that the final standard was only minimally changed from the draft. What features do 802.11n devices commonly provide? Table 5.13 lists them. The main feature is the ability to provide 300Mbps data rates for clients. This is achieved by using two spatial streams and double-wide 40MHz channels. (The standard defines up to four spatial streams, and some—but not most—devices can accommodate three streams, yet the overwhelming majority to date can use only two.) Furthermore, the WMM and aggregation optimizations go a long way towards closing the gap between the data rate and the actual highest application throughput. Ignoring the MIMO aspect, 802.11n is similar to 802.11ag. It too is OFDM, with exactly the same subcarrier setup as 802.11ag, except for increasing the number of data subcarriers by four to 52 (and thus slightly increasing the frequency width). Additionally, the symbol is usually still 4 microseconds; however, there is an optional mode known as short guard interval that shaves 400 nanoseconds off the symbol’s length, getting a slightly higher kick in throughput. There are still eight modulation and coding rates that establish the data rates, 164 Chapter 5 www.newnespress.com except that one BPSK mode is removed and another 64-QAM mode, with an even higher coding rate, is added. These eight rates, however, are multiplied into a much higher number of rates, based on the channel width, the number of spatial streams, and the guard interval. All together, the notion of data rates being signaled in the products by Mbps has been abandoned, and replacing it is the concept of the modulation and coding scheme (MCS), based on a small number representing what the actual parameters are in a table. This is similar to how simple channel numbers represent much more complicated frequencies. Table 5.14 contains the common two-stream set for 802.11n devices and encompasses 60 different data rate options, each with its own slightly different SNR requirement, channel width, or robustness. There are additional features that 802.11n has as options, which are not commonly implemented but could be of great benefit for voice mobility. Among these features are space-time block codes (STBC), transmit beamforming, and extended power save capabilities. However, because these features have not yet become commonplace, their use is rather limited. 5.5.3.1 MIMO As mentioned before, MIMO lets the devices transmit at, three times, or four times by using as many spatial streams. The general rule for MIMO, as a theory, is that the number of spatial streams usable is the lesser of the number of antennas that can be used simultaneously on the transmitter and the receiver. In theory, a 100-antenna transmitter and a 50-antenna receiver could allow for a 50 (< 100) spatial stream radio. In practice, there are limits. 802.11n defines only four spatial streams maximum. The Wi-Fi Alliance Draft 2.0 certification tested for only two spatial streams, and most devices today remain only capable of two streams. The reasons are rather simple. More antennas require more room for antennas, and it is hard to find room to place them. Also, more antennas Table 5.13: 802.11n Common features 802.11n Features Meaning Two spatial streams Doubles performance over non-MIMO Aggregation Allows for very high efficiency WMM Quality-of-service is mandatory (see Section 6.0.1) WPA2 High-grade encryption support is mandatory (see Section 5.6.1.1) 40MHz wide channels Doubles performance over 20MHz 300Mbps Top data rate MRC/Receive Beamforming Longer range in some cases Introduction to Wi-Fi 165 www.newnespress.com require more radio chains—an 802.11n radio is actually made of multiple copies of the parts needed to make an 802.11ag radio work—and those are expensive and draw power. Finally, practical considerations prevent higher numbers of simultaneous streams from working well. Now, for the description of why MIMO works. For this example, assume that the sender and receiver each have three antennas, but follow the industry norm of using only two spatial streams. The sender divides its signal into two spatial streams, then spreads its two streams across the three antennas. Those two streams from three signals bounce around the environment, and end up as three different signals at the receiver, one for each of the receiver’s antennas. Each of those three signals is some different combination of the three signals from the sender, and thus is some different combination of the two spatial streams. This is where the math sets in. The different combinations are usually very different. If you Table 5.14: The 802.11n data rates MCS Modulation Bits per Symbol per Subcarrier Coding Rate Spatial Streams Speed 20MHz Long GI 20MHz Short GI 40MHz Long GI 40MHz Short GI 0 BPSK 1 1 2 1 6.5 7.2 13.5 15 1 QPSK 2 1 2 13 14.4 27 30 2 QPSK 2 3 4 19.5 21.7 40.5 45 3 16-QAM 4 1 2 26 28.9 54 60 4 16-QAM 4 3 4 39 43.3 81 90 5 64-QAM 6 2 3 52 57.8 108 120 6 64-QAM 6 3 4 58.5 65 121.6 135 7 64-QAM 6 5 6 65 72.2 135 150 8 BPSK 1 1 2 2 13 14.4 27 30 9 QPSK 2 1 2 26 28.9 54 60 10 QPSK 2 3 4 39 43.3 81 90 11 16-QAM 4 1 2 52 57.8 108 120 12 16-QAM 4 3 4 78 86.7 162 180 13 64-QAM 6 2 3 104 115.6 216 240 14 64-QAM 6 3 4 117 130 243 270 15 64-QAM 6 5 6 130 144.4 270 300 166 Chapter 5 www.newnespress.com write out the two spatial streams, the effects of the sender’s spreading and the channel’s bouncing, and the receiver’s receiving, you can produce a matrix equation, from linear algebra. Because the combinations are different for each antenna—linearly independent, in fact—the receiver can undo the effects of the channel using linear algebra and produce the original streams. For further explanation, see the appendix at the end of this chapter. Basically, the MIMO receiver uses the preamble of the frame that is sent to discover what the effects of the channel are on the streams, and then uses that to undo those effects. The effect of having multiple antennas when only one spatial stream is used (and a main effect as a part of MIMO) is for beamforming. There are two parts to beamforming: receive beamforming, and transmit beamforming. The term beamforming arose from RADAR, where stationary equipment used electronics and a large number of antennas to set up interference patterns just right to concentrate a signal in a direction or to a point, as if the antennas were mounted on a swivel and were pointed, although they are not. Receive beamforming isn’t necessarily thought of the same way as transmit beamforming, but is the major reason why 802.11n has higher range than 802.11abg. There are a number of techniques for doing what could be called receive beamforming. One term used surprisingly often by vendors in describing their products in data sheets is maximum ratio combining (MRC). To understand it at a high level, the receiver is twiddling with how it combines the signals received on each of its antennas to maximize the power of the final signal it received. Because of the way interference patterns and combinations work, it turns out that the twiddling it does is a unique pattern (H, the channel matrix, if you read the math briefing) based on the client’s location. But because receive beamforming learns that pattern when it sees the preamble (same as MIMO), this is not a problem, and the end result is an apparent amplification of the signal, thus increasing range for reception. 802.11n clients with MIMO have longer range on legacy access points than legacy clients do, for that reason. Therefore, if you need to extend the range of a couple of clients, your best bet is to upgrade the clients to 802.11n. (Upgrading the access points without upgrading the clients may not increase range at all in many cases.) Transmit beamforming is also possible. 802.11n defines two types of beamforming, known as explicit and implicit beamforming. Explicit beamforming uses the cooperation of the receiver to determine the best way of combining signals across the transmitter’s antennas for forming the signal to the receiver. This cooperation requires features on the receivers that are not commonly implemented. Implicit beamforming simply requires the transmitter, assuming that the channel it sees from the receiver when that device transmits is the same as what the receiver sees from the transmitter. By assuming this reciprocity, the transmitter does not need to involve the receiver. However, it is forced into its guess, which may not be correct, and requires that the transmitter always keep tabs on the receiver’s channel conditions by either sending an RTS to it before every packet, thus eliciting a responding Introduction to Wi-Fi 167 www.newnespress.com CTS that will help uncover the channel conditions, or by winging it and hoping the receiver doesn’t move much. For this reason, some vendors are limiting their transmit beamforming support to only legacy, non-802.11n clients. Transmit beamforming is an interesting concept for voice mobility networks based on the microcell approach of reducing transmit power levels to begin with. 5.5.3.2 Legacy Support Legacy, in the context of 802.11n, means 802.11ag as well as 802.11b. To avoid the same problem of 802.11g possibly being interfered with by 802.11b here, with 802.11ag stepping on 802.11n, 802.11n uses a different form of protection. Instead of the wasteful CTS frames, and the protocol necessary to decide whether a CTS frame is needed, 802.11n uses a special preamble. The preamble first starts off as an 802.11ag preamble, not only at 6Mbps but signaling a 6Mbps data rate for the following data, and including a length that will encompass the entire 802.11n frame. However, as soon as the preamble is over, the 802.11n radio stops transmitting in 802.11ag, and switches over to 802.11n, where it continues with more preamble fields, including the real data rate. 802.11n devices see the entire frame. 802.11ag devices see the preamble only, but defer just as if the frame were all 802.11ag but out of range. Thus, the technologies do not interfere. For 802.11b clients, 802.11n still uses an 802.11b CTS frame. 5.5.3.3 Aggregation 802.11 frames have to have the preamble, but there was no particular reason that one preamble couldn’t cover multiple frames. 802.11n introduces this concept with frame aggregation, or A-MPDUs. A-MPDUs are a special type of jumbo frame that contains multiple 802.11 data frames sent from and to the same wireless device. Up to four milliseconds or roughly 64,000 bytes can be packed into one of these aggregates. In almost every sense, an A-MPDU can just be thought of as a concatenation of data: every byte is sent at the same data rate that the preamble calls out, and the A-MPDU is retransmitted in full if the expected acknowledgment does not come back immediately from the receiver. However, unlike a simply larger data frame, if some of the data frames within the aggregate are received and others are not, the receiver can indicate this by using a special Block Acknowledgment. This block acknowledgment specifies precisely which of the senders’ frames were received and which had errors in them. For those frames that were not received, the sender can choose to add those to later aggregates, thus not wasting time resending frames that already arrived. The main benefit of aggregation is to reduce the overhead of the preamble and backoff for 802.11n. 168 Chapter 5 www.newnespress.com 5.5.3.4 Double-Wide Channels (40MHz) Double-wide channels work by bonding together two adjacent 20MHz channels into one larger 40MHz channel. This 40MHz channel acts just like a wider 20MHz channel, but offering data rates slightly higher than twice that of the 20MHz channel. (The slight increase over twice comes from using up the gap between channels as usable bandwidth.) There is some inconsistency in the naming of this feature in the industry. Some devices call it “double-wide,” others call it “channel bonding,” and others call it just “40MHz.” We’ll use “double-wide” and “40MHz” for this description. Double-wide channels are named by the 20MHz channels that they occupy. There are a few nomenclatures in active use, but all are just slight variations of the basic concept. There is a primary channel and an extended channel in a 40MHz channel name. 40MHz channels can operate with 20MHz devices, but only on one 20MHz half of the double-wide channel. This one half is the primary channel. The other half is used only for the rest of the 40MHz transmissions, and is thus the extended channel. For example, a 40MHz channel selection that has a primary channel of 36 and an extended channel of 40 can be written as 36+1, (36, +1), 36U, 36H, and so on. The same 40MHz channel, but with the primary being on the other half, can be written as 40−1, (40, −1), or 40L. It is useful to keep in mind, however, that a 40MHz frame only sees one 40MHz channel, and is not split itself into two separate 20MHz frames for each half, unlike what the terminology suggests. There are a few considerations for double-wide channels. The first is that both 20MHz channels that are within the 40MHz should be empty. This is especially true for the extension channel, which should not have any other access points using that channel, unless a layered or virtualized over-the-air architecture is used that can support that. Certainly, no access point should be deployed such that its primary (or only) channel is 40 if 36+1 is being used anywhere around that access point. The second consideration is that some devices cannot support 40MHz, and so will use only the primary half of the channel. This is obviously true for legacy (non-802.11n) devices, but is also true for some 802.11n devices. Some 802.11n devices are not designed to take advantage of double-wide channels, and a few early 802.11n devices had support for double-wide channels in the 2.4GHz band turned off, though newer devices from those manufacturers have solved that problem. The third consideration is that there are only half as many 40MHz channels as there are 20MHz channels, and the direction a 40MHz channel may extend is limited. This is to avoid the overlap mentioned earlier, but ends up leading to channel waste unless channels Introduction to Wi-Fi 169 www.newnespress.com are carefully planned. For example, channels 36 and 40 can be bonded, but channels 40 and 44 can never be, by rule. If channel 36 is being used by an existing 20MHz network, then channel 40 can never be used with a 40MHz channel in that case. The 2.4GHz band does not have the limitation of which direction a 40MHz channel can extend, but because the bandwidth is so limited, only one 40MHz channel can be created successfully. The network or administrator can usually choose which. Keep in mind that dynamic architecture vendors recommend against using double-wide channels in the 2.4GHz band, because using it on that architecture eliminates a channel that is necessary for the alternating channel plan to work with. Layered and virtualized architectures do not have this limitation. 5.5.3.5 Coming Down the Road 802.11 is still growing. As of this writing, there is a push to expand the technology to increasing bands—such as the spectrum opened up by the end of analog television broadcasts in 2009. These technologies will not show up for a few years, however, and should not affect voice mobility networks that are being considered for deployment in the near future. 5.6 Security for 802.11 Security is a broad subject, and there is an entire chapter dedicated to the unique challenges with security for voice mobility later. But any component of voice mobility over Wi-Fi will require some use of 802.11’s built-in encryption. Keep in mind that securing the wireless link is not only critical, but may be the only encryption used to prevent eavesdroppers from listening in on sensitive voice calls for many networks. 802.11 security has both a rich and somewhat checkered past. Because of the initial application of 802.11 to the home, and some critical mistakes by some of the original designers, 802.11 started out with inadequate protection for traffic. But thankfully, all Wi-Fi- certified devices today are required to support strong security mechanisms. Nevertheless, administrators today do still need to keep in mind some of the older, less secure technologies—often because the mobile handset might not correctly support the latest security, and it may fall to you to figure out how to make an old handset work without compromising the security of the rest of the network. A secure wireless network provides at least the following (borrowed from Chapter 8): • Confidentiality: No wireless device other than the intended recipient can decrypt the message. . using them need to follow, let us start with those rules, and the state of the technology. 802.11n is a very new standard, and, for 2008 and most of 2009, was not yet finished. However, every major. streams, and the guard interval. All together, the notion of data rates being signaled in the products by Mbps has been abandoned, and replacing it is the concept of the modulation and coding. the Draft 2.0 feature set and program are essentially the same as those for the final standard. This makes sense, because the vendors that figure prominently in IEEE and the Wi-Fi Alliance had