802.11n offers two features to increase the utilization of the radio spectrum. 802.11n retains the common 20 MHz channel width used by prior 802.11 standards. Within the 20 MHz channel, however, 802.11n improves spectral efficiency by adding subcarriers that were unused in 802.11a/g, as shown in Figure 3-1. Even though 802.11n adds four data subcarriers, increasing throughput by about 8%, it does not need to add any pilot subcarriers. Pilot subcarriers are used to provide channel measurement and calibration, and are a form of overhead. Just as MIMO increases the efficiency of a data transmission, it increases the efficiency of the pilot carrier operation. In a MIMO system, each subcarrier will be received through each of the received radio chains, and thus, will provide more information on the state of the channel as in a SISO system.

Figure 3-1. Channel structure comparison between 802.11a/g (52 carriers) and 802.11n (56 carriers)

The second change made by 802.11n is that it supports operation in wider 40 MHz channels. Although the standard describes several methods of operating a 40 MHz channel, by far the most common one is that two adjacent 20 MHz channels are treated as one channel and treated as a single 40 MHz contiguous block.

Even though the center frequency of a 40 MHz channel moves upward, the “name” of the channel does not change. For example, a 20 MHz channel operating on channel 60 can be used next to a 20 MHz channel operating on channel 64. However, an AP advertising a 40 MHz bandwidth at channel 60 occupies the spectrum for both channel number 60 and channel number 64. Figure 3-2 illustrates the differing channel widths.

Figure 3-2. Channel comparison of 20 MHz 802.11n channel to 40 MHz 802.11n channel

802.11n’s 40 MHz channels do more than double the throughput when compared to the traditional narrow channels because the 40 MHz channel format decreases overhead as a fraction of the channel. Pilot carriers are an “overhead” expense required in OFDM, and do not transmit any data from higher protocol layers. 802.11n doubled the channel width from 20 MHz to 40 MHz, but only increased the number of pilot carriers by half, as described in Table 3-1. By using the increased effectiveness of pilot carriers in a MIMO system, the spectral efficiency increases by 4%.

As with anything related to radio transmission in 802.11, regulatory considerations are an important part of how much data a network can support. Although the 802.11 specifications define a large number of channels, especially in the various 5 GHz bands, they are only available for use if allowed by the regulatory authority in the country where an AP is installed. 802.11n operates in both the 2.4 GHz band used by 802.11b/g as well as the 5 GHz band used by 802.11a.^[13] In both cases, 802.11n reuses the channel numbering that was established by previous standards. Figure 3-3 illustrates the latest available information on regulations in the 5 GHz band. Channels are identified by the IEEE channel number across the top of the diagram, and blocks indicate whether a 20 MHz or 40 MHz channel is available for use. The broad band in the middle of the figure, 5.470–5.725 GHz, has been the subject of intense interest. It is generally available throughout the world for use, and represents a substantial increase in available spectrum for building wireless LANs.

Figure 3-3. Available channel map (5 GHz)

One of the reasons why the 5.470 GHz spectrum became available is that its previous use generally allowed for a secondary user of the band. It is widely used for weather radar. Wireless LAN use of the band is secondary, meaning that if radar signals are detected the wireless LAN must be automatically shut down. However, such a high frequency generally does not impinge on operation of indoor wireless LANs. This band is sometimes referred to as the Dynamic Frequency Selection (or DFS) band after the required procedures to detect radar and avoid interfering with the allocated use of the band. Certification to operate within the DFS band is generally granted by the radio regulator in a country after testing to ensure compliance with the relevant rules.^[14] Use of 40 MHz channels substantially increases the demand on spectrum resources, and the ability to use 40 MHz channels will typically depend on whether the network components can support channels on which DFS operations are required.

In 802.11n, the Modulation and Coding Set (MCS) number is a value that describes the number of spatial streams, modulation (BPSK, QPSK, 16-QAM, or 64-QAM), and error-correcting code used for a transmission. 802.11n supports both equal modulation, in which all spatial streams are transmitted in the same manner, and unequal modulation, in which the spatial streams may be modulated differently. 802.11n defines 77 different combinations of modulation and coding. To date, most products shipped support only equal modulation modes, which are the first 32 MCS values. Table 3-3 and Table 3-4 at the end of this chapter have a list of the common MCS values, along with the link rates.

Unequal modulation is useful when one spatial stream is significantly more impaired than others. In transmit beamforming, the mathematical operations that are used to separate out the spatial streams may result in the streams having significantly different signal-to-noise ratios, which requires that some spatial streams be transmitted using a more conservative modulation. Because transmit beamforming has not been widely implemented, unequal modulation has also not seen wide implementation.

Although it is not a rule in the legal sense, a good rule of thumb used by OFDM system designers is that the guard interval (GI) should be four times the highest multipath delay spread.^[16] When 802.11a was being designed, designers used a conservative value of 200 ns for the delay spread, and chose to make the guard interval 800 ns. Implementation experience has shown that most indoor environments rarely have delay spreads of even 100 ns, and often it is closer to 50-75 ns. To wring an extra bit of performance out of the radio link, 802.11n includes an option for the short guard interval, which cuts the guard interval down to 400 ns. Whether you can successfully use the short guard interval depends on the multipath spread, which is reported by advanced RF analysis tools. In general, multipath interference is worse when there are significant reflections due to metal. Most 802.11n networks are able to use the short guard interval without issue.

All other OFDM parameters remain the same. Overall speed is increased because the part of the symbol time devoted to data transmission is still 3.2 µs, but each symbol is shorter. The total symbol length shrinks from 4.0 µs with the long guard interval to 3.6 µs (3.2 µs + 0.4 µs), shrinking OFDM overhead by 10% (as illustrated in Figure 3-4).

Figure 3-4. Long and short guard interval compared

The fields in the HT Mixed Mode PLCP frame are:

Greenfield mode removes backward compatibility support by replacing the legacy fields with HT-specific versions of the same fields.

HT Signal Field

The HT Signal field, shown in Figure 3-6, serves the same purpose as the Signal field does in 802.11a/g. It is transmitted at a standard rate (MCS 0) and can be readily decoded by a receiver. In an HT-Mixed Mode format transmission, the presence of the HT Signal field is used to determine whether the payload is transmitted using the older 802.11a/g rates or the MIMO rates in 802.11. To assist receivers in determining whether the HT Signal field or an 802.11a/g Data field follows the training symbols, the constellation for the HT Signal field is rotated. 802.11a/g receivers are unable to decode the payload, but 802.11n receivers decode and use the data in the HT signal field aids in decoding the PSDU portion of the packet by describing how it is encoded and modulated.

Figure 3-6. HT Signal field

Modulation and Coding Scheme (7 bits)

This field describes the modulation and coding scheme, and implicitly gives the receiver the number of spatial streams. Although the MCS value can take on any one of 76 values, the 32 described in 802.11n Speed are by far the most commonly used.

Channel bandwidth (1 bit)

This bit is set to 0 for 20 MHz channels and 1 for 40 MHz channels.

HT Length (16 bits)

This field describes the length of the PSDU in bytes, and can range up to 65,535.

Not Sounding (1 bit)

Sounding frames are used in beamforming, a process that is described in more detail in Chapter 4. When this bit is set to 0, the frame is not a sounding frame, which is the typical operation. When set to 1, the frame is used for sounding operations.

Aggregation (1 bit)

This bit is set when the payload is an aggregate frame that contains several subframes. Aggregation is further described in Chapter 5.

STBC (2 bits)

Space-Time Block Coding allows multiple radio chains to be used to increase the sensitivity to a single data stream; it is further described in Chapter 4.

Forward Error Correction coding (1 bit)

When using a convolutional code, this bit is set to 0; to use LDPC, the bit is set to 1.

Short guard interval (1 bit)

When the short guard interval is in use, this bit is set to 1. When set to 0, it indicates the long guard interval is in use.

CRC (8 bits)

A CRC allows the receiver to detect corruption of the HT Signal field.

Tail bits (6 bits)

The HT Signal field is protected by a convolutional code, and requires a trailing six zeroes to “ramp down” the convolutional code.

Many client devices have implemented 802.11n for cost reasons. Once the technology basis for the industry moves, it is hard to get older chips. Even though many battery-powered devices have low overall throughput requirements and therefore, manufacturers would sometimes prefer to keep using 802.11b, chip vendors continuously refresh their product lines and pull their customers towards the latest technology. Even devices made with full attention to reducing battery draw are now made with 802.11n chips, though of course it is a 1x1 design to minimize the number of power-hungry radio chain components. Virtually every Wi-Fi enabled phone uses single-stream 802.11n, as do many tablet computing devices.

In the case of single-stream 802.11n, there is only a small performance gain in link rate. 802.11n’s channel has 8% more data subcarriers in a channel, so it can support link rates that are 8% higher, as shown in Table 3-2. Even though the link rate is not higher on paper, it often makes sense to use 802.11n for these devices because the antenna array at the AP can often boost performance through techniques like maximal ratio combining.

The second comparison that is often made is the difference in link rate between the familiar 20 MHz channel width and the new wider 40 MHz channels in 802.11n. Table 3-3 and Table 3-4 show the link rate that can be achieved for various combinations of modulation, coding, and spatial streams. The table shows data rates using the long guard interval; for speeds using the optional short guard interval, add 11%. In 802.11n, each combination is given an MCS number. These two tables only show the equal modulation MCS numbers; there are many more MCS numbers that make use of unequal modulation, but they are not widely supported.

The first wave of access points and laptops to hit the market supported only two spatial streams, and the common numbers used to describe the performance of these systems was “100 Mbps” (which is the net throughput of a set of frames aggregated together at 135 Mbps, plus a block acknowledgment) or “300 Mbps” (the link rate achieved with 40 MHz channels and a short guard interval).

In 2011, the second major wave of products was brought to market, supporting three spatial streams and speeds of up to 450 Mbps. These products now power all mainstream high-performance access points, and many new client devices now use three-stream 802.11n chipsets. Although four spatial streams have been standardized, it seems unlikely that four-stream products will ever be mainstream because most chip vendors have focused on building four-stream products using 802.11n’s even higher-speed successor.