OFDM PMD

The OFDM PHY uses a cocktail of different modulation schemes to achieve data rates ranging from 6 Mbps to 54 Mbps. In all cases, the physical layer uses a symbol rate of 250,000 symbols per second across 48 subchannels; the number of data bits per symbol varies. An OFDM symbol spans all 48 subchannels.

Encoding and Modulation

There are four rate tiers with the OFDM PHY: 6 and 9 Mbps, 12 and 18 Mbps, 24 and 36 Mbps, and 48 and 54 Mbps. Support is required for 6, 12, and 24 Mbps, which are lowest speeds in each of the first three tiers, and therefore the most robust in the presence of interference. The lowest tier uses binary phase shift keying (BPSK) to encode 1 bit per subchannel, or 48 bits per symbol. The convolutional coding means that either half or one quarter of the bits are redundant bits used for error correction, so there are only 24 or 36 data bits per symbol. The next tier uses quadrature phase shift keying (QPSK) to encode 2 bits per subchannel, for a total of 96 bits per symbol. After subtracting overhead from the convolutional code, the receiver is left with 48 or 72 data bits. The third and fourth tiers use generalized forms of BPSK and QPSK known as quadrature amplitude modulation (QAM). 16-QAM encodes 4 bits using 16 symbols, and 64-QAM encodes 6 bits using 64 symbols. The third tier uses 16-QAM along with the standard R=1/2 and R=3/4 convolutional codes. To achieve higher rates with 64-QAM, however, the convolutional codes use R=2/3 and R=3/4. Table 13-3 summarizes the coding methods used by each data rate in the OFDM PHY.

Radio Performance: Sensitivity and Channel Rejection

Like other physical layers, 802.11a specifies minimum performance requirements for the receiver, which are shown in Table 13-4. Minimum sensitivity has been discussed for the other physical layers. The only feature of note in 802.11a is that the wide range in data speeds means that a minimum performance requirement is quoted for each data speed. In comparison with the requirements laid down by the direct-sequence layers, 802.11a is just as stringent. 802.11a requires a -76 dBm sensitivity, which is comparable to the 18 Mbps and 24 Mbps data rates in 802.11a.[*]

[*] Note, however, that path loss is worse at the 802.11a frequencies.

More interesting is the specification of channel rejection. As with the other physical layers, begin by injecting a signal slightly (3 dB) above the minimum sensitivity on a given channel. On either rejection test, bring up a second signal on either an adjacent or nonadjacent channel. When the channel under test suffers a 10% frame error rate, note the difference in power between the two channels.

What the table says is that as the data rate increases, the more easily a signal is disrupted at the receiver. If an 802.11a network is built too dense, so that 54 Mbps signals from adjacent APs are regularly received in the middle of the two, it is possible that the client radio chipsets will not be able to decode the transmissions. However, this scenario is unlikely due to the relatively short range of 54 Mbps transmissions. Furthermore, many chipsets will perform better than the standard requires, but most vendors do not quote rejection in the data sheets for their cards.

Clear Channel Assessment

The OFDM PHY specification leaves implementers a great deal of latitude in selecting techniques for noting a busy channel. Received signal strength thresholds determine whether the channel is in use, but the main guideline for 802.11a equipment is that it must meet certain performance standards. Implementations are free to use the Packet Length field from the PLCP header to augment clear channel assessment, but this is not required.

Transmission and Reception

The block diagram for an 802.11a receiver is shown in Figure 13-17. When a frame is ready for transmission, the 802.11a interface runs the following procedure:

1. Select a transmission rate. Algorithms for selecting a data rate are implementation-dependent, so products sold by different vendors may select different rates under identical circumstances. The rate dictates the modulation and convolutional code, as well as the number of data bits per subcarrier. See Table 13-3.
2. Transmit the PLCP preamble, which consists of long and short training sequences.
3. Begin transmission of the PLCP header with the SIGNAL field, which is not scrambled. It is coded with the convolutional coder.
4. Create the data field of the packet.
   1. Create the SERVICE field, which is currently set to all zeros. It has seven zeros for scrambler initialization, and nine reserved bits set to zero.
   2. Append the data.
   3. Put on the six zero tail bits.
   4. Pad with zeros to a multiple of the data bit per subcarrier block size.

5. Scramble the data, which helps avoid long strings of zeros or ones.
6. Encode the data with the convolutional coder. If necessary, puncture the output of the convolutional code to generate an encoded string at a higher rate than 1/2.
7. Divide the coded data into blocks for processing. The block size depends on the modulation rate for the data symbols.
   1. Perform the interleaving process and map bits from the block on to the 48 subcarriers.
   2. Insert the four pilot subcarriers at the designated locations.in the channel.
   3. Use the inverse Fourier transform to convert the frequency-domain data into time-domain data for transmission.

8. Repeat step 5 for each data block until there are no more.

Figure 13-17. A transceiver block diagram

Acknowledgment

Support of the 6, 12, and 24 Mbps data rates is required by 802.11a. Upon receipt of a frame, the 802.11 MAC requires an acknowledgment. Acknowledgments must be sent at a supported data rate for all associated stations. Most devices send acknowledgments at 24 Mbps because it minimizes the overhead while obeying the stricture to transmit at a mandatory rate.

An example of OFDM encoding

OFDM encoding, as you can no doubt see by now, is an intense, multistep process. One of the additions that 802.11a made to the original specification was Annex G, an encoding of Schiller's Ode to Joy for transmission over an 802.11a network. Shortly after 802.11a was published, the IEEE 802.11 working group discovered several errors in the example and published a correction. If you are interested in learning about OFDM encoding in detail, you can refer to this example.