802.11 PHY Layer

The initial 802.11 standard, as ratified in 1997, supported three alternative PHY layers; frequency hopping and direct sequence spread spectrum in the 2.4 GHz band as well as an infrared PHY. All three PHYs delivered data rates of 1 and 2 Mbps.

The infrared PHY specified a wavelength in the 800–900 nm range and used a diffuse mode of propagation rather than direct alignment of infrared transceivers, as is the case in IrDA for example. A connection between stations would be made via passive ceiling reflection of the infrared beam, giving a range of 10–20 metres, depending on the height of the ceiling.

Pulse position modulation was specified, 16-PPM and 4-PPM respectively for the 1 and 2 Mbps data rates. Later extensions to the standard have focused on high rate DSSS (802.11b), OFDM (802.11a and g) and OFDM plus MIMO (802.11n). These PHY layers will be described in the following sections.

802.11a PHY Layer

The 802.11a amendment to the original 802.11 standard was ratified in 1999 and the first 802.11a compliant chipsets were introduced by Atheros in 2001. The 802.11a standard specifies a PHY layer based on orthogonal frequency division multiplexing (OFDM) in the 5 GHz frequency range.

In the US, 802.11a OFDM uses the three unlicensed national information infrastructure bands (U-NII), with each band accommodating four non-overlapping channels, each of 20-MHz bandwidth.

Maximum transmit power levels are specified by the FCC for each of these bands and, in view of the higher permitted power level, the four upper band channels are reserved for outdoor applications. In Europe, in addition to the 8 channels between 5.150 and 5.350 GHz, 11 channels are available between 5.470 and 5.725 GHz (channels 100, 104, 108, 112, 116, 120, 124, 128, 132, 136, 140).

European regulations on maximum power level and indoor versus outdoor use vary from country to country, but typically the 5.15–5.35 GHz band is reserved for indoor use with a maximum EIRP of 200 mW, while the 5.47–5.725 GHz band has an EIRP limit of 1W and is reserved for outdoor use.

As part of the global spectrum harmonisation drive following the 2003 ITU World Radio Communication Conference, the 5.470–5.725 GHz spectrum has also been available in the US since November 2003.

Each of the 20 MHz wide channels accommodates 52 OFDM subcarriers, with a separation of 312.5 kHz (= 20 MHz/64) between centre frequencies. Four of the subcarriers are used as pilot tones, providing a reference to compensate for phase and frequency shifts, while the remaining 48 are used to carry data.

Four different modulation methods are specified which result in a range of PHY layer data rates from 6 Mbps up to 54 Mbps. The coding rate indicates the error-correction overhead that is added to the input data stream and is equal to m/(m+n) where n is the number of error correction bits applied to a data block of length m bits.

For example, with a coding rate of 3/4 every 8 transmitted bits includes 6 bits of user data and 2 error correction bits. The user data rate resulting from a given combination of modulation method and coding rate can be determined as follows, taking the 64-QAM, 3/4 coding rate line as an example.

During one symbol period of 4 μS, which includes a guard interval of 800 nS between symbols, each carrier is encoded with a phase and amplitude represented by one point on the 64-QAM constellation. Since there are 64 such points, this encodes 6 code bits.

The 48 subcarriers together therefore carry 6 × 48 = 288 code bits for each symbol period. With a 3/4 coding rate, 216 of those code bits will be user data while the remaining 72 will be error correction bits. Transmitting 216 data bits every 4 μS corresponds to a data rate of 216 data bits per OFDM symbol × 250 OFDM symbols per second = 54 Mbps.

The 802.11a specifies 6, 12 and 24 Mbps data rates as mandatory, corresponding to 1/2 coding rate for BPSK, QPSK and 16-QAM modulation methods. The 802.11a MAC protocol allows stations to negotiate modulation parameters in order to achieve the maximum robust data rate.

Transmitting at 5 GHz gives 802.11a the advantage of less interference compared to 802.11b, operating in the more crowded 2.4 GHz ISM band, but the higher carrier frequency is not without disadvantages. It restricts 802.11a to near line-of-sight applications and, taken together with the lower penetration at 5 GHz, means that indoors more WLAN access points are likely to be required to cover a given operating area.

802.11b PHY Layer

The original 802.11 DSSS PHY used the 11-chip Barker spreading code together with DBPSK and DQPSK modulation methods to deliver PHY layer data rates of 1 and 2 Mbps respectively. The high rate DSSS PHY specified in 802.11b added complementary code keying (CCK), using 8-chip spreading codes.

The 802.11 standard supports dynamic rate shifting (DRS) or adaptive rate selection (ARS), allowing the data rate to be dynamically adjusted to compensate for interference or varying path losses.

When interference is present, or if a station moves beyond the optimal range for reliable operation at the maximum data rate, access points will progressively fall back to lower rates until reliable communication is restored.

This strategy is based on the implications of Eq. (4-1), which showed that SNR is proportional to the transmitted energy per bit, so that by falling back to a lower data rate, a higher SNR and lower BER can be achieved.

Conversely, if a station moves back within range for a higher rate, or if interference is reduced, the link will shift to a higher rate. Rate shifting is implemented in the PHY layer and is transparent to the upper layers of the protocol stack. The 802.11 standard specifies the division of the 2.4 GHz ISM band into a number of overlapping 22 MHz channels.

The FCC in the US and the ETSI in Europe have both authorised the use of spectrum from 2.400 to 2.4835 GHz, with 11 channels approved in the US and 13 in (most of) Europe. In Japan, channel 14 at 2.484 GHz is also authorised by the ARIB. Some countries in Europe have more restrictive channel allocations, notably France where only four channels (10 through 13) are approved.

The available channels for 802.11b operation are summarised in Table 6-8. The 802.11b standard also includes a second, optional modulation and coding method, packet binary convolutional coding (PBCCTM–Texas Instruments), which offers improved performance at 5.5 and 11 Mbps by achieving an additional 3 dB processing gain.

Rather than the 2 or 4 phase states or phase shifts used by BPSK/DQSK, PBCC uses 8-PSK (8 phase states) giving a higher chip per symbol rate. This can be translated into either a higher data rate for a given chipping code length, or a higher processing gain for a given data rate, by using a longer chipping code.

802.11g PHY Layer

The 802.11g PHY layer was the third 802.11 standard to be approved by the IEEE standards board and was ratified in June 2003. Like 802.11b, 11g operates in the 2.4 GHz band, but increases the PHY layer data rate to 54 Mbps, as for 802.11a.

The 802.11g uses OFDM to add data rates from 12 Mbps to 54 Mbps, but is fully backward compatible with 802.11b, so that hardware supporting both standards can operate in the same 2.4 GHz WLAN.

The OFDM modulation and coding scheme is identical to that applied in the 802.11a standard, with each 20 MHz channel in the 2.4 GHz band divided into 52 subcarriers, with 4 pilot tones and 48 data tones. Data rates from 6 to 54 Mbps are achieved using the same modulation methods and coding rates shown for 802.11a.

Although 802.11b and 11g hardware can operate in the same WLAN, throughput is reduced when 802.11b stations are associated with an 11g network (so-called mixed-mode operation) because of a number of protection mechanisms to ensure interoperability, as described below.

  • RTS/CTS - Before transmitting, 11b stations request access to the medium by sending a request to send (RTS) message to the access point. Transmission can commence on receipt of the clear to send (CTS) response. This avoids collisions between 11b and 11g transmissions, but the additional RTS/CTS signalling adds a significant overhead that decreases network throughput.
  • CTS to self - The CTS to self option dispenses with the exchange of RTS/CTS messages and just relies on the 802.11b station to check that the channel is clear before transmitting. Although this does not provide the same degree of collision avoidance, it can increase throughput significantly when there are fewer stations competing for medium access.
  • Backoff time - 802.11g backoff timing is based on the 802.11a specification (up to a maximum of 15 × 9 μS slots) but in mixed-mode an 802.11g network will adopt 802.11b backoff parameters (maximum 31 × 20 μS slots). The longer 802.11b backoff results in reduced network throughput.

A number of hardware manufacturers have introduced proprietary extensions to the 802.11g specification to boost the data rate above 54 Mbps. An example is D-Link’s proprietary “108G” which uses packet bursting and channel bonding to achieve a PHY layer data rate of 108 Mbps. Packet bursting, also known as frame bursting, bundles short data packets into fewer but larger packet to reduce the impact of gaps between transmitted packets.

Packet bursting as a data rate enhancement strategy runs counter to packet fragmentation as a strategy for improving transmission robustness, so packet bursting will only be effective when interference or high levels of contention between stations are absent.

Channel bonding is a method where multiple network interfaces in a single machine are used together to transmit a single data stream. In the 108G example, two non-overlapping channels in the 2.4 GHz ISM band are used simultaneously to transmit data frames.

Data Rates at the PHY and MAC Layer

In considering the technical requirements for a WLAN implementation in Chapter 7, it will be important to recognise the difference between the headline data rate of a wireless networking standard and the true effective data rate as seen by the higher OSI layers when passing data packets down to the MAC layer.

Each “raw” data packet passed to the MAC service access point (MAC SAP) will acquire a MAC header and a message integrity code and additional security related header information before being passed to the PHY layer for transmission.

The headline data rate, for example 54 Mbps for 802.11a or 11g networks, measures the transmission rate of this extended data stream at the PHY layer. The effective data rate is the rate at which the underlying user data is being transmitted if all the transmitted bits relating to headers, integrity checking and other overheads are ignored.

For example, on average, every 6 bits of raw data passed to the MAC SAP of an 802.11b WLAN will gain an extra 5 bits of overhead before transmission, reducing a PHY layer peak data rate of 11 Mbps to an effective rate of 6 Mbps. For 802.11g networks the MAC SAP data rate depends on the presence of 802.11b stations, as a result of the mixed-mode media access control mechanisms.