Engineers turn to MBOK, CCK, CCSK, PPM, QAM, OCDM, and OFDM modulation schemes to provide higher data rates in next-generation DSSS WLAN systems.
By Carl Andren, Senior Systems Engineer, Harris Semiconductor
There has been a swift move toward higher data rates in the wireless local-area network (WLAN) industry. Companies are quickly developing WLAN systems that provide data rates hitting 11 Mb/s or higher. To achieve these higher speeds, direct sequence spread spectrum (DSSS) WLAN design engineers must evaluate and implement different modulation techniques.
Currently, DSSS WLAN systems employ binary phase-shift keying (BPSK) and quadrature phase-shift keying (QPSK) modulation techniques. Although these modulation schemes are sufficient in 1 and 2 Mb/s systems, they do not meet the demands of higher data rate transmission schemes.
To develop higher data rate DSSS WLAN systems, engineers need to replace BPSK and QPSK schemes with more complex modulation techniques. Some of the most common complex techniques used in high speed DSSS WLANs include M-ary orthogonal keying (MOK), complementary code keying (CCK), cyclic-code shift keying (CCSK), pulse-position modulation (PPM), quadrature amplitude modulation (QAM), orthogonal code division multiplexing (OCDM), and orthogonal frequency division multiplexing (OFDM).
MOK
MOK modulation techniques are well known and have been shown to be a good solution for developing high data rate WLAN systems. MOK techniques were extensively studied in the 60s in analog implementations. MOK techniques, however, did not catch on in these early analog systems because of their complexity. But now, with integrated digital implementations, engineers can effectively use MOK modulation schemes and gain the benefits of this waveform.
In WLAN applications, a variation of MOK, called M-ary bi-orthogonal keying (MBOK), is being implemented in higher data rate systems. MBOK allows one more bit per symbol essentially free. In addition, it lets WLAN systems provide multi-channel operation in the ISM band by virtue of keeping the total spread bandwidth the same as the existing IEEE 802.11 standard.
The MBOK spectrum is filtered to meet a spectral mask of -35 dB at +/-11 MHz and -50 dB at +/-22 MHz using a filter with a 17 MHz 3 dB. This lets WLAN systems offer three non-interfering channels in the 2.4 to 2.483 GHz ISM band while accounting for spectral energy reduction at the band edges.
In the MBOK modulation scheme, the spread function is picked from a set of M orthogonal vectors by data word (See Figure 1). Since the in-phase (I) and quadrature (Q) channels can be considered independent when coherently processed, both can be modulated this way.
Figure 1: In MBOK modulation techniques, the spread function is picked from a set of M orthogonal vectors.
Bi-orthogonal keying adds one more bit to each of the I and Q channels by using both true and inverted versions of the spread function (i.e. BPSK modulation). This allows MBOK-based systems to pack 8 bits into each symbol.
To offer the same bandwidth as existing IEEE 802.11 DSSS modulation specifications, MBOK-based WLAN systems employ an 11 Mchips/s chipping rate and an increased symbol rate of 1.375 MSamples/s. This allows the WLAN to offer an overall 11 Mb/s data rate in the WLAN device while still providing interoperability with current IEEE 802.11 preamble and header specifications.
In WLAN applications, MBOK modulation has also been shown to deliver slightly better Eb/N0 through its embedded coding properties. This allows the MBOK-based systems to tolerate more interference than other systems, such as BPSK- and QPSK-based devices.
CCK
MBOK modulation was promoted to the IEEE standard committee for inclusion into the standard for high-speed WLANs. The standard committee did not adopt this modulation, but did adopt a similar modulation technique known as CCK.
Jointly developed by Harris and Lucent Technologies, CCK is a form of MOK modulation where the code symbols are four phase modulated. Since CCK's symbols are QPSK in nature, they simultaneously occupy both the I and Q channels. The code set of complementary codes, however, is much richer than the set of Walsh codes (See Figure 2), so a much higher M can be used in the M-ary process. Thus, by using a set of 64 codes, CCK-based WLANs can modulate 6 bits on the M index and 2 bits on QPSK to create an 8 b code symbol that, in effect, has 16 b of complexity. This functionality allows CCK-based WLAN systems to offer 11-Mb/s data rates.
Figure 2: CCK's complementary codes are created as complex symbols so that an 8 chip symbol has, in effect, 16 b of complexity.
One of the main benefits of CCK is its ability to handle multipath interference. In multipath conditions, the absence of simultaneous orthogonal signals in CCK minimizes cross rail interference. This allows CCK-based devices to be less susceptible to multipath interference, which in turn allows these WLAN devices to provide better system performance.
CCSK
MOK modulation can also be accomplished using a form of PPM called CCSK. CCSK is simpler to demodulate than MBOK because it only correlates on one sequence. As a result, it is starting to find strong acceptance in high-speed WLAN designs.
One problem with CCSK is that it is not quite as efficient as MBOK because its symbols are not entirely orthogonal (they are trans-orthogonal). This efficiency problem, however, can be easily handled. By using cyclically shifted Barker words, CCSK-based WLAN designs can achieve the same Eb/N0 as MBOK-based approaches (See Figure 3).
Figure 3: By using cyclically shifted Barker words, CCSK modulation schemes can achieve the same Eb/N0 as MBOK modulation approaches.
CCSK is very similar to PPM. The main difference between the two arises during the modulation process. PPM techniques modulate the whole symbol while CCSK techniques only modulate the correlation pulse. In CCSK, this results in lower amplitude modulation in the transmitted waveform and therefore a lower power amplifier cost.
The main problem with any variety of PPM, including CCSK, is their susceptibility to interference. In CCSK-based WLANs, multipath delay spread can be more than a chip's length, causing interference problems in the WLAN system.
PPM
PPM is another popular waveform used to increase data rates in WLAN products. By definition, PPM is a form of pulse-time modulation in which the position in time of a pulse is varied.
In general, higher data rate WLAN systems employ a PPM waveform that consists of DSSS symbols with 11-chip Barker words, which are time shifted to provide up to 3 bits in the time shift. These symbols can be BPSK or QPSK modulated to give 1 or 2 more bits per symbol. In addition, when using PPM, both the I and Q channels can be modulated independently using BPSK to make a total of 8 bits per symbol. This allows PPM-based WLAN systems to achieve a total bit rate of 8 Mb/s.
When using PPM, engineers must remember one important property. In PPM, adjacent symbols are overlapped or have gaps between them. This property can be troublesome in WLAN applications. It creates an amplitude modulation of 6 dB and makes the transmit power amplifier less efficient.
QAM
Quadrature modulation occurs when two carrier components, differing in phase by 90 deg., are each modulated by a different signal. QAM is a form of quadrature modulation where some form of pulse amplitude modulation is used for both inputs.
Some companies have evaluated QAM for high-speed WLAN designs. But QAM has not proven to be an effective solution in these high-speed systems.
QAM with spreading is straightforward in concept, but suffers from low efficiency. Since this modulation scheme requires a very clean, undistorted signal, it is very sensitive to multipath interference.
Eb/N0 performance also causes problems in QAM-based WLAN systems. The Eb/N0 performance of QAM is not as good as MOK because it has both phase and amplitude components. As a result, QAM-based WLAN systems are much more sensitive to distortion and require an equalizer to operate properly.
The equalizer causes two headaches for WLAN designers. First, the equalizer requires a training sequence. This training sequence increases the length of the preamble and the complexity of the WLAN. Second, the equalizer adds cost to the overall system design. This is always a problem in the cost-sensitive wireless market.
Due to these problems, QAM has not proven to be an efficient and effective solution for higher data rate WLAN systems.
OCDM
Some designers have also looked at OCDM modulation for high-speed WLAN systems. To increase data rates, this modulation scheme simultaneously provides multiple spread channels on the same frequency. As a result, this method sends multiple streams of data on the same orthogonal channel (See Figure 4).
Figure 4: OCDM-based WLAN systems send multiple streams of data on the same orthogonal channel.
Sharp is using OCDM modulation for its 10 Mb/s WLAN modem. Through the OCDM modulation technique, Sharp says it has increased data rates by using CCSK Barker words for the orthogonal pseudorandom noise (PN) spread channels. In essence, Sharp's approach uses multiple CDMA channels to send more data.
Golden Bridge is also using OCDM modulation techniques in its WLAN products. In Golden Bridge's WLAN devices, Walsh codes are used for signal spreading. Under this approach, 16 channels of 16 chip orthogonal symbols are BPSK modulated and summed in an analog sense.
There are a few problems with Golden Bridge's approach. First, OCDM produces a high degree of amplitude modulation as it sums 16 independent channels, forcing the engineer to implement a very linear, and very costly, power amplifier to meet the spectral mask. Second, OCDM draws more power and eliminates the ability to provide limiting capabilities in the receiver.
OFDM
OFDM has been adopted by one or more announced 10 Mb/s suppliers when developing higher data rate WLANs. Formally accepted as the method to achieve better than 20 Mb/s data rates in the 5 GHz WLAN systems, OFDM provides multiple frequency channels at regular spacing, each modulated by PSK.
MIL-STD 188C has specified OFDM for decades for use in wireline and radio modems. It is commonly radiated over single sideband radios since it is very tolerant to spectral notches caused by multipath fading.
Some form of diversity is necessary to make OFDM work in a WLAN environment since a narrowband fade can remove one or more of the carriers. By spreading symbol energy over multiple frequencies, a robust link can be made.
OFDM makes best use of the spectrum with the channel filled edge to edge somewhat uniformly. This creates the least interference between users.
Despite these advantages, there are some problems with OFDM modulation schemes in WLAN applications. The long symbols of OFDM are said to make it more multipath resistant. But, the summing of 16 independent carriers can produce large amplitude modulation. This makes the transmitter difficult to design and rules out offering limiting capabilities in the WLAN receiver.
Processing is another problem in OFDM-based WLAN systems. The processing of OFDM is traditionally done with Fast Fourier transforms (FFTs) and inverse FFTs. FFTs are generally more complex and power hungry than the simple correlation techniques used by the other waveforms in high speed WLAN systems. This increases power consumption in WLAN systems, which can be costly to today's wireless manufacturers.
Carl Andren, Senior Systems Engineer, Harris Semiconductor, MS 62-028, Box 883, Melbourne, FL 32905. Phone: 407-724-7535; Fax: 407-724-7886; e-mail: candren@harris.com.