Future Modulation Schemes

Future Modulation Schemes

The choice of modulation has always been a function of hardware implementation and
required modulation and bandwidth efficiency. In the 1980s, FM provided�"and still
provides today�"an elegant way of translating an analog waveform onto an (analog)
RF carrier.
In the 1990s, GMSK was used for GSM as a relatively simple way to digitally modulate
or demodulate an RF carrier without the need for linearity in the RF PA. Note that
GSM was developed as a standard in the early 1980s. US TDMA and IS95 CDMA were
specified/standardized toward the end of the 1980s, by which time four-level modulation
schemes (with AM components) were considered to provide a better efficiency
trade-off. Figure 1.7 compares the performance trade-offs of QPSK (1), MSK (2), and
GMSK (3). QPSK (1) carries 2 bits per symbol but has relatively abrupt phase changes
at the symbol boundaries. MSK (2) has a constant rate of change of phase but still manages
to maintain an open eye diagram at the symbol decision points. GMSK (3) has
additional filtering (a Gaussian baseband filter that effectively slows the transition
from symbol state to symbol state). The filtering ensures the modulation is constant
envelope; the disadvantage is that decision points are not always achieved, resulting in
a residual demodulated bit error rate.
QPSK is used in IMT2000MC and IMT2000DS on the downlink. HPSK is used on the
uplink to reduce linearity requirements. A variant of IMT2000MC known as 1xEV,
however, also has the option of using 8 PSK (also used in GSM EDGE implementation)
and 16-level QAM. This seems to be a sensible way to increase bandwidth efficiency,
given that eight-level modulation can carry 3 bits per symbol and 16 level can carry 4
bits per symbol.

It is necessary, however, to qualify the impact of the choice of modulation on the link
budget. For every doubling of modulation state, an additional 3 dB of link budget is
required to maintain the same demodulation bit error performance. Therefore, 8 PSK
needs 3 dB more link budget than QPSK, and 16-level QAM needs 3 dB more link budget
than 8 PSK. Provided you are close to the base station, you can take advantage of
higher-level modulation, but it will not deliver additional capacity at the edge of a cell.
It is also worth verifying the time domain performance of the demodulator.
The usual rule of thumb is that a demodulator can tolerate a quarter symbol shift in
terms of timing ambiguity without causing high demodulator error rates
In higher-level modulations, the symbol transition rate stays the same but the number
of symbol states increases. The symbol states become closer together in terms of
phase and frequency. Given that the vector is rotating, a timing error translates into a
phase or frequency error.
Multipath effects cause phase rotation and attenuation. In CDMA, these are partly,
though not totally, taken out by the RAKE receiver. Given that none of these adaptive
mechanisms are perfect, timing ambiguity translates into demodulator error rate. This
effect becomes more severe as bit rate and symbol rate increases. Thus, while higherlevel
modulation options promise performance gains, these gains are often hard to
realize in larger cells, particularly in edge-of-cell conditions where power is limited
and severe multipath conditions may be encountered.
An alternative is to use orthogonal frequency-division multiplexing (OFDM).
OFDM is sometimes described incorrectly as a modulation technique. It is more correctly
described as a multicarrier technique. Present examples of OFDM can be found
in wireline ADSL/VDSL, fixed access wireless, wireless LANs, and digital TV.
Standard terrestrial digital TV broadcasting in Europe and Asia uses QPSK. (Highdefinition
TV needs 16- or 64-level QAM and presently lacks the link budget for practical
implementation.) The QPSK modulation yields a 10.6 Mbps data rate in an 8 MHz
channel. The 8 MHz channel is divided into 8000 × 1 kHz subcarriers that are orthogonal
from each other. The OFDM signal is created using a Fast Fourier Transform. (Fast
Fourier Transforms were first described by Cooley and Tukey in 1963 as an efficient
method for representing time domain signals in the frequency domain.)
As there are now a total of 8000 subcarriers, the symbol rate per carrier is slow and
the symbol period is long compared to any multipath delays encountered on the channel.
Continuous pilot bits are spread randomly over each OFDM symbol for synchronization
and phase error estimation; scattered pilot bits are spread evenly in time and
frequency across all OFDM symbols for channel sounding.
The advantage of OFDM is that it provides a resilient channel for fixed and mobile
users. (DVB was always intended to provide support for mobility users.) The disadvantage
of OFDM is that it requires a relatively complex FFT to be performed in the
encoder and decoder. In digital TV, the power budget overheads associated with the
complex transform do not matter, in the context of transmitters producing kiloWatts of
RF power and receivers attached to a main supply.
Present implementation of an OFDM transceiver in a 3G cellular handset would,
however, not be economic in terms of processor and power budget overhead. OFDM is
however, a legitimate longer-term (4G) option providing a bandwidth efficient robust
way of multiplexing multiple users across 10, 15, or 20 MHz of contiguous bandwidth.