Modulation: Impact on Handset Design

Modulation: Impact on Handset Design

Information can be modulated onto an RF carrier by changing the amplitude of the carrier,
the frequency of the carrier, or the phase of the carrier. For example, using Minimum
Shift Keying (MSK), the carrier changes phase by +90° or -90° over a bit period
(see Figure 1.5).
The example shown in Figure 1.5 is a constant envelope phase modulation scheme.
Prior to modulation, the data stream passes through baseband filters. In Gaussian Minimum
Shift Keying (GMSK), these are Gaussian filters.

The advantage of GMSK, being constant envelope, is that it can be used with Class
C amplifiers, which typically have a power efficiency of between 50 and 55 percent.
The disadvantage is that with the GSM implementation of GMSK, because of the filtering,
decision points on the modulation trellis are not always obtained, resulting in
some residual bit errors. GMSK is a two-level modulation scheme�"that is, the two
phase states can represent a 0 or a 1.
Higher-level modulation states can be used to carry more bits per symbol. A fourstate
modulation scheme, for example, QPSK (Quadrature Phase Shift Keying) has 2
bits per symbol (00, 01, 11, 10), an eight-level modulation scheme can carry 3 bits per
symbol, a 16-level modulation scheme can carry 4 bits per symbol, a 1024-level modulation
scheme (used in fixed point-to-point, for example) can carry 10 bits per symbol.
However, as the number of modulation states increase, the distance between phase
states reduces and the likelihood of a demodulator error increases. Every time a modulation
level is doubled (for example, from two-level to four-level), an additional 3 dB of
signal energy is needed to maintain equivalent demodulator bit error rate performance.

Higher-level modulations also tend to contain amplitude components and can
therefore not be used with power-efficient Class C amplification. The modulation technique
used in IS54 TDMA is an example (see Figure 1.6).
This is a four-level modulation technique known as π/4DQPSK. DQPSK refers to
“differential quadrature phase shift keying,” the use of four differentially encoded
phase states to describe a 00, 01, 01, or 10. The π/4 indicates that the vector is indexed
by 45° at every symbol change. This makes it look like an eight-level modulation trellis,
which it isn’t. It shows that any change from phase state to phase state avoids passing
through the center of the trellis, which would imply a 100 percent AM component.
Instead the AM component is constrained to 70 percent.
Even so, the modulation requires a higher degree of linear amplification to avoid
spectral regrowth during and after amplification. While this is reasonably easily
accommodated in low-power handsets, it does result in larger�"and hotter�"RF amplifiers
in IS54 TDMA base stations.

Similarly, CDMA uses QPSK on the downlink and offset QPSK on the uplink (as
with π/4DQPSK, OQPSK reduces the AM components and relaxes the linearity
requirements of the handset PA). It is, however, difficult to realize efficiencies of more
than 7 to 8 percent in base station amplifiers (QPSK), substantially increasing the
power and heat dissipation needed, relative to GSM. This is why it has been easier to
produce very small picocellular base stations for GSM (1.5 kg) but harder to deliver an
equivalent form factor for IS54 TDMA or CDMA products.