Given that receive signal powers are often less than a picoWatt, it is clear that RF
duplex spaced bands tend to deliver better receive sensitivity and therefore tend to be
used for wide area coverage systems. Wide area two-way radio networks in the UHF
band typically use 8 MHz or 10 MHz duplex spacing, 800/900 MHz cellular networks
use 45 MHz duplex spacing, GSM 1800 uses 95 MHz duplex spacing, PCS 1900 uses
80 MHz, and IMT2000 (3G) uses 190 MHz duplex spacing. In the United States, there
are also proposals to refarm 30 MHz of TV channel bandwidth in the 700 MHz band for
3G mobile services.
Figure 1.1 shows the duplex spacing implemented at 800/900 MHz for GSM in
Europe, CDMA/TDMA in the United States, and PDC (Japan’s 2G Personal Digital Cellular
standard) in Japan. PDC was implemented with 130 MHz duplex spacing (and 25
kHz channel spacing), thus managing to be different than all other 2G cellular standards.
In Asia, countries with existing Advanced Mobile Phone System (AMPS), and
CDMA/TDMA allocations have a problem in that the upper band of AMPS overlaps
the lower band of GSM. As the GSM band is paired, this means the corresponding
bands in the upper band of GSM are unusable. The result is that certain countries
(Hong Kong being the most obvious example) had a shortage of capacity because of
how the spectrum had been allocated. Latin America has the same 800/900 MHz allocation
as the United States (also shown in Figure 1.1). In the United States and Latin
America, however, the AMPS 2 × 25 MHz allocations are bounded by politically sensitive
public safety specialist mobile radio spectrum, preventing any expansion of the US
800 MHz cellular channel bandwidth.
In Europe, the original (1G) TACS allocation was 2 × 25 MHz from 890 to 915 MHz
and 935 to 960 MHz (1000 × 25 kHz channels), which was later extended (E-TACS) to
33 MHz (1321 × 25 kHz channels). GSM was deployed in parallel through the early to
mid-1990s and now includes 25 MHz (original allocation), plus 10 MHz (E-GSM), plus
4 MHz for use by European railway operators (GSM-R), for a total of 39 MHz or 195 ×
200 kHz RF channels
Additional spectrum was allocated for GSM in the early 1990s at 1800 MHz
(GSM1800). This gave three bands of 25 MHz each to three operators (75 MHz�"that is,
375 × 200 kHz paired channels). As with all duplex spaced bands, handset transmit is
the lower band. (Because of the slightly lower free space loss, this is better for a powerlimited
handset.) Only a fraction of this bandwidth is actually used, rather undercutting
operator’s claims to be suffering from a shortage of spectrum.
In the United States and Latin America, 2 × 60 MHz was allocated at 1850 to 1910
and 1930 to 1990 MHz for US TDMA (30 kHz) or CDMA (1.25 MHz) channels or GSM
(200 kHz) channels (GSM 1900), as shown in Figure 1.2. Unfortunately, the upper band
of PCS 1900 overlaps directly with the lower band of IMT2000, the official ITU allocation
for 3G. The intention for the IMT allocation was to make 2 × 60 MHz available,
divided into 12 × 5 MHz channels, and this has been the basis for European and Asian
allocations to date. In addition, 3 × 5 MHz nonpaired channels were allocated at 2010
to 2025 MHz and 4 × 5 MHz nonpaired channels at 1900 to 1920 MHz. The air interface
for the paired bands is known as IMT2000DS, and for the nonpaired bands, it is
IMT2000TC. (We discuss air interfaces later in this chapter.)
Figure 1.3 shows the RF bandwidth that needs to be addressed if the brief is to produce
an IMT2000 handset that will also work in existing 2G networks (GSM 900, GSM
1800, GSM 1900) co-sharing with US TDMA and CDMA.
Some countries have the 60 MHz IMT2000 allocation divided among five operators.
Five licensees sharing a total of 60 MHz would each have 12 MHz of spectrum. As this
is not compatible with 5 MHz channel spacing, two operators end up with 3 × 5 MHz
paired bands and three operators end up with 2 × 5 MHz paired bands and a nonpaired
band (either in TDD1 or TDD2). It will therefore be necessary in some cases to
support IMT2000DS and IMT2000TC in a dual-mode handset. The handset configuration
would then be IMT2000DS, IMT2000TC, GSM 1900, GSM 1800, and GSM 900.
Table 1.2 shows that selectivity and sensitivity are increasingly achieved at baseband,
reducing the requirement for RF filters and relaxing the need for frequency stability.
The need for backward compatibility, however, makes this benefit harder to realize.
First-generation AMPS/ETACS phones were required to access a large number of
25 kHz RF channels. This made synthesizer design (the component used to lock the
handset onto a particular transmit and receive frequency pair) quite complex. Also,
given the relatively narrowband channel, frequency stability was critical. A 1 ppm
(part per million) temperature compensated crystal oscillator was needed in the handset.
It also made network planning (working out frequency reuse) quite complex.
In second generation, although relaxing the channel spacing to 200 kHz reduced the
number of RF channels, the need for faster channel/slot switching made synthesizer
design more difficult. However, adopting 200 kHz channel spacing together with the
extra complexity of a frequency and synchronization burst (F burst and S burst)
allowed the frequency reference to relax to 2.5 ppm�"a reduction in component cost.
In third generation, relaxing the channel spacing to 5 MHz reduces the number of
RF channels, relaxes RF filtering, makes synthesizer design easier, and helps relax the
frequency reference in the handset (to 3 ppm). Unfortunately, you only realize these
cost benefits if you produce a single-mode IMT2000 phone, and, at present, the only
country likely to do this�"for their local market�"is Japan.
Additionally you might choose to integrate a Bluetooth or IEEE 802 wireless LAN
into the phone or a GPS (Global Positioning System/satellite receiver). In the longer
term, there may also be a need to support a duplex (two-way) mobile satellite link at
1980 to 2010 and 2170 to 2200 MHz. In practice, as we will see in the following chapters,
it is not too hard to integrate different air interfaces at baseband. The problem
tends to be the RF component overheads.
A GSM 900/1800 dual-mode phone is relatively simple, particularly as the 1800
MHz band is at twice the frequency of the 900 band. It is the add-on frequencies (1.2,
1.5, 1.9, 2.1, 2.4 GHz) that tend to cause design and performance problems, particularly
the tendency for transmit power at transmit frequency to mix into receive frequencies
either within the phone itself or within the network (handset to handset, handset to
base station, base station to handset, and base station to base station interference). And
although we stated that it is relatively easy to integrate different air interfaces at baseband,
it is also true to say that each air interface has its own unique RF requirements.
Sections
Comments (0 posted)
Syndication
