Radio Bandwidth Quality/Frequency Domain Issues

We have just described how code domain processing is used in IMT2000DS to improve
radio bandwidth quality. Within the physical layer, we also need to comprehend frequency
domain and time domain processing. If we wished to be very specific, we
would include source coding gain (using processor bandwidth to improve the quality
of the source coded content), coherence bandwidth gain (frequency domain processing),
spreading gain (code domain processing), and processing gain (time domain processing,
that is, block codes and convolutional encoders/decoders).
Let’s first review some of the frequency domain processing issues (see Table 3.6). We
said that the IMT2000 spectrum is tidily allocated in two 60 MHz paired bands
between 1920-80 and 2110 and 2170 MHz with a 190 MHz duplex spacing. In practice,
the allocations are not particularly tidy and vary in minor but significant ways country
by country.

 License A (Hutchison) has 14.6 MHz (3 × 5 MHz less a guard band) paired band
allocation.
 License B (Vodafone) has 14.8 MHz (3 × 5 MHz less a guard band) paired band
allocation.
 License C (BT3G) has 10 MHz (2 × 5 MHz allocation in the paired band) and 5
MHz at 1910 MHz in the TDD1 nonpaired band.
 License D (One2One) has 10 MHz (2 × 5 MHz allocation in the paired band) and
5 MHz at 1900 MHz in the TDD1 nonpaired band.
 License E (Orange) has 10 MHz (2 × 5 MHz in the paired band) and 5 MHz at
1905 MHz in the TDD1 nonpaired band.
The German allocation is different in that 10 MHz of paired bandwidth is allocated
to six operators (6 × 10 MHz = 60 MHz), then all four of the TDD1 channels are allocated
(1900 to 1920 MHz), along with one of the TDD2 channels (see Figure 3.23).
3GPP1 also specifies an optional duplex split of 134.8 and 245.2 MHz to support possible
future pairing of the TDD1 and TDD2 bands. Although this is unlikely to be
implemented, the flexible duplex is supported in a number of handset and Node B
architectures.
The fact that 5 MHz channels are allocated differently in different countries means
that operators must be prepared to do code planning and avoid the use of codes that
cause adjacent channel interference to either other operators in the same country or
other operators in immediately adjacent countries. It is therefore important to explore
the interrelationship between particular combinations of spreading codes and adjacent
channel performance.

The three measurements used are as follows:
 ACLR (Adjacent Channel Leakage Ratio), formerly Adjacent Channel Power
Ratio
 ACS (Adjacent Channel Selectivity)
 ACIR (Adjacent Channel Interference Ratio), formerly Adjacent Channel Protection
Ratio.
OVSF code properties also determine peak-to-average ratios (PAR), in effect the AM
components produced as a result of the composite code structure. PAR in turn determines
RF PA (RF Power Amplifier) linearity requirements, which in turn determine
adjacent channel performance. In other words, peak-to-average power ratios are a consequence
of the properties of the offered traffic�"the instantaneous bit rate and number
of codes needed to support the per user multiplex.
We find ourselves in an interactive loop: We can only determine frequency domain
performance if we know what the power spectral density of our modulated signal will
be, and we only know this if we can identify statistically our likely offered traffic mix.
Out-of-channel performance is qualified using complementary cumulative distribution
functions�"the peak-to-average level in dB versus the statistical probability that
this level or greater is attained. We use CCDF to calculate the required performance of
particular system components and, for example, the RF PA.
ACLR is the ratio of transmitted power to the power measured after a receiver filter
in the adjacent RF channel. It is used to qualify transmitter performance. ACS is the
ratio of receiver filter attenuation on the assigned channel frequency to the receiver
filter attenuation on the adjacent channel frequency and is used to qualify receiver
performance. When we come to qualify system performance, we use ACIR�"the adjacent
channel interference ratio. ACIR is derived as follows:

(We review system performance in Chapter 11 on network hardware.)
As a handset designer, relaxing ACLR, in order to improve PA efficiency, would be
useful. From a system design perspective, tightening ACLR would be helpful, in order
to increase adjacent channel performance. ACLR is in effect a measure of the impact of
nonlinearity in the handset and Node B RF PA.
We can establish a conformance specification for ACLR for the handset but need to
qualify this by deciding what the PAR (ratio of the peak envelope power to the average
envelope power of the signal) will be. We can minimize PAR, for example, by scrambling
QPSK on the uplink (HPSK) or avoiding multicodes. Either way, we need to ensure the
PA can handle the PAR while still maintaining good ACL performance. We can qualify
this design trade-off using the complementary cumulative distribution function.
A typical IMT2000DS handset ACLR specification would be as follows:
 33 dBc or -50 dBm at 5 MHz offset
 43 dBc or -40 dBm at 10 MHz offset
94