Downlink Spreading

With the exception of the synchronization
channels (SCHs), the downlink channels are spread to the chip rate and
scrambled, as shown in Figure 6-6. Each channel to be spread is split into
two streams—the I branch and the Q branch. The even symbols are mapped
to the I branch and the odd symbols are mapped to the Q branch. The
I branch is treated as a stream of real-valued bits, whereas the Q branch is treated as a stream of imaginary bits. Each of the two streams is spread
by the same channelization code. The spreading code/channelization code
to be used is taken from the same code tree as used in the uplink—that is,
OVSF codes that are chosen to maintain the orthogonality between different
channels transmitted from the same base station. The spreading rate
for a given channel depends on the channel in question.

The I and Q streams are then combined such that each I and Q pair of
chips is treated as a single complex value, such that the result of combining
them is a stream of complex-valued chips. This stream of chips is then subjected
to a complex downlink scrambling code, identified as Sdl,n in Figure 6-6.
One important difference occurs between spreading in the uplink and
downlink. In the uplink, the data for a given physical channel (such as a
single DPDCH) is directed either to the I branch or the Q branch, as shown
in Figure 6-2. Thus, on the uplink, no serial-to-parallel conversion takes
place. Therefore, for a spreading factor of, say, 8, the data rate of the physical
channel is simply 3,840,000/8  480 Kbps.

On the downlink, however, each channel (with the exception of the synchronization
channel) is subjected to a serial-to-parallel conversion, as
shown in Figure 6-6. For a given spreading factor, the serial-to-parallel conversion
effectively doubles the data rate of the physical channel. In other
words, half of the channel’s data is carried on the I branch with half on the
Q branch, and both of these are spread with the same spreading factor.

If, for example, we have a spreading factor of 8, then the data rate on the
I channel is 480 Kbps and the data rate on the Q channel is also 480 Kbps.
The net data rate is 960 Kbps—twice that achieved on the uplink for the
same value of spreading factor. In reality, however, the data rate on the
downlink is not quite twice that on the uplink. This is due to the fact, as is
explained later, that control information is time-multiplexed with a user
data on the downlink. This reduces the net throughput for a given downlink
data channel. Nonetheless, for a given spreading factor on the downlink, the
effective throughput is significantly greater than the corresponding
throughput on the uplink for the same spreading factor. 234