Page 264 - From GMS to LTE

250 From GSM to LTE-Advanced Pro and 5G

or a different user. In practice, however, such QoS attributes are not widely used except
for VoLTE and hence most bearers for Internet traffic have the same priority on the
radio interface.
For bearers with the same priority, other factors may influence the scheduler’s deci-
sion as to when to schedule a user and how many RBs are allocated to them in each
subframe. If each bearer of the same priority was treated equally, some capacity on the
air interface would be wasted. With this approach, mobile devices that currently or
permanently experience bad radio conditions, for example, at the cell edge, would have
to be assigned a disproportionate number of RBs because of the low modulation and
coding scheme required. The other extreme is to always prefer users that experience
very good radio conditions, as this would lead to very low datarates for users experienc-
ing bad radio conditions. As a consequence, proportional fair schedulers take the
overall radio conditions into account, observe changes for each user over time and try
to find a balance between the best use of the cell’s overall capacity and the throughput
for each user.
Scheduling downlink data for a user works as follows. For each subframe the eNode‐B
decides the number of users it wants to schedule and the number of RBs that are
assigned to each user. This then determines the required number of symbols on the
time axis in each subframe for the control region. As shown in Figure 4.8, there are a
total of 2 × 7 = 14 symbols available on the time axis if a short cyclic prefix is used.
Depending on the system configuration and the number of users to schedule, one to
four symbols are used across the complete carrier bandwidth for the control region.
The number of symbols can either be fixed or changed as per the demand.
The eNode‐B informs mobile devices about the size of the control region via the
PCFICH, which is broadcast with a very robust modulation and coding scheme. The
two bits describing the length of the control region are secured with a code rate of 1/16,
which results in 32 output bits. QPSK modulation is then used to map these bits to 16
symbols in the first symbol column of each subframe.
With the mobile device aware of the length of the control region, it can then calculate
where its search spaces are located. As described above, search spaces have been intro-
duced to reduce the mobile device’s processing load in order to save battery capacity. In
mobile device (UE)‐specific search spaces that are shared by a subset of mobile devices
or in common search spaces that have to be observed by all mobile devices for broad-
cast messages, the mobile decodes all PDCCH messages. Each message has a checksum
in which the mobile’s identity is implicitly embedded. If the mobile can correctly calcu-
late the checksum, it knows that it is the intended recipient of the message. Otherwise
the message is discarded.
The length of a PDCCH message is variable and depends on the content. For easier
decoding, a number of fixed‐length PDCCH messages have been defined. A message is
assembled as follows. On the lowest layer, four symbols on the frequency axis are grouped
into a Resource Element Group (REG). Nine REGs form a CCE. These are further aggre-
gated into PDCCH messages, which can consist of 1, 2, 4 or 8 CCEs. A PDCCH message
consisting of two CCEs, for example, contains 18 REGs and 18 × 4 = 72 symbols. In other
words, it occupies 72 subcarriers. With QPSK modulation, the message has a length of
144 bits. The largest PDCCH message with eight CCEs has a length of 576 bits.
A PDCCH message can be used for several purposes and as their lengths differ, several
message types exist. In the standards, the message type is referred to as Downlink