Page 257 - From GMS to LTE

Long Term Evolution (LTE) and LTE-Advanced Pro 243

scheme is too conservative and hence capacity on the air interface is wasted. In practice,
the air interface is best utilized if about 10% of the packets have to be retransmitted
because they have not been received correctly. The challenge of this approach is to
report transmission errors quickly and to ensure that packets are retransmitted as
quickly as possible to minimize the resulting delay and jitter. Further, the scheduler
must adapt the modulation and coding scheme quickly to keep the error rate within
reasonable limits. As in HSPA, the HARQ scheme is used in the Medium Access Control
(MAC) layer for fast reporting and retransmission. In LTE, the mechanism works as
described in the following sections.

HARQ Operation in the MAC Layer
In the downlink direction, asynchronous HARQ is used, which means that faulty data
does not have to be retransmitted straight away. The eNode‐B expects the mobile device
to send an acknowledgment (ACK) if the data within each 1‐millisecond subframe has
been received correctly. A negative acknowledgment (NACK) is sent if the data could
not be decoded correctly. HARQ feedback is sent either via the PUSCH or via the
PUCCH if the mobile device has not been assigned any uplink resources at the time the
feedback is required. This can be the case, for example, if more data is transmitted to a
mobile device in the downlink direction than the mobile device itself has to send in the
uplink direction.
If an ACK is received, the eNode‐B removes the subframe data from its transmission
buffer and sends the next chunk of data if there is more data waiting in the buffer. In
case a NACK is received, the eNode‐B attempts to retransmit the previous data block.
The retransmission can occur immediately or can be deferred, for example, owing to
the channel currently being in a deep fading situation for a particular user.
Before a data block is sent, redundancy bits are added to the data stream that can be
used to detect and correct transmission errors to a certain degree. How many of those
redundancy bits are actually sent depends on the radio conditions and the scheduler.
For good radio conditions, most redundancy is removed from the data stream again
before transmission. This is also referred to as puncturing the data stream. If a transmission
error has occurred and the added redundancy is not sufficient to correct the data, the
eNode‐B has several options for the retransmission:

It can simply repeat the data block.
●
It sends a different redundancy version (RV) that contains a different set of redun-
●
dancy bits, that is, some of those bits that were previously punctured from the data
stream. On the receiver side, this data stream can then be combined with the previous
one, thus increasing the number of available error detection and correction information.
The network can also decide to change the modulation and coding scheme for the
●
transmission to increase the chances for proper reception.
Repeating a data block requires time for both the indication of the faulty reception
and the repetition of the data itself. In LTE, the ACK/NACK for a downlink transmis-
sion is sent after four subframes to give the receiver enough time to decode the data.
The earliest repetition of a faulty block can thus take place five subframes or 5 millisec-
onds after the initial transmission. The eNode‐B can also defer the transmission to a
later subframe if necessary. Depending on the radio conditions and the modulation and
coding selected by the scheduler, some data blocks might have to be retransmitted
several times. This, however, is rather undesirable as it reduces the overall effectiveness.