20 Harq

HARQ
17 August 2019
ARQ and HARQ
 Automatic Repeater reQuest ARQ is used to handle transmission errors.
 The received uses an error detection code typically Cyclic Redundancy Check (CRC) to
check if the received packet is in error or not.
 LTE employs a combination of forward error –correction coding and ARQ known as
Hybrid ARQ.
 In Hybrid ARQ with soft combining, the erroneously received packet is stored in
buffer memory and later combined with retransmission to obtain a single combined
packet that is more reliable than its constituents.
 If decoding with CRC code fails, a retransmission is requested.
 HARQ is supported only for DLSCH and ULSCH
2015 © Samsung Electronics 2

ARQ and HARQ
 Difference between ARQ and HARQ
 ARQ:
 Works at RLC layer
 If the received data has an error (as detected by ARQ) then it is discarded,
and a new re-transmission is requested from the sender
 HARQ:
 Works at PHY layer but controlled by MAC layer
 If the received data has an error then the Receiver buffers the data and
requests a re-transmission from the sender.
 When the receiver receives the re-transmitted data, it then combines it
with buffered data prior to channel decoding and error detection. This
helps the performance of the re-transmissions.

LTE Protocol Architecture (Downlink)

Multiple HARQ Processes
 LTE HARQ uses multiple parallel stop-and-wait processes.
 Upon reception of a transport block, the receiver makes an attempt to decode the
transport block and informs the transmitter through a single acknowledgement bit.
Multiple ARQ processes combines the
simplicity of a stop-and-wait protocol
while still allowing continuous
transmission of data
Multiple Hybrid ARQ Processes
In example above - TB5 is decoded before TB1 which requires retransmissions – then in sequence delivery
is ensured by the RLC layer
Downlink retransmissions may occur at any time after initial transmission - the protocol is asynchronous
HARQ SAW
 The HARQ protocol relies upon the sender receiving acknowledgements from the
receiver. The round trip time, which includes both the sender and receiver processing
times as well as the propagation delays, means that these acknowledgements are not
received instantaneously. In general, the propagation delays are negligible compared
to the processing delays
 The impact of waiting for an acknowledgement is shown in Figure below. The sender
becomes inactive while waiting for the acknowledgement so the average throughput
is relatively low. This corresponds to using a single Stop And Wait (SAW) process. A
SAW process stops and waits for an acknowledgement before proceeding to transfer
any further data
Data transfer using a single Stop And Wait (SAW) process

HARQ SAW
 Multiple parallel SAW processes are used to avoid the round trip time having an
impact upon throughput, i.e. additional SAW processes transfer data while the first
SAW process is waiting for its acknowledgement. Larger round trip times require an
increased number of parallel SAW processes. The concept of multiple parallel SAW
processes is illustrated in Figure
 The example in Figure below requires 8 parallel SAW processes to maintain a constant
flow of data. These SAW processes are also referred to as HARQ processes. The HARQ
entity within the MAC layer manages these multiple HARQ processes
Data transfer using multiple parallel Stop And Wait (SAW) process
HARQ SAW
 The duration associated with each transmission corresponds to the Transmission Time
Interval (TTI) which is represented by the 1 ms subframe in LTE.
 Figure in previous slide illustrates 8 parallel HARQ processes so there is an 8 ms cycle
between sending data from a specific SAW process and being able to send a
retransmission or new data from the same SAW process.
 This allows for a total round trip time of 8 ms (UMTS HSDPA uses 6 HARQ processes and a
2 ms TTl to allow for a 12 ms round trip time)
 The sender buffers the transmitted data until a positive acknowledgement has been
received in case a retransmission is required.
 Data is cleared from the transmit buffer once a positive acknowledgement has been
received or the maximum number of allowed retransmissions has been reached. New
data can be sent by a specific HARQ process once its transmit buffer has been cleared
 The uplink and downlink of LTE are both specified to use parallel HARQ processes.
 However, the uplink and downlink differ in some respects, e.g. the uplink uses
synchronous HARQ whereas the downlink uses asynchronous HARQ. Asynchronous
HARQ provides greater flexibility but requires additional overhead in terms of signalling
which HARQ process is being used at any point in time
 3GPP References: TS 36.213, TS 36.321

User Plane Protocol Stack
 LTE Layer 2 user – plane protocol stack is composed of three Sublayers :
 PDCP: Packet Data Convergence Protocol. This layer processes Radio Resource Control
(RRC) packets messages in the control plane and IP packets in the user plane. Main
functions Header Compression, Security (integrity protection and ciphering) and
support for re-ordering and retransmission during handover.
 RLC: Radio Link Control. Main functions are segmentation and reassembly of upper
layer packets in order to adapt them to the size which can be transmitted over radio
interface. The RLC layer performs re-transmissions to recover from packet losses.
Additionally the RLC layer performs reordering to compensate for out of order reception
due to HARQ ( Hybrid Automatic Repeat reQuest) operation in the layer below.
 MAC :Medium Access Control. This layer performs multiplexing of data from different
radio bearers. The MAC layer aims to achieve the negotiated Quality of Service (QoS) for
each radio bearer. For the up link this process includes to report to the eNodeB the
amount of buffered data for transmission.

User Plane Architecture
At the transmitting side, each layer receives SDU (Service Data Unit) from higher layer and
provides PDU (Protocol Data Unit) to lower layer. Example : RLC layer received PDCP PDUs from
PDCP point of view and RLC SDUs from RLC point of view.
At the receiving side the process is reversed.
Length of SDUs and PDUs is always in multiples of 8 bits (byte aligned)
Data Flow through various layers
A Transport Block is equivalent
to a MAC PDU

Radio Link Control RLC
 The RLC layer is location between the PDCP layer and MAC layer
 The RLC layer reformats the PDCP PDUs in order to fit them into the size indicated
by the MAC layer, that is, the RLC transmitter segments and/or concatenates the
PDCP PDUs and the RLC receiver reassembles the RLC PDUs to reconstruct the PDCP
PDUs
 In addition the RLC reorders the RLC PDUs if they are received out of sequence due
to HARQ operation performed in the MAC layer.
 An RLC entity is configured in one of the three data transmission modes –

Transparent mode TM, Unacknowledged Mode UM and Acknowledged Mode AM.
In AM special functions are defined to support retransmission.
When two code words are transferred, they need not be of equal size.
CQI reporting, Link Adaption and HARQ run independently for each code word

RLC ARQ
 In LTE, two retransmission mechanisms are used.
 Retransmissions of missing data or data units in error are handled primarily by the
hybrid –HRQ mechanism in MAC layer , complemented by the ARQ retransmission
functionality of the RLC layer

MAC and RLC layers
RLC entity is responsible for

 Segmentation , concatenation and re-assembly of PDUs
 RLC re-transmissions
 In-sequence delivery and duplicate detection for the corresponding logical channel
RLC can operate in three different modes depending on the requirements from the
application :
TM – Transparent Mode
UM- Unacknowledged Mode
AM- Acknowledged Mode

RLC Modes
 Transparent Mode TM : RLC is essentially bypassed. No retransmissions, no
segmentations/ reassembly and no in-sequence delivery takes place. Used for Control
Plane Broadcast Channels (BCCH,CCCH and PCCH)
 Unacknowledged Mode UM : supports segmentation/reassembly and in-sequence

delivery but not retransmissions – used for Voice over IP and broadcast transmissions
using MBSFN
 Acknowledged Mode AM : used for TCP/IP packets on DL-SCH.

Segmentation/reassembly, in-sequence delivery and retransmissions all are
supported

Acknowledged Mode AM
 In- Sequence Delivery : SDU n should be delivered prior to SDU n+1. PDUs are stored
in a buffer until all PDUs with lower sequence number are delivered.
Re-ordering timer is
Duplicate detection is also responsibility of RLC – if packer n+2
started for missing PDU.
arrives again, it is discarded
Re-transmission is
Status reports are sent on expiry of re-ordering timer
requested with a status
Status prohibit timer is used to control status report which
report if this timer
cannot be transmitted more than once per timer interval
expires
Acknowledged Mode AM (Contd)
(a) In the above example, at t3 PDU up to n+5 have been transmitted.

(b) Only PDU n+5 arrived and PDU n+3 and n+4 are missing .
(c ) Re-ordering timer starts. No PDU arrives prior to expiry of the timer. At time t4, the
receiver sends a control PDU containing a status report indicating missing PDUs to its peer
entity. Control PDUs have higher priority than data PDUs.
(d) Upon receipt of the status report at time t5, the transmitter knows that PDUs up to n+2
have been received correctly and the transmission window is advanced. The missing PDUs n+3
and n+4 are retransmitted.
At t6, all PDUs have been successfully delivered.
The transmitted sends a flag in the header of the last RLC data PDU to which the receiver will
respond with the requested status report acknowledging all PDUs upto n+5.
Upon receipt of the same the transmission window is advanced by the transmitter.
MAC Layer
 The MAC layer consists of a HARQ entity, a multiplexing / de-multiplexing entity, a logical
channel prioritization entity, a random access control entity and a controller which performs
various control functions
 The HARQ entity is responsible for the transmit and receive HARQ operations.
 The transmit HARQ operation includes transmission and retransmission of TBs and reception
and processing of ACK/NACK signaling
 The receive HARQ operation includes reception of TBs, combining of the received data and
generation of ACK/NACK signaling.
 In order to enable continuous transmission while previous TBs are being decoded, up to
eight HARQ processes in parallel are used in FDD to support multi-process ‘Stop And Wait’
SAW HARQ operation .
 SAW operation means that upon transmission of a TB, a transmitter stops further
transmission and awaits feedback from the receiver. When NACK is received or when certain
time elapses without receiving any feedback the transmitter retransmits the TB

HARQ Schemes
 HARQ schemes can be categorized as either synchronous or asynchronous with the
retransmission in each case being either adaptive or non adaptive
 In a synchronous HARQ scheme the retransmissions for each process occur at a predefined
time relative to the initial transmission. In this way there is no need to signal information
such as HARQ process number as this can be inferred from the transmission timing
 In an asynchronous HARQ scheme, the retransmissions can occur at any time relative to the
initial transmission so additional explicit signaling is required to indicate the HARQ process
number to the receiver so that the receiver can correctly associate each retransmission with
the corresponding initial transmission.
 Thus synchronous HARQ reduces signaling overhead while asynchronous HARQ allow more
flexibility in scheduling
 In an adaptive HARQ scheme, transmission attributes such as Modulation and Coding (MCS)
and transmission resource allocation in the frequency domain can be changed at each
retransmission in response to variations in the radio channel conditions .
 In non adaptive HARQ scheme, the retransmission are performed without explicit signaling
of new transmission attributes – either by using the same transmission attributes as those of
previous transmission or by changing the attributes according to a predefined rule. Thus
adaptive schemes bring more scheduling gain at the expense of increased signaling
overhead.
 In LTE asynchronous adaptive HARQ is used for downlink and synchronous HARQ for the
uplink. In the uplink the retransmission may either be adaptive or non adaptive.
Synchronous and Asynchronous HARQ
Uplink
Down link
Asynchronous Adaptive HARQ for Downlink

Synchronous HARQ for Uplink (Adaptive & Non Adaptive)
For HARQ, a 24 bit CRC is attached to the Transport Block for error detection at the receiver
Adaptive HARQ
 Asynchronous and Adaptive HARQ is used in Downlink to avoid collisions with
transmission of system information and MBSFN subframes
 Synchronous HARQ is used in uplink due to lower signaling overhead required
 NDI New Data Indicator , HARQ process number (three bits for FDD and four bits for TDD),
Redundancy Version (two bits) are signaled as part of Down Link Control Signaling
 Downlink Spatial Multiplexing implies transmission of two transport blocks in parallel on a
component carrier. Each Transport Block has its NDI and RV indication. The process
number for the second block is offset from that of the first block.
 In case of spatial multiplexing, each transport block has its HARQ bit else for a single
transport block there is only one acknowledgement bit.
 For uplink basic mode is synchronous non adaptive HARQ but in some cases adaptive
HARQ may be used as shown below where the resource block set and MCS is changed for
retransmission (to make room for another terminal in n+16). Redundancy Version follows a
predefined pattern for uplink.

HARQ in LTE – FDD Downlink
 The Physical layer in LTE supports HARQ on the physical downlink and uplink shared
channels with separate control channels to send associated acknowledgement feedback
 In FDD eight Stop and Wait SAW – HARQ processes are available in both downlink and
uplink with a typical Round Trip Time (RTT) of 8 ms.
 Each HARQ process requires a separate soft buffer allocation in the receiver for the
purpose of combining the retransmissions
Downlink SAW HARQ Protocol Timing

diagram
Re-transmissions are scheduled at
HARQ in LTE – FDD Uplink fixed time intervals.
Uplink SAW HARQ Protocol Timing diagram

 In FDD the HARQ process to which a transport block belongs is identified by a unique three
bit HARQ process Identifier (HARQ ID).
 In TDD the number of HARQ processes depends on the uplink/downlink configuration and
four bits are used to identify the process
 There are several field in the downlink control information to aid the HARQ operation :
 NDI New Data Indicator – whether new transmission or retransmission
 RV Redundancy Version – indicates the RV selected for the transmission or retransmission
 MCS Modulation and Coding Scheme
 Downlink HARQ is asynchronous and adaptive – therefore every downlink transmission is
accompanied by explicit signaling of control information
 The Uplink HARQ is synchronous and either non adaptive or adaptive. The uplink non adaptive HARQ
operation requires a predefined RV sequence 0,2,3,1,0,2,3,1……for successive transmissions of a packet
due
2015 © toElectronics
Samsung absence of explicit control signaling 23
Synchronous HARQ in UL (FDD)
 Synchronous for UL ( Uplink Data transmission Processor or UL-HARQ )
 UE sends data to eNodeB in the form of Transport Block. If transport block receives
correctly, ACK is sent otherwise NACK. PHICH physical channel is used to carry HARQ in
the downlink direction for the received uplink data.
 Re-transmissions are scheduled at fixed time intervals.
 Uses the specific process in a specific subframe. Every 8 subframes UE repeats the
process id. As a result, eNB knows exactly when each HARQ process comes.
 UL HARQ Process ID = (SFN x 10 + subframe) modulo 8
 Always works in cycle even if no resources are allocated during a specific sub frame;
which means that the 1st process will repeat itself after every 8 ms.
 Since UE have to use specific HARQ process ID at specific subframe, the receiver
(eNode B) knows exactly which HARQ process comes when. And eNodeB can also
knows about RV because UL Grant (DCI 0) from eNodeB can specify RV using MCS
field.
 As mentioned, above HARQ is synchronous in the UL. Whenever, a re-
transmission occurs in UL, it can be either Adaptive and Non-adaptive.

FDD Uplink HARQ
 HARQ for the uplink refers to transferring uplink data on the PUSCH while receiving
downlink acknowledgement on the PHICH
 HARQ is synchronous in the uplink direction – means retransmissions are scheduled at
fixed time intervals relative to initial transmission – does not require HARQ identity to be
signaled so generates a lower overhead
 There is one HARQ entity at the UE for each serving cell. There are multiple serving cells
and multiple HARQ entities when LTE Advanced Carrier Aggregation is used
 Each HARQ entity maintains a number of synchronous parallel HARQ processes. The total
number of HARQ processes is summarized in Table below. TTI bundling is not used when
Transmission Mode 2 (UL) is configured, i.e. it is not used with uplink MIMO
 In the case of Transmission Mode 2 (for UL), the number of HARQ processes doubles for
both 2x2 MIMO and 4x4 MIMO.
 The number of HARQ does not quadruple for 4x4 MIMO because it uses two transport
blocks which are subsequently segmented into four blocks prior to transmission on the air
interface
MIMO is not used for TTI Bundling

There is no TTI bundling in down link HARQ processes for Uplink FDD
Uplink HARQ Timing for FDD
The use of 8 HARQ processes for FDD Transmission Mode 1 is illustrated in Figure below.
8 parallel HARQ Stop And Wait (SAW) processes are used to provide continuous data transmission while allowing for a
maximum round trip time of 8 ms
Requirement for acknowledgements doubles when spatial multiplexing is used

Value
2015 of Electronics
© Samsung k is always 4 for FDD 26
Uplink HARQ Timing for FDD
 HARQ acknowledgements are provided on the PHICH.
 They are sent 4 subframes after uplink transmission on the PUSCH.
 A retransmission or new data transmission using the same HARQ process can then
follow 4 subframes after the acknowledgement
 The requirement for acknowledgements doubles when spatial multiplexing is used
and 2 transport blocks are transferred during each subframe.
 The pair of acknowledgements associated with the pair of transport blocks are
transferred during the same subframe but using different PHICH resources
(combination of PHICH group and orthogonal sequence)
 The variable 'k' defines the difference between the subframe number associated with
the HARQ acknowledgement and the subframe number associated with the PUSCH
transmission. This variable is always equal to 4 for FDD

FDD Uplink HARQ
 Incremental redundancy is always possible in the uplink direction because the eNode B has a
relatively large memory capability.
 The physical layer puncturing pattern is determined by the Redundancy Version (RV) which
uses a value of 0 for the first transmission to provide the systematic bits with priority:
• the RV cycles through the values 2, 3, 1, 0 when retransmissions are triggered by a
negative acknowledgement on the PHICH (non-adaptive retransmission)
• the RV is signalled explicitly as part of the Modulation and Coding Scheme (MCS) when
retransmissions are triggered by the PDCCH (adaptive retransmission)
 Retransmissions can be either non-adaptive or adaptive:

• non-adaptive retransmissions are triggered by a negative acknowledgement on the
PHICH. Non-adaptive retransmissions use the same set of resources as the previous
transmission, i.e. the modulation scheme and the set of allocated Resource Blocks
remain unchanged. However, the RV does change between transmissions
• adaptive retransmissions can be triggered by the PDCCH Downlink Control Information
(DCI) formats 0 and 4. The New Data Indicator (NDI) flag triggers a retransmission if its
value is not toggled relative to the previous transmission. Adaptive retransmissions
allow the set of allocated Resource Blocks as well as the RV to change
FDD Uplink HARQ
 Table below summarizes the difference between non-adaptive and adaptive retransmissions
 PDCCH DCI formats 0 and 4 takes precedence over the PHICH. The PHICH can indicate a
positive acknowledgement but a retransmission is still triggered if a PDCCH is received
during the relevant subframe with the NDI bit untoggled. Both PHICH acknowledgements
and PDCCH instructions are received 4 subframes after transmission on the PUSCH. The rules
for interpreting the combination of PHICH and PDCCH acknowledgements are summarized in
Table
Retransmission triggering using the PHICH and PDCCH

UE HARQ Buffer Size
 A major contributor to the UE implementation complexity is the UE HARQ soft buffer size
 For each UE category the soft buffer size is determined based on the instantaneous peak
rate supported per subframe multiplied by eight (using eight HARQ processes).
 For FDD the soft buffer is split into eight equal partitions (one partition per TB) if the UE
is configured to receive PDSCH in TM Mode other than 3 or 4 (one TB per HARQ process)
or sixteen equal partitions for TM 3 or 4 (two TB per HARQ process).
 A Transport Block is equivalent to a MAC PDU
 It is important to note that all transmission Modes are supported by all UE categories
(the only exception are that PDSCH transmission modes 7 and 8 are optional for FDD UEs
and PUSCH TM 2 is optional for UE categories 1-7)
 A category 1 UE could be in TM 3 or 4 but limited to rank-1 operation as it can support
only one layer in spatial multiplexing
 Soft Buffer sizes for TDD were chosen same as FDD
 The number of HARQ processes in TDD varies between 4 and 15 according to the
downlink / uplink configuration
 Hence soft buffer size is adapted and split into min(MDL_HARQ , Mlimit) equal partitions if
the UE is configured to receive PDSCH in other than TM 3 or 4 or 2.min(MDL_HARQ , Mlimit)
when configured for TM 3 or 4, where MDL_HARQ , is the number of downlink HARQ
processes and Mlimit is equal to eight.
 Thus for TDD when the number of downlink HARQ processes is greater than 8,
statistical buffer management can be used for efficient UE implementation
Transmission Modes for PDSCH
1. When applied for user data – one or at most two TBs can be transmitted per UE per sub frame
depending on the transmission mode selected for the PDSCH for each UE
2. In case of closed loop precoding, the network selects the pre-coder matrix based on the feedback
from the terminal.
3. Based on measurements of CRS, the terminal selects suitable transmission rank and pre-coder matrix
and reports RI Rank Indication and PMI Pre-coder Matrix Indication
4. A limited set of pre-coder matrices called as codebook is defined for each transmission rank – only
index of the selected matrix then needs to be signaled
5. Open Loop Pre-coding – the pre-coder matrix is selected in a predefined manner – used for high
mobility scenarios. Codebook based pre-coding allows maximum of 4 antenna ports & 4 layers
6. Non code book processing uses DM RS for channel estimation & demodulation and no PMI is sent
FDD Down Link HARQ
 HARQ for the downlink refers to transferring downlink data on the PDSCH while
receiving uplink acknowledgements on the PUCCH or PUSCH
 There is one HARQ entity at the UE for each serving cell. There are multiple serving
cells and multiple HARQ entities when LTE Advanced Carrier Aggregation is used
 Each HARQ entity maintains a number of asynchronous parallel HARQ processes. The
total number of HARQ processes is summarized in Table below. TTI bundling is not
applicable to the downlink
Total number of HARQ processes for the downlink of FDD

FDD Down Link HARQ
receiving uplink acknowledgement on the PUCCH of PUSCH
 Two main differences with uplink FDD HARQ are as under :
• in the case of the downlink, there is no dependency upon whether or not spatial
multiplexing is used.
• In the downlink direction, when 2 transport blocks are transferred during the
same subframe using spatial multiplexing, then those 2 transport blocks are
associated with the same HARQ process rather than 2 different HARQ
processes. This avoids the requirement to double the number of HARQ
processes when using spatial multiplexing
• in the case of the downlink, the number of HARQ processes is 'up to' 8 and 'up
to' 4 rather than exactly 8 or exactly 4.
• This results from HARQ being asynchronous in the downlink direction so
fewer HARQ processes can be used when a connection is not scheduled data
during every subframe

FDD Down Link HARQ
 HARQ is asynchronous in the downlink direction. The eNode B provides the UE with
instructions regarding which HARQ process to use during each subframe that
resources are allocated.
 The HARQ process identity is included within PDCCH DCI formats 1, 1A, 1B, 1D, 2, 2A,
2B and 2C
 Asynchronous HARQ increases the signalling overhead as a result of the requirement
to include the HARQ process identity within the DCI. However, asynchronous HARQ
also increases flexibility because retransmissions do not have to be scheduled during
specific subframes.
 The eNode B may wish to postpone a retransmission because other UE are being
scheduled during that subframe, or because the channel conditions have become
relatively poor
 Up to 8 parallel HARQ Stop And Wait (SAW) processes are used to provide continuous
data transmission while allowing for a maximum round trip time of 8 ms.
 The use of asynchronous HARQ means that less than 8 HARQ processes can be used if
the UE is not scheduled during every subframe.
 This is in contrast to synchronous HARQ in the uplink which always cycles through the
set of 8 HARQ processes irrespective of whether or not resources are allocated
during a specific subframe

FDD Down Link HARQ
 When data transfer is continuous and the UE is scheduled during every subframe then 8
HARQ processes are required and the timing is the same as for the uplink, i.e. as illustrated
in Figure for FDD uplink
 Both chase combining and incremental redundancy are used in the downlink. Incremental
redundancy is the preferred method but chase combining is selected when incremental
redundancy is not possible. 3GPP TS 36.306 specifies a 'Total Number of Soft Channel Bits'
capability for each UE category. This represents a buffer size used to store the data
received by the set of HARQ processes. The eNode B keeps track of the UE buffer
occupancy and switches to chase combining when there is insufficient capacity to
support incremental redundancy
 Chase combining is most likely to be required when UE are allocated their maximum
throughput capability, i.e. the quantity of data filling the UE buffer is large. When using
chase combining, the Redundancy Version (RV) is set to 0 to provide the systematic bits
with the greatest priority
 The RV value is signalled explicitly with the downlink resource allocations within PDCCH
Downlink Control Information (DCI) formats 1, 1A, 1B, 1D, 2, 2A, 2B and 2C.
 DCI format 1 C does not specify an RV value because this DCI is used to schedule resources
for random access responses, paging messages, and system information. Random access
responses and paging messages always use an RV value of 0. System information uses RV
values which are calculated from the System Frame Number (SFN) and subframe number
FDD Down Link HARQ
 The selection of RV values is implementation dependent but in general the uplink
pattern specified by 3GPP can be used for the downlink, i.e. original transmission
uses RV=0 while retransmissions cycle through the values 2, 3, 1, 0
 HARQ retransmissions in the downlink can always be adaptive because the PDCCH
can allocate a different set of resources for the original transmission and each
retransmission.
 The MCS indices 29, 30 and 31 within TBS Table illustrate that the modulation
scheme can be changed between the original transmission and each retransmission.
This is in contrast to the uplink MCS indices 29, 30 and 31 within TBS Table which
only allow the RV value to change
 The UE returns acknowledgements to the eNode B using either the PUCCH or PUSCH.
In the case of FDD without Carrier Aggregation, it is necessary to transfer 1 HARQ
acknowledgement per subframe when spatial multiplexing is not used, and 2 HARQ
acknowledgements per subframe when spatial multiplexing is used. These
requirements can be satisfied by the capacity of the PUCCH without any additional
processing
 The eNode B uses the New Data Indicator (NDI) flag within the PDCCH DCI to inform
the UE of whether it is sending a new transport block or a retransmission. Toggling
the value relative to the previous value used for the same HARQ process indicates
that new data is being sent rather than a retransmission
PCFICH
 Three different CFI values are used in LTE; Each code word of CFI is 32 bits in length
and mapped to 16 Res using QPSK
 These 16 REs are arranged in groups of 4 known as REGs (Resource Element Groups)
 The four REGs carrying PCFICH are distributed in frequency domain; A CCE consists of
nine REGs
 It carries the number of symbols that can be used for control channels (PDCCH and
PHICH).
 Mapped to the first OFDM symbol in each of the downlink sub-frame
 This contains the information on number of OFDM symbols for PDCCH and PHICH
 PCFICH data is carried by 4 REGs and these four REGs are evenly distributed across
the whole band regardless of the bandwidth.
 The exact position of PCFICH is determined by cell ID and bandwidth.

PCFICH Location
 PCFICH carried the CFI number which can be 1 or 2 or 3.
 Regardless of System Bandwidth, PCFICH is always carried by 4 REGs (16 REs) at the first symbol of each
subframe. The exact location of these four REGs for PCFICH is determined by Physical Cell ID and System BW
according to formula as shown below.

PHICH
 PHICH Physical Hybrid ARQ Indicator Channel for the uplink data
 The MIB information consists of the downlink system bandwidth, the PHICH size, most
significant eight bits of System Frame Number SFN
 The PHICH carried the HARQ ACK/NACK (0 for ACK and 1 for NACK). This information is then
repeated in each of the three BPSK symbols. Multiple PHICHs are mapped to the same set of
REs constituting a PHICH group. Orthogonal Walsh Sequences are used to separate the
PHICHs.
 Each PHICH is uniquely identified by a PHICH index. which indicates both the group and the
sequence.
 The number of UEs receiving their acknowledgements on the same set of downlink REs can
be up to twice the sequence length. The sequence length is four for normal prefix or two for
extended cyclic prefix.
 Each transport block is independently acknowledged using PHICH. The PHICH index for the
second codeword is related to the first by an offset.
REGs required for one PHICH
 ACK and NACK is encoded by 3 bits (111 for ACK, 000 for NACK).
 According to Table 6.9.1-2 of 36.211, each bit of PHICH is spread by 4 bits (SF=4)
when we use 'normal cyclic prefix'. So each PHICH after spreading with a 4 bits
orthogonal sequence becomes 12 bits.
 PHICH is modulated in BPSK and this means 'one symbol carries one bit'. And this in
turn means we need 12 symbols for each PHICH (each ACK or NACK).
 Each RE (Resource Elements) carries one symbol. So we need 12 REs to carry one
PHICH (one ACK or NACK).
 One REG is made up of 4 REs. So we need 3 REGs to carry one PHICH.
 These three REGs for one PHICH is distributed evenly across the whole bandwidth.

3 REGs required for a PHICH
 PHICH Signal Construction
Each of the three instances of the orthogonal code of a PHICH is mapped to an REG on
one of the first three OFDM symbols of each subframe.
Four Symbols are mapped to each REG.

PHICH Group
 One PHICH is transmitted per received transport block and TTI – this is when uplink
spatial multiplexing is used, two PHICH are used to acknowledge each transport block
 The error rate of PHICH should be low for proper operation of HARQ. Typically error
rates are 10-2 and 10-4 for ACK and NACK respectively . Since NACK to ACK error would
imply loss of transport block at MAC level which has to be recovered through RLC
retransmissions and associated delays while ACK to NACK error only implies an
unnecessary retransmission of an already decoded transport block
 A set of PHICHs transmitted on the same set of resource elements is called a PHICH
group. A PHICH group consists of eight PHICH in case of Normal prefix and four PHICHs
for extended cyclic prefix.

PHICH Location
 PHICH is carried by the first symbol of each subframe. (It is located in the same
symbol as PCFICH).
 One PHICH is carried by multiple REG.
 Multiple PHICH can be carried by the same set of REG and these multiple PHICH
being carried by the same REGs are called PHICH group. These multiple PHICHs are
multiplexed by orthogonal codes.
 Therefore, to identify a specific PHICH we need to know PHICH group number and
orthogonal code index.
 The UE figures out these two numbers from the lowest PRB index of the first slot of
the PUSCH transmission and DMRS cyclic shift.

Multiplexing of PHICH
 Multiple PHICH can be mapped to a same set of resource elements and this group of
PHICH being carried by the same set of resource element is called PHICH Group.
 One of the most common way of multiplexing in wireless communication would be

to use "orthogonal sequences".
 PHICH multiplexing also uses the same method, meaning they are multiplexed with a
set of predefined orthogonal sequences. The set of orthogonal sequence defined in
3GPP 36.211("6.9 Physical hybrid ARQ indicator channel").

Multiplexing of PHICH
 Multiple PHICH data can be located in the same physical locations (in the same REs).
This multiple PHICH that are multiplexed in the same location is called 'PHICH group'.
 The number of PHICH can belong to a single PHICH group is determined by a parameter
from higher layer (phich-Resource IE in MIB) and also determined by whether it is FDD
or TDD. In case of TDD, this varies depending on subframe configuration.

PHICH Groups
 PHICHs can be carried by one PHICH group : Maximum 8 PHICHs can be multiplexed
into a PHICH group when we use normal CP and Maximum 4 PHICHs can be
multiplexed into a PHICH when we use the extended CP. Zero PHICH in a PHICH group
is also allowed. When multiple PHICHs get multiplexed, they use a certain length of
orthogonal code. In case of normal CP, the length of the orthogonal code is 4 and in
case of extended CP, the length is 2.
 PHICH groups can be supported by a system bandwidth: This can be determined by

the system bandwidth (N_RB) and a special parameter called Ng. These N_RB and Ng
value is carried by MIB .
 With Ng and the N_DL_RB (maximum number of RB for a system bandwidth), you can
calculate the N_group_PHICH as in the following table.

Calculation of PHICH Groups
 In FDD, PHICH group can be calculation as follows :
 In TDD, PHICH group can be calculation as follows :
"MIB (Master Information Block : has an IE

(information element) called "phich-Resource" as
shown below.
This IE represents Ng

PCHICH Groups Vs System Bandwidth
 With Ng and the N_DL_RB (maximum number of RB for a system bandwidth), we can
calculate the N_group_PHICH as in the following table.
For 10 MHz, with Ng as 1, we can have 7 PHICH groups which means 84 REs
since each PHICH group requires 12 Resource Elements
For 20 MHz with Ng as 1, we need 156 REs.

Example PHICH resource allocation
 Example of PHICH resource allocation (PHICH Group, Orthogonal Sequence Index) for a 3
MHz and 2 PHICH Groups. There are eight sequences within each group so this provides
16 combinations of group and sequence – sufficient to support 15 acknowledgements for
15 connections when each connection is allocated a single resource Block

PHICH Duration
 The Master Information Block on the PBCH indicates whether the PHICH uses a
normal or extended duration. A normal duration means that the PHICH uses the first
OFDMA symbol belonging to a subframe . An extended duration means that the
PHICH uses the first three symbols belonging to a subframe
 The PHICH duration puts a lower limit on the value signaled by the PCFICH. If the
PHICH occupies the first 3 OFDMA symbols belonging to a subframe, then the
PCFICH must signal a value of at least 3

PHICH Assignment
 For PUSCH transmissions scheduled from serving cell c in subframe n, a UE shall
determine the corresponding PHICH resource of serving cell in subframe ,
where is always 4 for FDD and is given in table 9.1.2-1 for TDD.
For example, for TDD configuration 2, an

uplink transmission on PUSCH received in
subframe 2 should be acknowledged on
PHICH in subframe 2+6=8.

PHICH Assignment
 nDMRS is mapped from the cyclic shift for DMRS field (according to Table 9.1.2-2) in the most
recent PDCCH with uplink DCI format for the transport block(s) associated with the
corresponding PUSCH transmission. shall be set to zero, if there is no PDCCH with uplink DCI
format for the same transport block, and
— if the initial PUSCH for the same transport block is semi-persistently scheduled, or
— if the initial PUSCH for the same transport block is scheduled by the random access response grant
.

PHICH Assignment

PDCCH Capacity Dimensioning
 Let us take "PDCCH Dimensioning (10 MHz)" channel bandwidth as an example. We
assume 1 or 2 or 3 OFDM symbols are allocated for PDCCH.
 If the Control Format Indicator (CFI) is 1, only the first OFDM symbol in a subframe is
used for the PDCCH.
 There are a total of 50 PRBs in a 10MHz channel.
 Each PRB has 12 subcarriers and has 12 Resource Elements (REs) in the first OFDM
symbol. So 50 PRBs have 50*12=600 REs in the first OFDM symbol.
 Each PRB has 4 REs for Reference Signals (RS), so 50 PRBs have 50*4 =200 REs
reserved for RS, assuming 2 transmit antennas at the eNB.
 A total of 16 REs is reserved for PCFICH, regardless of system bandwidth.
 The PHICH occupies 84 Res, assuming Ng=1.
 The number of REs available for the PDCCH = Total REs - RS REs - PCFICH REs - PHICH
REs = 600 - 200 - 16 - 84 = 300 REs
 Each CCE consists of 36 REs, so the number of CCEs is floor(300 / 36) = 8 CCEs

PDCCH Capacity Dimensioning

HARQ Hybrid Automatic Repeat Request
 The HARQ in LTE is based on stop and wait HARQ
 Once down link packet is sent from eNode B, the UE will decode it and provide feedback
in the PUCCH. For negative acknowledgement (NACK) based on CRC check, the eNode B
will send a retransmission
 Eight HARQ are defined for Down Link and Up Link with minimum delay of 7 ms. The UE
will send the ACK/NACK for packet in frame n, in the uplink frame n+4. This leaves 3 ms
processing time for the UE depending on Timing Offset controlled by Timing Advance
Procedure.
Propagation delays are negligible

as compared to processing delays

Chase Combining and Incremental Redundancy
 Hybrid ARQ refers to a retransmission protocol in which the receiver checks for errors in
the received data and if an error is detected then the receiver buffers the data and
requests a retransmission from the sender.
 A HARQ receiver is then able to combine the buffered data with the re-transmitted data
prior to channel decoding and error detection. This improves the performance of the
retransmission
 HARQ retransmissions can benefit from either Chase Combining or Incremental
Redundancy
 Chase Combining means the physical layer applies the same puncturing pattern to both
the original transmission and each retransmission. This results in retransmissions which
include the same set of physical layer bits as the original transmission as shown below.
The benefits of chase combining are its simplicity and lower UE memory requirements
The RV does not
change between
transmissions of the
same transport block if
Chase Combining is
used. The RV changes
between transmissions
of the same transport
block if Incremental
Redundancy is used
Chase Combining
HARQ Process Number
 HARQ Process Number specifies the Hybrid Automatic Repeat Request (HARQ) process for
the transport block being sent within the allocated Resource Blocks
 The set of 3 bits for FDD provide support for a maximum of 8 HARQ Processes
 The set of 4 bits for TDD provide support for a maximum of 16 HARQ process
 In practice, the number of Downlink HARQ processes for TDD depends upon the uplink –
downlink configuration
 Uplink – downlink configuration 5 has the greatest requirement and uses a maximum of
15 HARQ processes
Data Transfer using multiple parallel SAW Processes (Stop and Wait)

 Turbo Coding generates systematic, parity 1 and parity 2 bits. The systematic bits are the
same as the original bit sequence. These bits are the most important to the receiver and are
provided with the greatest priority i.e. the parity 1 and parity 2 bits are punctured in
preference to the systematic bits
 A receiver using chase combining benefits from a soft combining gain because the received
symbols are combined prior to demodulation. This improves the received signal to noise
ratio if it is assumed that the signal power is correlated while the nose power is
uncorrelated.
 Incremental redundancy means the physical layer applies different puncturing patterns to
the original and retransmission. This results in retransmission which include a different set
of physical layer bits to the original transmission. The drawback associated with
incremental redundancy are its increased complexity and increased UE memory
requirements.
Incremental Redundancy

 The first transmission provides the systematic bits with the greatest priority while
subsequent retransmission can provide either the systematic or the parity 1 and parity 2
bits with the greatest priority.
 The performance of incremental redundancy is similar to the performance of chase

combining when the coding rate is low i.e. there is little puncturing. The performance
of incremental redundancy becomes greater when there is a increased quantity of
puncturing (code rate is high).
 Incremental redundancy performs better than chase combining when the coding rate
is high because channel gain is greater than soft combining gain.
The sender buffers the transmitted data until a positive acknowledgement has been
received in case a retransmission is required .
Data is cleared from the transmit buffer once a positive acknowledgement has been
received or the maximum number of allowed retransmissions has been reached .
New data can be sent by a specific HARQ process once its transmit buffer has been cleared
Switching Points TDD

TDD HARQ
 Transmission of downlink or uplink data with HARQ requires that an ACKnowledgement
or Negative ACKnowledgement ACK/NACK be sent in the opposite direction to inform the
transmitting side of the success or failure of the packet reception
 In case of FDD, acknowledgement indicators related to data transmission in a subframe k
are transmitted in the opposite direction during subframe k+4 such that a one-to-one
mapping exists between the instant at which a transport block is transmitted and its
corresponding acknowledgement.
 However in case of TDD, subframes are designed on a cell specific basis as uplink or
down link or mixed (special subframe). Thus the scheme used for FDD cannot be directly
used for TDD
 Since TDD is asymmetric for downlink-uplink, it is inevitable that one of the link
directions there are insufficient subframe to support one-to-one mapping between
transmitted data and its acknowledgment. The design for TDD supports grouped
ACK/NACK to carry multiple acknowledgments within one subframe

TDD HARQ
 In TDD-LTE, the HARQ process is different from that in the FDD implementation of
LTE.
 In the case of FDD, for a transmission on subframe #n, an ACK/NACK message is
sent on subframe # n+4.
 The reason for the 4 subframe delay in the transmission of a ACK/NACK message is
due to the processing delay of about 3 ms at the receiver.
 If it is a NACK, the retransmission is scheduled on subframe # n+8 for UL
transmissions while the DL retransmission can be asynchronous.
 In TDD-LTE, the time association between the data transmission and the ACK/NACK
cannot be maintained due to the variable numbers of DL and UL subframes being
present in a frame.
 The UL and DL delay between data and ACK is dependent on the TDD configuration
chosen.
 Hence, a fixed delay between a transmission and the HARQ ACK/NACK is not
possible in TDD-LTE.

TDD HARQ
 Similar to FDD, HARQ for TDD is synchronous in the uplink direction.
 The fixed time intervals between transmissions and retransmissions depend on the
uplink-downlink configuration – Synchronous HARQ does not require HARQ process
identity to be signaled thus saving on signaling overheads

TDD Uplink
 HARQ for the uplink refers to transferring uplink data on the PUSCH while receiving
downlink acknowledgements on the PHICH
 Similar to FDD, HARQ for TDD is synchronous in the uplink direction. This means that
acknowledgements and retransmissions are scheduled at fixed time intervals relative to the
initial transmission. In the case of TDD, those fixed time intervals depend upon the uplink-
downlink subframe configuration. Synchronous HARQ does not require the HARQ process
identity to be signalled so generates a lower overhead
 There is one HARQ entity at the UE for each serving cell. There are multiple serving cells
and multiple HARQ entities when LTE Advanced Carrier Aggregation is used
 Each HARQ entity maintains a number of synchronous parallel HARQ processes. The total
number of HARQ processes as a function of the TDD uplink-downlink subframe
configuration is summarized in Table below. TTI bundling is not used when Transmission
Mode 2 is configured, i.e. it is not used with uplink MIMO
Total number of
HARQ
processes for
the uplink of
TDD

TDD Uplink
 In the case of uplink-downlink subframe configurations 1 to 5, when using normal
operation with Transmission Mode 1, the number of HARQ processes is equal to the
number of uplink subframes within each radio frame. In the case of uplink-downlink
subframe configurations 0 and 6 an additional HARQ process is required
 The HARQ process timing for uplink-downlink subframe configuration 1 is illustrated
in Figure below. The delay between the UE sending some uplink data on the PUSCH
and receiving an acknowledgement on the PHICH varies depending upon the
subframe being used. This delay is shown using the variable 'k' in Figure . The value of
'k' never decreases below 4
Timing of HARQ
processes for
TDD uplink-
downlink
subframe
configuration 1
(normal
operation,
transmission
mode 1)

TDD Uplink
 The pattern is less regular than FDD because acknowledgements cannot always be sent after
4 subframes, i.e. the 4th subframe can be an uplink subframe so the eNode B has to wait
until the next available downlink subframe before being able to send the acknowledgement.
HARQ acknowledgements can be sent during the DwPTS of the special subframes.
 Similarly, the delay between receiving an acknowledgement on the PHICH and being able to
send either new data or a retransmission using the same HARQ process is not constant
because the 4th subframe after an acknowledgement can be a downlink subframe so the UE
has to wait until the next available uplink subframe
 Figure below illustrates the HARQ process timing for uplink-downlink subframe
configuration 6 when assuming only 5 HARQ processes are used. This is not a real
configuration but illustrates the reason for requiring 6 HARQ processes
Timing of HARQ processes for

TDD uplink-downlink
subframe configuration 6 if
assuming only 5 HARQ
processes
(note this is not a real
configuration- real
configuration uses 6 HARQ
processes)
TDD Uplink
 Within Figure in previous slide, HARQ processes 1 to 4 appear normal but HARQ process 5
has only 3 subframes difference between the UE receiving its acknowledgement and having
to send the next transmission. A minimum of 4 subframes is required to allow sufficient
processing time so this triggers the requirement for an additional HARQ process
 Figure below illustrates the real HARQ process timing for uplink-downlink subframe
configuration 6. This demonstrates that an additional HARQ process allows a minimum of 4
subframes difference between receiving an acknowledgement and sending the next block
of data
Timing of HARQ
processes for TDD
uplink-downlink
subframe
configuration 6 (real
configuration)

TDD Uplink
 The value of the 'k' variable for each uplink-downlink subframe configuration is
presented in Table below. This table shows that when an acknowledgement is
received on the PHICH within the nth subframe, then it refers to the PUSCH data
sent during subframe n – k
 The uplink-downlink subframe configuration 0 represents a unique case because two
sets of acknowledgements are received in subframes 0 and 5, i.e. acknowledgements
are received in these subframes which refer to PUSCH transmissions 7 subframes
earlier, and 6 subframes earlier. This approach is necessary because there are 6 uplink
subframes available for the PUSCH but only 4 downlink subframes available for the
PHICH
'k' variable for

each uplink-
downlink
subframe
configuration

TDD Uplink
 Figure below illustrates the HARQ process timing for uplink-downlink subframe
configuration 0. The 2 PHICH acknowledgements sent during subframe 0 (and also
subframe 5) are differentiated by their PHICH group. 3GPP TS 36.213 specifies which
PHICH group is allocated to each acknowledgement. This allows the UE to link each
acknowledgement to the appropriate PUSCH transmission
Timing of HARQ
processes for
TDD uplink-
downlink
subframe
configuration 0

TDD Uplink
 Table below presents the number of HARQ acknowledgements per subframe on the PHICH
for each uplink-downlink subframe configuration. Within the context of this table, special
subframes are counted as downlink subframes. Multiple HARQ acknowledgements can be
multiplexed within the same subframe by allocating different PHICH groups and
orthogonal sequences
 Without spatial multiplexing, the PHICH has to transfer a maximum of 2 HARQ
acknowledgements per subframe per connection. The majority of uplink-downlink subframe
configurations only require a single HARQ acknowledgement per subframe because there
are more downlink subframes than uplink subframes
 The requirement for acknowledgements doubles when spatial multiplexing is used and 2
transport blocks are transferred during each subframe. The pair of acknowledgements
associated with the pair of transport blocks are transferred during the same subframe but
using different PHICH resources (combination of PHICH group and orthogonal sequence)
The number
of HARQ
acknowledge
ments per
subframe on
the PHICH

TDD Downlink
receiving uplink acknowledgements on the PUCCH or PUSCH
 There is one HARQ entity at the UE for each serving cell. There are multiple serving
cells and multiple HARQ entities when LTE Advanced Carrier Aggregation is used
 Each HARQ entity maintains a number of asynchronous parallel HARQ processes. The
total number of HARQ processes is summarized in Table in next slide. TTI bundling is
not applicable to the downlink because it is aimed at improving uplink coverage
 Similar to the downlink of FDD, Table in next slide illustrates 2 main differences
relative to the uplink
 in the case of the downlink, there is no dependency upon whether or not
spatial multiplexing is used. In the downlink direction, when 2 transport blocks
are transferred during the same subframe using spatial multiplexing, then those
2 transport blocks are associated with the same HARQ process rather than 2
different HARQ processes
 in the case of the downlink, the number of HARQ processes is 'up to' x and 'up to'
y, rather than exactly x or exactly y. This results from HARQ being asynchronous
in the downlink direction so fewer HARQ processes can be used when a
connection is not scheduled data during every subframe

TDD Downlink
 Similar to the downlink of FDD, HARQ for TDD is asynchronous in the downlink direction.
This means that the eNode B provides the UE with instructions regarding which HARQ
process to use during each subframe that resources are allocated. The HARQ process
identity is included within PDCCH DCI formats 1, 1A, 1B, 1D, 2, 2A, 2B and 2C
 Asynchronous HARQ increases the signalling overhead as a result of the requirement to
include the HARQ process identity within the DCI. However, asynchronous HARQ also
increases flexibility because retransmissions do not have to be scheduled during specific
subframes. The eNode B may wish to postpone a retransmission because other UE are
being scheduled during that subframe, or because the channel conditions have become
relatively poor
Total number of HARQ processes for the downlink of TDD

TDD Downlink
 Table below presents the number of HARQ acknowledgements per subframe which would
ideally be supported by the PUCCH (in practice, the PUCCH does not have this much
capacity). Within the context of this table, special subframes are counted as downlink
subframes
Requirement for HARQ acknowledgements

 Many of these figures exceed the capacity of the PUCCH. The number of
acknowledgements supported by the PUCCH. Channel selection can be used with PUCCH
format 1 b to allow up to 4 HARQ acknowledgements to be transferred during a single
subframe. However, uplink-downlink subframe configuration 5 requires 9
acknowledgements to be sent within a single subframe (assuming spatial multiplexing is
not used and downlink data is sent during every downlink subframe)
TDD Downlink

Grouping of ACK/NACK
 Two mechanisms are provided for grouping the acknowledgement information
carried in the uplink in TDD operation termed as ‘ACK/NACK bundling ‘ and
‘ACK/NACK multiplexing’
Grouping of ACK/NACK information in TDD

ACK/NACK Bundling & Multiplexing in TDD
 ACK/NACK bundling reuses the same 1 and 2 bit Physical Uplink Control Channel PUCCH
formats 1a and 1b which are used for FDD. For each downlink codeword (up to two if
downlink spatial multiplexing is used), only a single acknowledgement is derived by
performing logical ‘AND’ operation of the acknowledgments across group of downlink
subframes associated with that uplink subframe; this indicates whether zero or more than
zero transport blocks in the group were received in error
 For ACK/NACK multiplexing a separate acknowledgement is returned for each associated

downlink subframes. However to limit the amount of signaling information,
acknowledgement from multiple code words on different spatial layers within a
subframe are first condensed into a single acknowledgment again by means of logical
AND operation (known as ‘spatial ACK/NACK bundling’)
 For the more extreme asymmetries, however, there remains a need to transmit more than
two bits of ACK/NACK information in one uplink subframe. This is achieved using the
normal 1- and 2-bit PUCCH formats augmented with a code selection scheme whereby
the PUCCH code selected by the UE conveys the surplus information to the eNodeB

 A disadvantage of these lossy compression schemes for grouped acknowledgements is
that the eNodeB does not know exactly which transport block(s) failed in decoding. In
the event of a NACK, all transport blocks in the same group must be resent, increasing
retransmission overheads and reducing link throughput. A more subtle impact is that
the average HARQ round trip time (and hence latency) can be increased due to the fact
that some blocks cannot be acknowledged until the remainder of the group have been
received.
 A further complication arises because the PDCCH control signalling is not 100% reliable
and there is some possibility that the UE will miss some downlink resource
assignments.
 This would introduce the possibility of HARQ protocol errors, including the erroneous
transmission of ACK in the case when one or more downlink assignments were missed
in a group of subframes. In order to help avoid this problem, a ‘Downlink Assignment
Index’ (DAI) is included in the PDCCH to communicate to the UE the number of
subframes in a group that actually contain a downlink transmission. In the case of
ACK/NACK bundling, this helps the UE to detect missed downlink assignments and
avoid returning ACK if one or more downlink assignments were missed, while in the
case of ACK/NACK multiplexing the DAI helps the UE to determine how many bits of
ACK/NACK information should be returned.
 Nevertheless, this mechanism alone cannot safeguard against all possible error cases.
 In Release 10, where even larger numbers of ACK/NACK bits may need to be transmitted in a
single subframe due to carrier aggregation, new PUCCH mechanisms are provided.
 The presence of link asymmetry in TDD generally increases the overall HARQ round trip
times (compared to the 8 ms of FDD) as a result of the additional ‘waiting times’ needed for
the appropriate link direction to become available for transmission of an acknowledgement,
grant or retransmission.
 This, together with the degree of asymmetry, also means that the number of HARQ
processes in TDD varies depending on the TDD uplink-downlink subframe configuration: in
the downlink, the number of HARQ processes varies from 4 for configuration 0 to 15 for
configuration 5, while in the uplink it varies from 1 for configuration 5 to 7 for
configuration 0

 The signalling of the granted uplink transmission resources also requires some
specific attention for TDD.
 For FDD, the downlink subframe in which an uplink grant is sent implicitly also signals
the specific uplink subframe that has been assigned (located four subframes later),
whereas for TDD this relationship cannot always hold due to the various uplink-
downlink configurations.
 An alternative linkage is therefore formulated for each specific uplink-downlink
configuration to associate each uplink subframe with one preceding downlink
subframe that controls it (maintaining the same four-subframe spacing as FDD
wherever possible).
 For configuration 0, there are more uplink subframes per 10 ms than the number of
subframes available for PDCCH, and here an additional ‘UL Index’ field in the uplink
grant message is used to indicate the uplink subframe to which the grant relates. In
addition, one value of the ‘UL Index’ field can be used to enable a single PDCCH
message to grant uplink resources in two uplink subframes

 3GPP TS 36.213 specifies 2 HARQ feedback modes for TDD which reduce the number
of acknowledgements to be transferred:
 HARQ-ACK bundling
 HARQ-ACK multiplexing
 All uplink-downlink subframe configurations support both feedback modes, with the
exception of uplink-downlink subframe configuration 5 which only supports HARQ-
ACK bundling. HARQ multiplexing is not supported for configuration 5 because there
is only a single uplink subframe to support the transfer of acknowledgements for 9
downlink subframes (including the DwPTS field within the special subframe ). This
means that HARQ-ACK bundling is required to reduce the number of
acknowledgements
 A UE can be instructed to use either HARQ-ACK bundling or HARQ-ACK multiplexing
within an RRC Connection Setup, RRC Connection Reconfiguration or RRC
Connection Re-establishment message. The instruction is provided as part of the
PUCCH configuration information
 HARQ-ACK bundling reduces the number of HARQ acknowledgements by performing
a logical AND operation between the acknowledgements belonging to multiple
downlink subframes, i.e. bundling is completed by applying a logical AND operation in
the time domain

ACK/NACK Bundling in TDD
 All uplink – down link configurations support both bundling and multiplexing with
exception of configuration 5 which only supports HARQ- ACK bundling (since there is a
single uplink subframe against nine downlink subframes with DwPTS)
 A UE can be configured to use bundling or multiplexing within an RRC connection setup or
RRC connection reconfiguration
 Example below shows for configuration 2, the downlink subframes are divided into two
groups and a single acknowledgement generated from each group. In case of spatial
multiplexing two acknowledgements are generated which can be transferred using
format 1b on PUCCH
HARQ Bundling in Configuration 2

 The example illustrated in Figure in previous slide assumes that spatial multiplexing
is not used. This results in a single HARQ acknowledgement for each uplink subframe.
 When spatial multiplexing is used, the logical AND operations are completed
independently for the two transport blocks, i.e. there is one logical AND operation for
the set of transport block 1 's, and a second logical AND operation for the set of
transport block 2's. This results in 2 HARQ acknowledgements for each uplink
subframe
 When spatial multiplexing is not used, and a single HARQ acknowledgement is
generated, that HARQ acknowledgement can be transferred using PUCCH format 1a.
When spatial multiplexing is used, and 2 HARQ acknowledgements are generated,
those HARQ acknowledgements can be transferred using PUCCH format 1 b
 Receiving a single transport block in error results in the UE returning a negative
acknowledgement for the complete 'bundle' of downlink transport blocks. This
leads to the eNode B re-transmitting all transport blocks within the bundle
 The benefit of HARQ-ACK bundling is a reduced capacity requirement for the PUCCH

 Table below presents the number of downlink transmissions within each 'bundle' for
each uplink-downlink subframe configuration.
 Uplink-downlink configurations 0 and 6 have sufficient uplink subframes to avoid the
requirement for bundling
The number of downlink transmissions within each 'bundle' for HARQ-ACK bundling

 The 1 or 2 'bundled' acknowledgements transferred during subframe 'n' are
generated from the downlink data received during subframes 'n- k', where k is
specified in Table
'k' variable for each uplink-downlink subframe configuration

ACK/NACK Multiplexing in TDD
 HARQ multiplexing reduces the number of HARQ acknowledgements by performing logical
AND operation between pair of transport blocks associated with spatial multiplexing
 HARQ bundling applies AND operations in the time domain and HARQ multiplexing applies
AND operation in the spatial domain
 Figure illustrates the concept HARQ-ACK multiplexing for uplink-downlink subframe
configuration 2. A total of 16 acknowledgements per radio frame is reduced to a total of 8
acknowledgements. These acknowledgements are transferred using 2 uplink subframes.
PUCCH format 1 b can be used to transfer 4 acknowledgements when channel selection is
used, i.e. the PUCCH payload transfers 2 acknowledgments and the selection of PUCCH
resources (1 resource out of a set of 4 resources) transfers a further 2 acknowledgements
HARQ Multiplexing in Configuration 2

ACK/NACK Multiplexing in TDD
 HARQ-ACK multiplexing is only applicable to uplink subframes which provide
acknowledgements for more than a single downlink subframe, i.e. entries within Table
earlier (The number of downlink transmissions within each 'bundle' for HARQ-ACK
bundling) which have values greater than 1.
 It is not necessary to apply HARQ multiplexing for uplink subframes which acknowledge
only a single downlink subframe because there is only 1 or 2 acknowledgements to
transfer and PUCCH formats 1 a and 1 b can accommodate these acknowledgements
without HARQ-ACK multiplexing
 Receiving a single transport block in error results in the UE returning a negative
acknowledgement for the pair of downlink transport blocks. This leads to the eNode B re-
transmitting both transport blocks
 The benefit of HARQ-ACK multiplexing is a reduced capacity requirement for the PUCCH
 3GPP References: TS 36.213, TS 36.321, TS 36.331

DAI Downlink Assignment Index
 In case PDCCH control signaling is not 100% reliable, there is a possibility that the UE
will miss some downlink resource assignments. This would introduce the possibility of
an erroneous ACK in case when one or more downlink assignments were missed in a
group of subframes.
 To avoid this problem, a ‘Downlink Assignment Index’ DAI is included in the PDCCH to
communicate to the UE the number of subframes in a group that actually contain a
downlink transmission.
 In case of ACK/NACK bundling this helps the UE to detect missed downlink
assignments and avoid returning ACK if one or more downlink assignments were
missed
 In case of ACK/NACK multiplexing the DAI helps the UE to determine how many bits
of ACK/NACK needs to be returned.

HARQ Processes
 The link asymmetry increases the overall HARQ RTT in TDD compared to 8 ms in FDD
 The Number of HARQ processes in TDD depends on the downlink/uplink
configuration.
 In downlink the number of HARQ processes varies from 4 for configuration 0 to 15
for configuration 5
 In uplink it varies from 1 for configuration 5 to 7 for configuration 0
 Data is removed from buffer when:
a. ACK is received
b. Max number of re-transmission has reached
New data can then be send by the same HARQ process once its transmit buffer is
empty.

HARQ Timing
 In case of FDD, it is pretty simple and obvious for UE to transmit HARQ ACK or NACK. UE
start preparing the response as soon as it completes the decoding PDSCH and transmit
it 4 ms (4 TTI) later.
 But in TDD, UE cannot transmit the response in such a fixed timing as in FDD. It has to
wait until it gets the next chance for UL transmission and the next chance will be
different depending on UL/DL configuration.
 Even when UE gets the chance to transmit the UL, it is may not always possible to
transmit all the necessary response.
 For example, if UE gets too many DL subframe before the UL subframe, it will be
difficult to transmit the all the reply in the UL transmission because PUCCH space is not
big enough to accommodate all the HARQ ACK/NACK.
 The HARQ is as per table in following slide
UMTS HSDPA uses 6 HARQ processes and a 2 ms TTI to allow for a 12 ms round trip time

TDD HARQ
No of Hybrid ARQ processes and Acknowledgement Timing k for different TDD Configurations
In TDD the acknowledgement of a transport block in subframe n is transmitted in subframe

n+k, where k>=4 and is selected such that n+k is an uplink subframe when
acknowledgement is transmitted from the terminal on PUCCH or PUSCH and a downlink
subframe when acknowledgement is transmitted from the eNodeB on PHICH.
For example, for TDD configuration 2, an uplink transmission on PUSCH received in

subframe 2 should be acknowledged on PHICH in subframe 2+6=8. Similarly downlink
transmission on PDSCH in subframe 0 should be acknowledged on PUCCH (or PUSCH) in
subframe 0+7=7
More HARQ processes required for Configuration with lesser uplink subframes
Downlink Association Index
Indicates
the
subframe
in which
UE
transmits
the
acknowle
dgement
K Indicates the
subframe prior to this
subframe (n-k) for
which
acknowledgement is
being transmitted
Case 3 ACK/NACK from UE for PDSCH
 Case 3 : UL/DL Configuration 2
 In case of UL/DL Configuration 2, Ack/Nack response timing for the PDSCH that is
received by UE is transmitted according to the following rule.



ACK/NACK from eNB for PUSCH
Indicates the
subframe in which
UE transmits
PUSCH
Indicates the
subframes
(n+kPHICH) after
which the eNodeB
transmits the
acknowledgement

Case 3 : ACK/NACK from eNB for PUSCH
UL/DL Config 2:
UE shall transmit PUSCH in subframe 2,7
eNB shall send ACK/NACK in subframe (2+6=8),(7+6=3)

Case 1: ACK/NACK from eNB for PUSCH
UE shall transmit PUSCH in subframe 2,3,4,7,8,9

eNB shall send ACK/NACK in subframe (2+4=6),(3+7=0),(4+6=0),(7+4=1),(8+7=5),(9+6=5)

TDD Configuration 0 - HARQ Timing

Case 2 : ACK/NACK from eNB for PUSCH
UE shall transmit PUSCH in subframe 2,3,7,8

eNB shall send ACK/NACK in subframe (2+4=6),(3+6=9),(7+4=1),(8+6=4)

Troubleshooting Example
 For example : DL/UL Configuration 2 and UE sent a NACK at Subframe 2.
 How did you know whether the NACK is for PDSCH at subframe 4 or 5 or 6 or 8 ?
 In FDD.. the answer is simple since the ACK/NACK from the UE is always for the
PDSCH that it received 4 subframe before. If it is FDD, the answer is supposed to be 'it
is for PDSCH received at subframe 8 in previous SFN), but in TDD case it is different
 We can use the following procedure.

 First, check UCI info at specific SFN and subframe number (let's label this as
'SFN_n:Subframe_2') and locate the HARQ process number that caused NACK.
 Go to transmitted PDSCH list 'around' SFN_n:Subframe_2
 Look through several subframes upwards and downwards to find the subframe that is
marking the same HARQ process number as at first step above. That is the subframe that
caused NACK.

Ack/Nack Feedback Mode
 in TDD LTE ibe subframe can transmit ACK/NACK for multiple subframe as shown below.
In the following figure as an example, UE send ACK/NACK for 4 PDSCHs in subframe 2.
What should eNB do if the subframe 2 send NACK ? Does it have to retransmit the whole
4 PDSCHs ? or transmit only PDSCH which is NACKed ?
Depending on tdd-
AckNackFeedbackMode
setting in RRC message
(e.g, RRC Connection
Setup or RRC Connection
Reconfiguration) :
If it is set to be 'bundling',
eNB should retransmit all
the PDSCH. If it is sent to
be 'multiplexing', eNB
should retransmit the
only PDSCH which is
NACKed.

HARQ Parameters
Parameter Description Range Value
maxHARQ- Maximum number of UL HARQ transmissions. Enum{n1, n5
Tx n2, n3, n4,
Corresponds to parameter maxHARQ-Tx n5, n6, n7,
specified in 3GPP TS 36.331 section 6.3.2 and in n8,n10,
3GPP TS 36.321 section 5.4.2.2 n12, n16,
n20, n24,
Higher value provides delay in RLF n28,
Trade off end-to-end delay vs. capacity. spare2,
spare1}
For volte, strongly recommend it to n4, for BE, it
can be relaxed.
Lower value enables faster detection of an RLF if

it exists.

HARQ Parameters
 HARQ information: HARQ information consists of New Data Indicator (NDI), Transport
Block (TB) size. For DL-SCH transmissions the HARQ information also includes HARQ
process ID. For UL-SCH transmission the HARQ info also includes Redundancy Version
(RV). In case of spatial multiplexing on DL-SCH the HARQ information comprises a set
of NDI and TB size for each transport block.
 HARQ RTT Timer: This parameter specifies the minimum amount of subframe(s)
before a DL HARQ retransmission is expected by the UE. For FDD the HARQ RTT Timer
is set to 8 subframes. For TDD the HARQ RTT Timer is set to k + 4 subframes, where k
is the interval between the downlink transmission and the transmission of associated
HARQ feedback
 The UE is configured with a Maximum number of HARQ transmissions and a
Maximum number of Msg3 HARQ transmissions by RRC: maxHARQ-Tx and
maxHARQ-Msg3Tx respectively. For transmissions on all HARQ processes and all
logical channels except for transmission of a MAC PDU stored in the Msg3 buffer, the
maximum number of transmissions shall be set to maxHARQ-Tx. For transmission of a
MAC PDU stored in the Msg3 buffer, the maximum number of transmissions shall be
set to maxHARQ-Msg3Tx.

HARQ Entity
 There is one HARQ entity at the UE for each Serving Cell which maintains a number
of parallel HARQ processes.
 Each HARQ process is associated with a HARQ process identifier.
 The HARQ entity directs HARQ information and associated TBs received on the DL-
SCH to the corresponding HARQ processes
 When the physical layer is configured for downlink spatial multiplexing , one or two
TBs are expected per subframe and they are associated with the same HARQ
process. Otherwise, one TB is expected per subframe.
 The UE shall:
 If a downlink assignment has been indicated for this TTI:
 allocate the TB(s) received from the physical layer and the associated HARQ
information to the HARQ process indicated by the associated HARQ information.
 If a downlink assignment has been indicated for the broadcast HARQ process:
 allocate the received TB to the broadcast HARQ process.

REPITITION of HARQ in Uplink
 The eNodeB can use RRC connection setup, RRC connection reconfiguration or RRC
connection re-establishment message to instruct the UE to repeat its HARQ
acknowledgements sent in the up link direction
 This is intended to improve the reliability of signaling for UE in relatively weak
coverage
 The eNodeB can request 1,3,or 5 repetitions of the HARQ acknowledgements after
the original transmission
 Repetition is applicable for both FDD and TDD but is only applicable in TDD when
using HARQ-ACK Bundling

Misc Points
 for TDD configuration 2, for an uplink transmission in subframe 'n', the ACK/NACK is
sent in down link in subframe 'n+6' (and not n+4). Thus using DSUDDDSUDD
(subframe 0 to 9) for uplink transmission in subframe 2 the ACK/NACK would come in
down link in subframe 2+6=8
 for Spatial Multiplexing (MIMO) in FDD uplink each code word has a separate
ACK/NACK - thus we may have 16 HARQ process
 for FDD Downlink with spatial multiplexing(MIMO), same HARQ is used for both the
code words - thus we may have 8 HARQ process (not 16 as in uplink)
 for TDD using bundling, one ACK/NACK generated for each codeword - thus with
spatial multiplexing (MIMO) two bits generated for the two code words bundled for
four down link subframes each
 for TDD using multiplexing with spatial multiplexing (MIMO), AND Operation is done
for the two code words - thus for four down link subframes with two code words we
have four bits of ACK/NACK
 HARQ RTT for FDD is 8 ms and HARQ RTT for TDD Configuration 2 is 12 ms.

3GPP 36.213 10.1.3
 For TDD UL/DL configuration 5 and a UE that does not support aggregating more than one
serving cell, only HARQ-ACK bundling is supported.
 If a UE that supports Carrier Aggregation is configured by higher layers to use HARQ-ACK
bundling, PUCCH format 1b with channel selection according to the set of Tables 10.1.3-
2/3/4 or according to the set of Tables 10.1.3-5/6/7, or PUCCH format 3 for transmission of
HARQ-ACK shall be used.
 PUCCH format 1b with channel selection according to the set of Tables 10.1.3-2/3/4 or
according to the set of Tables 10.1.3-5/6/7 is not supported for TDD UL/DL configuration 5.
 TDD HARQ-ACK bundling is performed per codeword across M multiple DL subframes
associated with a single UL subframe n, by a logical AND operation of all the individual
PDSCH transmission (with and without corresponding PDCCH/EPDCCH) HARQ-ACKs and ACK
in response to PDCCH/EPDCCH indicating downlink SPS release
 For TDD HARQ-ACK multiplexing and a subframe n with M >1, spatial HARQ-ACK bundling
across multiple codewords within a DL subframe is performed by a logical AND operation of
all the corresponding individual HARQ-ACKs. PUCCH format 1b with channel selection is
used in case of one configured serving cell
 For TDD HARQ-ACK multiplexing and a subframe n with M = 1, spatial HARQ-ACK bundling
across multiple codewords within a DL subframe is not performed, 1 or 2 HARQ-ACK bits are
transmitted using PUCCH format 1a or PUCCH format 1b, respectively for one configured
serving cell.
Value of M
 M is defined to be "the number of elements in the set K defined in Table 10.1.3.1-1".

FDD and TDD Comparison

FDD HARQ
 In FDD, we are using 8 HARQ process.
 i) For Downlink
a) it can use the 8 HARQ processes in any order (Asynchronous Process).
b) UE does not know anything about HARQ process information for DL data
before it gets it. So Network send these information (Process ID, RV) in PDCCH
(DCI).
 ii) For Uplink
a)it have to use the specific process in a specific subframe (Synchronous
Process). UE has to use the same HARQ process number every 8 subframes.
b) Since UE have to use specific HARQ process ID at specific subframe, the
receiver (eNode B) knows exactly which HARQ process comes when. And
eNodeB can also knows about RV because UL Grant (DCI 0) from eNodeB can
specify RV using MCS field.
c)it has two mode of operation : Adaptive and Non-Adaptive HARQ

Adaptive HARQ Example - FDD
 Following is an example of Adaptive UL HARQ Process (Key idea is that Each UL
retransmission uses different RV and the RV is determined by DCI 0).

Non Adaptive HARQ Example FDD
 Following is an example of Non Adative UL HARQ Process (Key idea is that Each UL
retransmission uses different RV and the RV is determined by predefined sequence
specified in TS36.321 "5.4.2.2 HARQ process").

Misc
 The UE knows if it is supposed to do Adaptive retransmission and Non-Adaptive
retransmission by the following :
 UE would do "Adaptive retransmission" if it detect DCI 0 and NDI is not toggled.

(In this case, UE overrides the "HARQ feedback (PHICH)“ and it retransmit based
on DCI 0 information).
 UE figures out that it needs to use "Non-Adaptive retransmission" if it got "HARQ

feedback (PHICH=NACK)" but does not get DCI 0.

HARQ Process ID Synchronization
 When transferring data via HARQ process, the receiver and transmitter should know
'some information' about Process ID for each of the HARQ process, so that the
receiver can successfully keeping each process data without getting them mixed up.
 In case of Asynchronous HARQ (e.g, PDSCH transmission in LTE), the sender should
inform the receiver of HARQ processor number explicitly. In case of LTE, DCI 1 and 2
carries this information.
 In case of Synchronous HARQ it is not required to inform Process ID in this case since
the process ID can be inferred from the transmission time (In LTE UL HARQ case, this
timing is expressed in SFN and subframe number).
 HARQ process ID can be inferred from SFN and subframe number :
 In LTE, there is no specific formula is defined in the 3GPP specification, but
following can be one of the simplest rule in LTE case.
 UL HARQ Process ID = (SFN x 10 + subframe) modulo 8

Redundancy Versions

Copyright and Confidentiality
Copyright © 2015, SAMSUNG Electronics Co., Ltd. SAMSUNG Electronics reserves the right to make changes to the specifications of
the products detailed in this document at any time without notice and obligation to notify any person of such changes. Information
in this document is proprietary to SAMSUNG Electronics Co., Ltd. No information contained here may be copied, translated,
transcribed or duplicated by any form without the prior written consent of SAMSUNG Electronics.

20 Harq

Uploaded by

Document Informationclick to expand document informationharq

Document Informationclick to expand document information

Copyright:

Available Formats

20 Harq

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

20 Harq

Uploaded by

Copyright:

Available Formats

HARQ

2015 © Samsung Electronics 2

2015 © Samsung Electronics 3

2015 © Samsung Electronics 4

Multiple Hybrid ARQ Processes

Data transfer using a single Stop And Wait (SAW) process

2015 © Samsung Electronics 8

2015 © Samsung Electronics 9

2015 © Samsung Electronics 11

 An RLC entity is configured in one of the three data transmission modes –

2015 © Samsung Electronics 12

2015 © Samsung Electronics 13

RLC entity is responsible for

2015 © Samsung Electronics 14

 Unacknowledged Mode UM : supports segmentation/reassembly and in-sequence

 Acknowledged Mode AM : used for TCP/IP packets on DL-SCH.

2015 © Samsung Electronics 15

(a) In the above example, at t3 PDU up to n+5 have been transmitted.

2015 © Samsung Electronics 18

Asynchronous Adaptive HARQ for Downlink

2015 © Samsung Electronics 21

Downlink SAW HARQ Protocol Timing

Uplink SAW HARQ Protocol Timing diagram

2015 © Samsung Electronics 24

MIMO is not used for TTI Bundling

Requirement for acknowledgements doubles when spatial multiplexing is used

2015 © Samsung Electronics 27

 Retransmissions can be either non-adaptive or adaptive:

Retransmission triggering using the PHICH and PDCCH

Total number of HARQ processes for the downlink of FDD

2015 © Samsung Electronics 32

2015 © Samsung Electronics 33

2015 © Samsung Electronics 34

2015 © Samsung Electronics 37

2015 © Samsung Electronics 38

 One REG is made up of 4 REs. So we need 3 REGs to carry one PHICH.

2015 © Samsung Electronics 40

2015 © Samsung Electronics 41

2015 © Samsung Electronics 42

 One PHICH is carried by multiple REG.

2015 © Samsung Electronics 43

 One of the most common way of multiplexing in wireless communication would be

2015 © Samsung Electronics 44

2015 © Samsung Electronics 45

 PHICH groups can be supported by a system bandwidth: This can be determined by

2015 © Samsung Electronics 46

 In TDD, PHICH group can be calculation as follows :

"MIB (Master Information Block : has an IE

2015 © Samsung Electronics 47

2015 © Samsung Electronics 48

2015 © Samsung Electronics 49

2015 © Samsung Electronics 50

For example, for TDD configuration 2, an

2015 © Samsung Electronics 51

2015 © Samsung Electronics 52

2015 © Samsung Electronics 53

2015 © Samsung Electronics 54