LTE-HARQ协议解读

NDI toggled，就代表NDI翻转（0变为1，或者 1变为0），NDI翻转则表示这次要是一次新的传输，而与NDI每次具体的值无关，只看其值翻转与否1. PDCCH和PDSCH在一个时隙中占用不同的OFDM symbol，前面1~3个OFDM symbol是控制域，用于存放PBCH、PHICH、PCFICH和PDCCH，后面的是PDSCH；2. 接收端确定PDCCH位置的过程为：首先接收位置固定的PBCH，获得PHICH的格式和占用情况；接着收取位置固定的PCFICH，获得PDCCH在一个子帧中的 OFDM 数目；最后在控制域内剔除PCFICH和PHICH，获得PDCCH的起点；知道了起点和OFDM symbol数目，PDCCH就可以正确接收了，而PDSCH属于数据域，位置可以认为是固定的。

上报方法要在上行控制信息UCI里面找。

ACK/NACK反馈模式要在PUCCH的反馈过程中有描述。

大概思路：1 首先根据高层配置选择绑定模式还是复用模式；2 如果是绑定，则绑定范围内多个子帧的同一个codeword的Crc结果相与，得到和codeword数目相同的bit，每个对应一个codeword的绑定ACK/NACK信息，根据比特数目为1或者2分别选择1a或者1b；3 如果是复用模式，则每个子帧内可能的codeword的crc校验结果相与，得到一个bit的ACK信息，然后再根据复用范围内每个子帧的ACK信息，查表获得最终的上报比特。

花雨小筑主页博客相册|个人档案 |好友查看文章LTE有关下行链路HARQ的几个问题2010-06-29 15:49如果只是从标准流程上看HARQ和Rate Matching的话，恐怕根本无法理解其实质吧，虽然HARQ貌似是个提出很久的理论，但在LTE的实施中其设计还是非常精巧的，以至于我至今也没能完全理解它和Rate Matching之间的关系。

其实重点不是在HARQ，而是HARQ的冗余版本如何控制Rate Matching中数据比特选择和修剪的，其中有几个概念一度让我非常混淆。

HARQ基本原理LTE HARQ快速重传，只涉及到L2/L1层，重传合并产生合并增益N-process Stop-And-WaitDL:自适应异步HARQUL ACK/NAK 在PUCCH/PUSCH发送PDCCH 携带 HARQ 进程号重传总是通过PDCCH调度，这是因为它采用异步自适应HARQUL:同步HARQ，相对于第一次传输，会在固定的地方重传最大传输次数是针对UE的而不是RB在PHICH 发送DL ACK/NAK概述增量冗余HARQ除了传统的Chase合并的HARQ技术，LTE还采用了增量冗余（IR）HARQ，既通过第一次传输发送信息bit和一部分的冗余bit，而通过重重发送额外的冗余bit，如果第一次传输没有成功解码，则可以通过重传更多的冗余bit降低信道的编码率，从而实现更高的解码成功率。

如果加上重重的冗余bit仍无法正确解码，则进行再次重传，随着重重次数的增加，冗余bit不断积累，信道编码率不断降低，从而可以获得更好的解码效果。

HARQ正对每个传输块进行重传。

进程实现模式下行HARQ采用多进程的“停止-等待”HARQ实现方式，即对于某一个HARQ进程，在等待ACK/NACK反馈之前，此进程暂时中止传输，当收到反馈后，再根据反馈的是ACK还是NACK 选择发送新的数据还是重传。

同步/异步HARQ按照重传发生的时刻来区分，可以将HARQ可以分为同步和异步两类。

这也是目前在3G LTE中讨论比较多的话题之一。

同步HARQ是指一个HARQ进程的传输(重传)是发生在固定的时刻，由于接收端预先已知传输的发生时刻，因此不需要额外的信令开销来标示HARQ进程的序号，此时的HARQ进程的序号可以从子帧号获得；异步HARQ是指一个HARQ进程的传输可以发生在任何时刻，接收端预先不知道传输的发生时刻，因此HARQ进程的处理序号需要连同数据一起发送。

由于同步HARQ的重传发生在固定时刻，在没有附加进程序号的同步HARQ在某一时刻只能支持一个HARQ进程。

20140307 （HARQ、HARQ process、HARQ information、同步/异步、自适应/非自适应、ACK/NACK反馈、上行HARQ（1））一、HARQ介绍HARQ（Hybrid Automatic Repeat reQuest），混合式自动重传请求，是一种结合FEC（Forward Error Correction）与ARQ（Automatic Repeat reQuest）方法的技术。

FEC通过添加冗余信息，使得接收端能够纠正一部分错误，从而减少重传的次数。

对于FEC无法纠正的错误，接收端会通过ARQ机制请求发送端重发数据。

接收端使用检错码，通常为CRC校验，来检测接收的数据包是否出错。

如果无错，则发送一个肯定的确认（ACK）；如果出错，则接收端会丢弃数据包，并发送一个否定的确认（NACK）给发送端，发送端收到NACK后，会重发相同的数据。

前面介绍的ARQ机制采用丢弃数据包并请求重传的方式。

然而，虽然这些数据包无法被正确解码，但其中还是包含了有用的信息，如果丢弃了，这些有用的信息就丢失了。

通过使用HARQ with soft combining，接收到的错误数据包会保存在一个HARQ buffer中，并与后续接收到的重传数据包进行合并，从而得到一个比单独解码更可靠的数据包。

然后对合并后的数据包进行解码，如果还是失败，则再请求重传，再进行软合并。

根据重传的bit信息与原始传输是否相同，HARQ with soft combining 分为Chase combining和incremental redundancy（IR，增量冗余）两类。

Chase combining中重传的bit信息与原始传输相同；增量冗余中重传的bit信息不需要与原始传输相同。

这里我们只介绍增量冗余，因为LTE中使用的是这种机制。

在增量冗余中，每一次重传并不需要与初始传输相同。

相反，会生成多个coded bit的集合，每个集合都携带相同的信息。

[转载]LTE：下⾏HARQ简介原⽂地址：LTE：下⾏HARQ简介作者：⾦辉_LTE下⾏HARQ简介下⾏HARQ使⽤异步（asynchronous）、⾃适应（adaptive）的⽅式意味着重传可能发⽣在任意时刻和频域上的任意位置。

下⾏HARQ使⽤异步、⾃适应⽅式的原因是为了避免与系统信息、MBSFN⼦帧之类的传输产⽣冲突。

通过在时域或频域上重新调度重传，避免与系统信息、MBSFN⼦帧在⽆线资源上发⽣重叠。

也就是说，下⾏重传总是通过PDCCH来调度的。

下⾏PDSCH传输对应的ACK/NACK在PUCCH或PUSCH上发送。

对TDD⽽⾔，DwPTS可⽤于发送PCFICH/PHICH/PDCCH和PDSCH。

也就是说，特殊⼦帧可⽤于发送UL grant和对应PUSCH数据的ACK/NACK等。

这⾥有个特例：对于带normal downlink CP的特殊帧配置0和5，或带extended downlink CP的特殊帧配置0和4（见36.211的Table 4.2-1。

），特殊⼦帧的DwPTS是不⽤于传输PDSCH的（见36.213的7.1.7节。

这些配置下，DwPTS的持续时间都很短，占3个OFDM symbol，所以不⽤于传输PDSCH）。

对TDD⽽⾔，UpPTS可⽤于发送SRS和PRACH（preamble format 4），但不能⽤于发送PUCCH和PUSCH（见36.213的8.2节；36.211的5.4、5.7节。

UpPTS的持续时间都很短，所以不⽤于传输PUCCH和PUSCH）。

也就是说，特殊⼦帧不能⽤于发送对应于PDSCH传输的ACK/NACK。

因此，对于下⾏数据⽽⾔，只能使⽤⾮特殊⼦帧来发送对应的ACK/NACK。

通常eNodeB和UE侧的解码处理时间⾄少为3ms。

对于下⾏数据⽽⾔，UE根据接收到的PDCCH来决定如何处理HARQ。

图1：下⾏DCI中与HARQ相关的字段（以DCI format 2A为例）图1介绍了下⾏DCI中与HARQ相关的字段，我在《LTE：HARQ（三）---HARQ information》中介绍过这些字段，这⾥就不再重复了。

LTE：上行HARQ（二）接下来，我们来看看上行是如何进行同步的。

首先需要说明的是，如果UE需要在PUSCH上发送数据，UE需要满足以下两个条件之一： 1）收到一个有效的UL Grant：该UL grant可以来自动态调度的PDCCH（DCI format 0/4，本文只介绍这种情况）、或来自RAR，或通过半静态semi-persistently配置。

2）收到一个PHICH且指示为NACK：对应非自适应重传。

接下来，我们分FDD、TDD 1~6、TDD 0三种配置来介绍上行HARQ在时域上的同步关系！每种配置都包含2部分：1）UL grant/PHICH与对应的PUSCH传输之间的timing关系；2）PUSCH传输与对应的PHICH（ACK/NACK）之间的timing关系。

1) FDD对FDD而言，如果UE在子帧n收到了UL grant（DCI format 0/4，对应新传或自适应重传）或PHICH（只收到NACK，对应非自适应重传），则UE会在n + 4子帧发送对应的PUSCH。

（见36.213的8.0节）对FDD而言，如果UE在子帧n收到了PHICH，则该PHICH对应UE 在上行子帧n - 4发送的PUSCH。

（见36.213的8.3节）如图3所示。

图3：FDD中的上行传输，UL grant、PUSCH、ACK/NACK之间的timing关系子帧n，n+4、n+8、n+12、n+16都对应同一HARQ process。

只要确定了子帧n所使用的HARQ process number，根据timing关系，也就知道后续子帧n+4、n+8、n+12、n+16所使用的HARQ process。

TDD的情况类似，但是timing关系略有不同。

（注：每个子帧只对应一个HARQ process，空分复用的情况下是2个，每个对应一个TB。

）2) TDD 1~6对TDD UL/DL configuration 1~6而言，如果UE在子帧n收到了UL grant（DCI format 0/4，对应新传或自适应重传）或PHICH（只收到NACK，对应非自适应重传），则UE会在n + k子帧发送对应的PUSCH。

▊HARQ混合自动重传1、HARQ技术LTE中HARQ技术主要是系统端对编码数据比特的选择重传以及终端对物理层重传数据合并。

在这里涉及到2个方面，一个就是自动重传请求也就是ARQ技术，另外一个就是前向纠错技术FEC。

也可以这么说HARQ=ARQ+FECFEC是一种编码技术，编码的作用主要就是保证传输的可靠性，具有自动纠错的能力。

举个例子，如果我要传输信息0，我可以发0000，如果收到干扰变成了0001或者1000的话，FEC可以纠正为0000，从而增加了容错率，而只发一个0的话一旦干扰成了1就会造成误码。

而假如接收端收到的是1100，由于1和0一样多所以，会认为是错码，从而要求重传，触发ARQ。

而ARQ技术则是收到信息后，会通过CRC校验位进行校验，如果发现错误了或者压根就没收到这个包会回NAK要求重传，否则回ACK说明已经收到了。

2、HARQ有两种运行方式：⑴跟踪（Chase）或软合并（Soft Combining）方式－即数据在重传时，与初次发射时的数据相同；⑵递增冗余(Incremental Redundancy)方式－即重传时的数据与发射的数据有所不同。

后一种方式的性能要优于第一种，但在接收端需要更大的内存。

终端的缺省内存容量是根据终端所能支持的最大数据速率和软合并方式设计的，因而在最大数据速率时，只可能使用软合并方式。

而在使用较低的数据速率传输数据时，两种方式都可以使用。

3、HARQ流程下面是一个软合并的流程图图HARQ流程从图1看，ENB先发一个packet1给UE，UE没有解调出来，回NAK给ENB。

这时候ENB将packet1另外一部分发给UE，UE通过两次发送的包进行软合并，解出来回ACK，ENB收到后继续发packet2。

这里要的一点是，HARQ发端每发一个包都会开一个timer，如果timer到时了还没有下一个包到来，ENB会认为这是最后一个包，会发一个指示给UE，告诉它发完了，防止最后一个包丢失。

20140307 （HARQ、HARQ process、HARQ information、同步/异步、自适应/非自适应、ACK/NACK反馈、上行HARQ（1））一、HARQ介绍HARQ（Hybrid Automatic Repeat reQuest），混合式自动重传请求，是一种结合FEC（Forward Error Correction）与ARQ（Automatic Repeat reQuest）方法的技术。

FEC通过添加冗余信息，使得接收端能够纠正一部分错误，从而减少重传的次数。

对于FEC无法纠正的错误，接收端会通过ARQ机制请求发送端重发数据。

接收端使用检错码，通常为CRC校验，来检测接收的数据包是否出错。

如果无错，则发送一个肯定的确认（ACK）；如果出错，则接收端会丢弃数据包，并发送一个否定的确认（NACK）给发送端，发送端收到NACK后，会重发相同的数据。

前面介绍的ARQ机制采用丢弃数据包并请求重传的方式。

然而，虽然这些数据包无法被正确解码，但其中还是包含了有用的信息，如果丢弃了，这些有用的信息就丢失了。

通过使用HARQ with soft combining，接收到的错误数据包会保存在一个HARQ buffer中，并与后续接收到的重传数据包进行合并，从而得到一个比单独解码更可靠的数据包。

然后对合并后的数据包进行解码，如果还是失败，则再请求重传，再进行软合并。

根据重传的bit信息与原始传输是否相同，HARQ with soft combining分为Chase combining和incremental redundancy（IR，增量冗余）两类。

Chase combining中重传的bit信息与原始传输相同；增量冗余中重传的bit信息不需要与原始传输相同。

这里我们只介绍增量冗余，因为LTE中使用的是这种机制。

在增量冗余中，每一次重传并不需要与初始传输相同。

相反，会生成多个coded bit的集合，每个集合都携带相同的信息。

LTE技术分析报告-- HARQ技术1DL HARQ1.1 DL Assignment receptionDownlink assignments transmitted on the PDCCH indicate if there is a transmission on the DL-SCH for a particular UE and provide the relevant HARQ information.When the UE has a C-RNTI, Semi-Persistent Scheduling C-RNTI, or Temporary C-RNTI, the UE shall for each TTI during which it monitors PDCCH:1.1.1PDCCH for the UE‟s C-RNTI, or Temporary- if a downlink assignment for this TTI has been received on the PDCCH for the UE‟s C-RNTI, or Temporary C-RNTI:- if this is the first downlink assignment for this Temporary C-RNTI:- consider the NDI to have been toggled.- if the downlink assignment is for UE‟s C-RNTI and if the previous downlink assignment indicated to the HARQ entity of the same HARQ process was either a downlink assignment received for the UE‟sSemi-Persistent Scheduling C-RNTI or a configured downlink assignment:- consider the NDI to have been toggled regardless of the value of the NDI.- indicate the presence of a downlink assignment and deliver the associated HARQ information to the HARQ entity for this TTI.1.1.2PDCCH for the UE‟s Semi-Persistent Scheduling C-RNTI- else, if a downlink assignment for this TTI has been received on the PDCCH for the UE‟s Semi-Persistent Scheduling C-RNTI:- if the NDI in the received HARQ information is 1:- consider the NDI not to have been toggled;- indicate the presence of a downlink assignment and deliver the associated HARQ information to the HARQ entity for this TTI.- else, if the NDI in the received HARQ information is 0:- if PDCCH contents indicate SPS release:- clear the configured downlink assignment (if any);- if timeAlignmentTimer is running:- indicate a positive acknowledgement for the downlink SPS release to the physical layer.- else:- store the downlink assignment and the associated HARQ information as configured downlink assignment;- initialise (if not active) or re-initialise (if already active) the configured downlink assignment to start in this TTI and to recur according to rules in subclause 5.10.1;- set the HARQ Process ID to the HARQ Process ID associated with this TTI;- consider the NDI bit to have been toggled;- indicate the presence of a configured downlink assignment and deliver the stored HARQ information to the HARQ entity for this TTI.1.1.3no measurement gap in this TTI:- else, if a downlink assignment for this TTI has been configured and there is no measurement gap in this TTI:- instruct the physical layer to receive, in this TTI, transport block on the DL-SCH according to the configured downlink assignment and to deliver it to the HARQ entity;- set the HARQ Process ID to the HARQ Process ID associated with this TTI;- consider the NDI bit to have been toggled;- indicate the presence of a configured downlink assignment and deliver the stored HARQ information to the HARQ entity for this TTI.1.1.4HARQ Process ID for UE‟s Semi-Persistent Scheduling C-RNTIFor downlink assignments received on the PDCCH for the UE‟s Semi-Persistent Scheduling C-RNTI and for configured downlink assignments, the HARQ Process ID associated with this TTI is derived from the following equation:HARQ Process ID = [floor(CURRENT_TTI/(Downlink Semi-Persistent Scheduling Interval))] modulo Number of Configured SPS Processes,where CURRENT_TTI=[(SFN * 10) + subframe number], Downlink Semi-Persistent Scheduling Interval is the periodicity of semi-persistent scheduling signalled via RRC and Number of Configured SPS Processes is the number of HARQ processes allocated for semi-persistent scheduling signalled via RRC.1.1.4.1ASN-TEXTSPS_Configuration ::= SEQUENCE {semiPersistSchedC_RNTI C_RNTI OPTIONAL, -- need ODsps_ConfigurationDL SPS_ConfigurationDL OPTIONAL, -- need ONsps_ConfigurationUL SPS_ConfigurationUL OPTIONAL -- need ON}SPS_ConfigurationDL ::= CHOICE{disable NULL,enable SEQUENCE {semiPersistSchedIntervalDL ENUMERA TED {sf10, sf20, sf32, sf40, sf64, sf80, sf128, sf160, sf320, sf640, spare6, spare5, spare4, spare3, spare2, spare1}, numberOfConfSPS_Processes INTEGER (1..8),n1Pucch_AN_Persistent INTEGER (0..2047),...}}SPS_ConfigurationUL ::= CHOICE {disable NULL,enable SEQUENCE {semiPersistSchedIntervalUL ENUMERA TED {sf10, sf20, sf32, sf40, sf64, sf80,sf128, sf160, sf320, sf640, spare6,spare5, spare4, spare3, spare2, spare1}, implicitReleaseAfter ENUMERA TED {e2, e3},p0_Persistent SEQUENCE {p0_NominalPUSCH_Persistent INTEGER (_126..24),p0_UePUSCH_Persistent INTEGER (_8..7)} OPTIONAL, -- need OP...}}1.1.5read BCCHWhen the UE needs to read BCCH, the UE may, based on the scheduling information from RRC:- if a downlink assignment for this TTI has been received on the PDCCH for the SI-RNTI;- if the redundancy version is not defined in the PDCCH format:- the redundancy version of the received downlink assignment for this TTI is determined by RV K = ceiling(3/2*k) modulo 4, where k depends on the type of system information message: forSystemInformationBlockType1 message, k = (SFN/2) modulo 4, where SFN is the system frame number;for SystemInformation messages, k=i modulo 4, i =0,1,…, n s w–1, where i denotes the subframe numberwithin the SI window n s w;- indicate a downlink assignment and redundancy version for the dedicated broadcast HARQ process to the HARQ entity for this TTI.1.2 HARQ EntityThere is one HARQ entity at the UE 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 (see subclause 5.3.2.2).The number of DL HARQ processes is specified in [36213], clause 7.When the physical layer is configured for spatial multiplexing [2], 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.NOTE: In case of BCCH a dedicated broadcast HARQ process is used.1.2.1Physical downlink shared channel related proceduresFor FDD, there shall be a maximum of 8 HARQ processes in the downlink.For TDD, the maximum number of HARQ processes in the downlink shall be determined by the UL/DL configuration (Table 4.2-2 of [3]), as indicated in table 7-1.The dedicated broadcast HARQ process defined in [8] is not counted as part of the maximum number of HARQ processes for both FDD and TDD.Table 7-1: Maximum number of DL HARQ processes for TDD1.3 HARQ processFor each subframe where a transmission takes place for the HARQ process, one or two (in case of spatial multiplexing) TBs and the associated HARQ information are received from the HARQ entity.For each received TB and associated HARQ information, the HARQ process shall:- if the NDI, when provided, has been toggled compared to the value of the previous received transmission corresponding to this TB; or- if the HARQ process is equal to the broadcast process and if this is the first received transmission for the TB according to the system information schedule indicated by RRC; or- if this is the very first received transmission for this TB (i.e. there is no previous NDI for this TB):- consider this transmission to be a new transmission.- else:- consider this transmission to be a retransmission.The UE then shall:- if this is a new transmission:- replace the data currently in the soft buffer for this TB with the received data.- else if this is a retransmission:- if the data has not yet been successfully decoded:- combine the received data with the data currently in the soft buffer for this TB.- if the TB size is different from the last valid TB size signalled for this TB:- the UE may replace the data currently in the soft buffer for this TB with the received data.- attempt to decode the data in the soft buffer for this TB;- if the data in the soft buffer was successfully decoded for this TB:- if the HARQ process is equal to the broadcast process:- deliver the decoded MAC PDU to upper layers.- else if this is the first successful decoding of the data in the soft buffer for this TB:- deliver the decoded MAC PDU to the disassembly and demultiplexing entity.- generate a positive acknowledgement (ACK) of the data in this TB.- else:- generate a negative acknowledgement (NACK) of the data in this TB.- if the HARQ process is associated with a transmission indicated with a Temporary C-RNTI and the Contention Resolution is not yet successful (see subclause 5.1.5); or- if the HARQ process is equal to the broadcast process; or- if timeAlignmentTimer is stopped or expired:- do not indicate the generated positive or negative acknowledgement to the physical layer.- else:- indicate the generated positive or negative acknowledgement for this TB to the physical layer.The UE shall ignore NDI received in all downlink assignments on PDCCH for its Temporary C-RNTI when determining if NDI on PDCCH for its C-RNTI has been toggled compared to the value in the previous transmission.1.3.1Downlink Conrol Information（36321）1.3.1.1Format 1 (PDSCH)DCI format 1 is used for the scheduling of one PDSCH codeword.1.3.1.2Format 1A (PDSCH)DCI format 1A is used for the compact scheduling of one PDSCH codeword and random access procedure initiated by a PDCCH order.1.3.1.3Format 1B (PDSCH)DCI format 1B is used for the compact scheduling of one PDSCH codeword with precoding information.1.3.1.4Format 1D (PDSCH)DCI format 1D is used for the compact scheduling of one PDSCH codeword with precoding and power offset information.1.3.1.5Format 2 (PDSCH)DCI format 2 is used for scheduling PDSCH to UEs configured in closed-loop spatial multiplexing mode.1.4 HARQ Feedback Procedure1.4.1UE procedure for reporting ACK/NACK (36213)For FDD, when both ACK/NACK and SR are transmitted in the same sub-frame a UE shall transmit the ACK/NACK on its assigned ACK/NACK PUCCH resource for a negative SR transmission and transmit the ACK/NACK on its assigned SR PUCCH resource for a positive SR transmission.For TDD and all UL-DL configurations except configuration 5, two ACK/NACK feedback modes are supported by higher layer configuration.- ACK/NACK bundling, and -ACK/NACK multiplexingFor TDD UL-DL configuration 5, only ACK/NACK bundling is supported .For TDD, the UE shall upon detection of a PDSCH transmission or a PDCCH indicating downlink SPS release (defined in section 9.2) within subframe(s) k n -, where K k ∈ and K is defined in Table 10.1-1 intended for the UE and for which ACK/NACK response shall be provided, transmit the ACK/NACK response in UL subframe n .Table 10.1-1: Downlink association set index K : {}110,,-M k k k for TDDTDD UL/DL Configuration = 0Downlink Subframe Special subframe UpLink SubframeFor TDD UL-DL configurations 1-6, the value of the Downlink Assignment Index (DAI) in DCI format 0, ULDAI V ,detected by the UE according to Table 7.3-X in subframe 'k n -, where 'k is defined in Table 7.3-Y, represents the total number of subframes with PDSCH transmissions and with PDCCH indicating downlink SPS release tothe corresponding UE within all the subframe(s)k n -, where K k ∈. The value ULDAI V includes all PDSCHtransmission with and without corresponding PDCCH within all the subframe(s) k n -. In case neither PDSCH transmission, nor PDCCH indicating the downlink SPS resource release is intended to the UE, the UE can expectthat the value of the DAI in DCI format 0, ULDAI V , if transmitted, is set to 4.For TDD UL-DL configurations 1-6, the value of the DAI in DCI format 1/1A/1B/1D/2/2A denotes theaccumulative number of PDCCH(s) with assigned PDSCH transmission(s) and PDCCH indicating downlink SPS release up to the present subframe within subframe(s) k n -, where K k ∈, and shall be updated from subframe to subframe(这句话非常关键). Denote DLDAI V as the value of the DAI in PDCCH with DCI format 1/1A/1B/1D/2/2A detected by the UE according to Table 7.3-X in subframe m k n -, where m k is the smallest value in the set K (defined in Table 10.1-1) such that the UE detects a DCI format 1/1A/1B/1D/2/2A. Denote DAI U as the total number of PDCCH(s) with assigned PDSCH transmission(s) and PDCCH indicating downlink SPS release detected by the UE within the subframe(s) k n -, where K k ∈. Denote SPS N , which can be zero or one, as the number of PDSCH transmissions without a corresponding PDCCH within the subframe(s) k n -, where K k ∈.For TDD ACK/NACK bundling or ACK/NACK multiplexing and a subframe n with 1=M , the UE shall generate one or two ACK/NACK bits by performing a logical AND operation per codeword across M DL subframes associated with a single UL subframe, of all the corresponding SPS DAI N U + individual PDSCH transmission ACK/NACKs and individual ACK in response to received PDCCH indicating downlink SPS release, where M is the number of elements in the set K defined in Table 10.1-1. The UE shall detect if at least one downlink assignment has been missed, and for the case that the UE is transmitting on PUSCH the UE shall also determine the parameter bundled N . For TDD UL-DL configuration 0, bundled N shall be 1 if UE detects the PDSCH transmission with or without corresponding PDCCH within the subframe k n -, where K k ∈. -For the case that the UE is not transmitting on PUSCH in subframe n and TDD UL-DL configurations 1-6,if 0>DAI U and ()14mod 1+-≠D AI D LD AI U V , the UE detects that at least one downlink assignment has been missed. -For the case that the UE is transmitting on PUSCH and the PUSCH transmission is adjusted based on a detected PDCCH with DCI format 0 intended for the UE and TDD UL-DL configurations 1-6, if()14mod 1+-+≠SPS D AI U L D AI N U V the UE detects that at least one downlink assignment has been missed and the UE shall generate NACK for all codewords where bundled N is determined by the UE as2bundled +=U L D AI V N . If the UE does not detect any downlink assignment missing, bundled N is determined by the UE as U L D AI V N =bundled . UE shall not transmit ACK/NACK if 0=+SPS DAI N U and 4=ULDAI V .-For the case that the UE is transmitting on PUSCH, and the PUSCH transmission is not based on a detected PDCCH with DCI format 0 intended for the UE and TDD UL-DL configurations 1-6, if0>DAI U and ()14mod 1+-≠D AI D LD AI U V , the UE detects that at least one downlink assignment has been missed and the UE shall generate NACK for all codewords. The UE determines)(bundled SPS D AI N U N += as the number of assigned subframes. The UE shall not transmit ACK/NACKif 0=+SPS DAI N U .For TDD ACK/NACK bundling, when the UE is configured by transmission mode 3 or 4 defined in Section 7.1and ACK/NACK bits are transmitted on PUSCH, the UE shall always generate 2 ACK/NACK bits assuming both codeword 0 and 1 are enabled. For the case that the UE detects only the PDSCH transmission associated with codeword 0 within the bundled subframes, the UE shall generate NACK for codeword 1.For TDD ACK/NACK multiplexing and a subframe n with 1>M , spatial ACK/NACK bundling across multiple codewords within a DL subframe is performed by a logical AND operation of all the corresponding individual ACK/NACKs. In case the UE is transmitting on PUSCH, the UE shall determine the number of ACK/NAK feedback bits ACKOand the ACK/NACK feedback bits ACKn o , 1,,0-=ACK O n to be transmitted insubframe n in case the UE is transmitting on PUSCH. -If the PUSCH transmission is adjusted based on a detected PDCCH with DCI format 0 intended for theUE, then UL DAI ACK V O = unless 4=ULDAI V and 0=+SPS D AI N U in which case the UE shall nottransmit ACK/NACK. The spatially bundled ACK/NACK for a PDSCH transmission with acorresponding PDCCH or for a PDCCH indicating downlink SPS release in subframe k n - is associatedwith ACKk D AI o 1)(- where DAI(k) is the value of DAI in DCI format 1A/1B/1D/1/2/2A detected in subframek n -. For the case with0>SPS N , the ACK/NACK associated with a PDSCH transmission without acorresponding PDCCH is mapped to ACKO ACK o 1-. The ACK/NACK feedback bits without any detected PDSCH transmission or without detected PDCCH indicating downlink SPS release are set to NACK. -If the PUSCH transmission is not adjusted based on a detected PDCCH with DCI format 0 intended for the UE, M OACK=, and ACK i o is associated with the spatially bundled ACK/NACK for DL subframei k n -, where K k i ∈. The ACK/NACK feedback bits without any detected PDSCH transmission orwithout detected PDCCH indicating downlink SPS release are set to NACK.For TDD when both ACK/NACK and SR are transmitted in the same sub-frame, a UE shall transmit the bundledACK/NACK or the multiple ACK/NAK responses (according to section 10.1) on its assigned ACK/NACK PUCCH resources for a negative SR transmission. For a positive SR, the UE shall transmit )1(),0(b b on its assigned SR PUCCH resource using PUCCH format 1b according to section 5.4.1 in [3]. The value of )1(),0(b b are generated according to Table 7.3-1 from the SPS DAI N U + ACK/NACK responses including ACK in response to PDCCH indicating downlink SPS release by spatial ACK/NAK bundling across multiple codewords within each PDSCH transmission.Table 7.3-1: Mapping between multiple ACK/NACK responses and )1(),0(b bFor TDD when both ACK/NACK and CQI/PMI or RI are configured to be transmitted in the same sub-frame on PUCCH, a UE shall transmit CQI and )1(),0(b b using PUCCH format 2b for normal CP or PUCCH format 2 forextended CP, according to section 5.2.3.4 in [4] with 10,a a '''' replaced by )1(),0(b b . The value of )1(),0(b b are generated according to Table 7.3-1 from the SPS DAI N U + ACK/NACK responses including ACK in response to PDCCH indicating downlink SPS release by spatial ACK/NACK bundling across multiple codewords within eachPDSCH transmission.When only ACK/NACK or only a SR is transmitted a UE shall use PUCCH Format 1a or 1b for the ACK/NACK resource and PUCCH Format 1 for the SR resource as defined in section 5.4.1 in [3]. 1.4.2UE procedure for determining physical uplink control channel assignmentUplink control information (UCI) in subframe n shall be transmitted- on PUCCH using format 1/1a/1b or 2/2a/2b if the UE is not transmitting on PUSCH in subframe n-on PUSCH if the UE is transmitting on PUSCH in subframe n unless the PUSCH transmission corresponds to a Random Access Response Grant or a retransmission of the same transport block as part of the contention based random access procedure, in which case UCI is not transmittedThroughout this section, subframes are numbered in monotonically increasing order; if the last subframe of a radio frame is denoted k , the first subframe of the next radio frame is denoted 1+k . The following combinations of uplink control information on PUCCH are supported: - HARQ-ACK using PUCCH format 1a or 1b- HARQ-ACK using PUCCH format 1b with channel selection - Scheduling request (SR) using PUCCH format 1 - HARQ-ACK and SR using PUCCH format 1a or 1b - CQI using PUCCH format 2-CQI and HARQ-ACK using PUCCH format- 2a or 2b for normal cyclic prefix - 2 for extended cyclic prefixThe scrambling initialization of PUCCH format 2, 2a and 2b is by C-RNTI.The parameter Simultaneous-AN-and-CQI provided by higher layers determine if a UE can transmit a combination of CQI and HARQ-ACK on PUCCH in the same subframe.For TDD, two ACK/NACK feedback modes are supported by higher layer configuration.- ACK/NACK bundling and -ACK/NACK multiplexingFor TDD UL-DL configuration 5, only ACK/NACK bundling is supported.TDD ACK/NACK bundling is performed per codeword across M multiple DL subframes associated with a single UL subframe n , where M is the number of elements in the set K defined in Table 10.1-1, by a logical AND operation of all the individual PDSCH transmission (with and without corresponding PDCCH) ACK/NACKs and ACK in response to PDCCH indicating downlink SPS release. The bundled 1 or 2 ACK/NACK bits are transmitted using PUCCH format 1a or PUCCH format 1b, respectively.For TDD ACK/NACK multiplexing and a subframe n with 1>M , where M is the number of elements in the set K defined in Table 10.1-1, spatial ACK/NACK bundling across multiple codewords within a DL subframe is performed by a logical AND operation of all the corresponding individual ACK/NACKs and PUCCH format 1b with channel selection is used. For TDD ACK/NACK multiplexing and a subframe n with 1=M , spatial ACK/NACK bundling across multiple codewords within a DL subframe is not performed, 1 or 2 ACK/NACK bits are transmitted using PUCCH format 1a or PUCCH format 1b, respectively.For FDD, the UE shall use PUCCH resource )1(PUCCH n for transmission of HARQ-ACK in subframe n , where- for a PDSCH transmission indicated by the detection of a corresponding PDCCH in subframe 4-n , or for a PDCCH indicating downlink SPS release (defined in section 9.2) in subframe 4-n , the UE shall use(1)PUCCH CCE )1(PUCCH N n n +=, where CCE n is the number of the first CCE used for transmission of the corresponding DCI assignment and (1)PUCCH N is configured by higher layers.- for a PDSCH transmission where there is not a corresponding PDCCH detected in subframe 4-n , the value of)1(PUCCH n is determined according to higher layer configuration and Table 9.2-2.For TDD ACK/NACK bundling or TDD ACK/NACK multiplexing and a subframe n with 1=M where M isthe number of elements in the set K defined in Table 10.1-1, the UE shall use PUCCH resource )1(PUCCH n fortransmission of HARQ-ACK in subframe n , where-If there is PDSCH transmission indicated by the detection of corresponding PDCCH or there is PDCCH indicating downlink SPS release within subframe(s) k n -, where K k ∈ and K (defined in Table 10.1-1) isa set of M elements{}110,,-M k k k depending on the subframe n and the UL-DL configuration (defined inTable 4.2-2 in [3]), the UE first selects a p value out of {0, 1, 2, 3} which makes 1+<≤p CCE p N n N andshall use)1(1)1()1(PU CCH CCE p p PU CCH N n N m N m M n ++⨯+⨯--=+, where (1)PUCCH N is configured byhigher layers, ⎣⎦{}36/)]4([,0max RBsc D L RB -⨯⨯=p N N N p , and CCE n is the number of the first CCE used fortransmission of the corresponding PDCCH in subframem k n -and the corresponding m , where m k is thesmallest value in set K such that UE detects a PDCCH in subframe m k n -.-If there is only a PDSCH transmission where there is not a corresponding PDCCH detected within subframe(s)k n -, where K k ∈and K is defined in Table 10.1-1, the value of )1(PUCCH n is determined according tohigher layer configuration and Table 9.2-2.Table 10.1-1: Downlink association set index K : {}110,,-M k k k for TDDFor TDD ACK/NACK multiplexing and sub-frame n with 1>M , where M is the number of elements in the set K defined in Table 10.1-1, denote )1(i PUCCH,n as the ACK/NACK resource derived from sub-frame i k n - and HARQ-ACK(i) as the ACK/NACK/DTX response from sub-frame i k n -, where K k i ∈(defined in Table 10.1-1) and 10-≤≤M i .-For a PDSCH transmission or a PDCCH indicating downlink SPS release in sub-framei k n - whereK k i ∈, the ACK/NACK resource )1(,1)1(,)1(PU CCH i CCE p p i PU CCH N n N i N i M n ++⨯+⨯--=+, wherep is selected from {0, 1, 2, 3} such that 1+<≤p CCE p N n N , ⎣⎦{}36/)]4([,0max RB sc D L RB -⨯⨯=p N N N p ,i CCE,n is the number of the first CCE used for transmission of the corresponding PDCCH in subframe i k n -,and (1)PUCCH N is configured by higher layers.- For a PDSCH transmission where there is not a corresponding PDCCH detected in subframe i k n -, the valueof)1(i PUCCH,n is determined according to higher layer configuration and Table 9.2-2.The UE shall transmit )1(),0(b b on an ACK/NACK resource )1(PUCCH n in sub-frame n using PUCCH format 1b according to section 5.4.1 in [3]. The value of )1(),0(b b and the ACK/NACK resource )1(PUCCH n are generated bychannel selection according to Table 10.1-2, 10.1-3, and 10.1-4 for M = 2, 3, and 4, respectively. In case )1(),0(b b are mapped to “N/A,” then the UE shall not transmit ACK/NACK response in sub -frame n .Table 10.1-4: Transmission of ACK/NACK multiplexing for M = 4ACK/NACK repetition is enabled or disabled by a UE specific parameter ackNackRepetition configured byhigher layers. Once enabled, the UE shall repeat any ACK/NACK transmission with a repetition factorANRep N , where ANRep N is provided by higher layers and includs the initial ACK/NACK transmission, untilACK/NACK repetition is disabled by higher layers. For a PDSCH transmission without a corresponding PDCCH detected, the UE shall transmit the corresponding ACK/NACK response ANRep N times usingPUCCH resource )1(PUCCH n configured by higher layers. For a PDSCH transmission with a correspondingPDCCH detected, or for a PDCCH indicating downlink SPS release, the UE shall first transmit the corresponding ACK/NACK response once using PUCCH resource derived from the corresponding PDCCH CCE index (as described in Section 10.1), and repeat the transmission of the corresponding ACK/NACK response 1A N Rep -N times always using PUCCH resource )1(A N Rep PU CCH ,n , where )1(A N Rep PU CCH ,n is configured by higher layers.For TDD, ACK/NACK repetition is only applicable for ACK/NACK bundling and is not applicable for ACK/NACK multiplexing.The scheduling request (SR) shall be transmitted on the PUCCH resource )1(SRI PU CCH ,)1(PU CCH n n =as defined in [3], where )1(SRI PUCCH,n is UE specific and configured by higher layers. The SR configuration for SR transmission periodicity and subframe offset is defined by SR configuration index SR I in Table 10.1-5. SR transmission instances are the subframes satisfying ⎣⎦()0mod 2/10,=-+⨯y Periodicit SR O FFSET s f SR N n n , where f n is the system frame number, and s n = {0,1,…, 19} is the slot index wit hin the frame, and N OFFSET,SR is the SR subframe offset defined in Table 10.1-5 and SR Periodicity is the SR periodicity defined in Table 10.1-5.Table 10.1-5: UE-specific SR periodicity and subframe offset configuration1.4.2.1 PUCCH formats 1, 1a and 1b (36211)For PUCCH format 1, information is carried by the presence/absence of transmission of PUCCH from the UE. In the remainder of this section, 1)0(=d shall be assumed for PUCCH format 1.For PUCCH formats 1a and 1b, one or two explicit bits are transmitted, respectively. The block of bits )1(),...,0(bit -M b b shall be modulated as described in Table 5.4.1-1, resulting in a complex-valued symbol )0(d . The modulation schemes for the different PUCCH formats are given by Table 5.4-1.Table 5.4.1-1: Modulation symbol )0(d for PUCCH formats 1a and 1b.The complex-valued symbol )0(d shall be multiplied with a cyclically shifted length 12PU CCHseq =N sequence )()(,n r v u α according to1,...,1,0 ),()0()(PU CCHseq)(,-=⋅=N n n r d n y v u α （使用一个序列代替原有的PUCCH 信息）where )()(,n r v u α is defined by section 5.5.1 (Generation of the reference signal sequence) with PU CCHseqRS sc N M =. The cyclic shift α varies between symbols and slots as defined below.The block of complex-valued symbols )1(),...,0(PU CCH seq -N y y shall be scrambled by )(s n S and block-wise。

20140310 HARQ，BSRLTE：上行HARQ（二）接下来，我们来看看上行是如何进行同步的。

首先需要说明的是，如果UE需要在PUSCH上发送数据，UE需要满足以下两个条件之一：1）收到一个有效的UL Grant：该UL grant可以来自动态调度的PDCCH（DCI format 0/4，本文只介绍这种情况）、或来自RAR，或通过半静态配置。

2）收到一个PHICH且指示为NACK：对应非自适应重传。

接下来，我们分FDD、TDD 1~6、TDD 0三种配置来介绍上行HARQ在时域上的同步关系！每种配置都包含2部分：1）UL grant/PHICH与对应的PUSCH传输之间的timing 关系；2）PUSCH传输与对应的PHICH（ACK/NACK）之间的timing关系。

1) FDD对FDD而言，如果UE在子帧n收到了UL grant（DCI format 0/4，对应新传或自适应重传）或PHICH（只收到NACK，对应非自适应重传），则UE会在n + 4子帧发送对应的PUSCH。

（见36.213的8.0节）对FDD而言，如果UE在子帧n收到了PHICH，则该PHICH对应UE在上行子帧n - 4发送的PUSCH。

（见36.213的8.3节）如图3所示。

图3：FDD中的上行传输，UL grant、PUSCH、ACK/NACK之间的timing关系子帧n，n+4、n+8、n+12、n+16都对应同一HARQ process。

只要确定了子帧n所使用的HARQ process number，根据timing关系，也就知道后续子帧n+4、n+8、n+12、n+16所使用的HARQ process。

TDD的情况类似，但是timing关系略有不同。

（注：每个子帧只对应一个HARQ process，空分复用的情况下是2个，每个对应一个TB。

）2) TDD 1~6对TDD UL/DL configuration 1~6而言，如果UE在子帧n收到了UL grant（DCI format 0/4，对应新传或自适应重传）或PHICH（只收到NACK，对应非自适应重传），则UE会在n + k子帧发送对应的PUSCH。

SAW并行的HARQ 进程eNodeB UELTE 系统的HARQeNodeBUE 八个FDD 下行FDDTDD确认模式●TDD模式下，发给UE的携带ACK/NACK的PHICH在i号子帧上发送，针对在i-k号子帧上发送的PUSCH提供确认信息，其中k的值取决于TDD配置模式表。

另外，TDD 还定义了如下两种ACK/NACK反馈模式：☐ACK/NACK捆绑式反馈。

☐ACK/NACK复用式反馈。

下行HARQ下行HARQ有如下特征：●下行使用非同步自适应HARQ。

●针对下行初传和重传的上行ACK/NACK通过PUCCH或者PUSCH发送。

●PDCCH上发送HARQ进程号，指示进行初传或者重传。

●固定通过PDCCH进行重传的调度。

上行HARQ上行HARQ有如下特征：●上行使用同步HARQ。

●最大重传次数面向UE设置，而非面向无线承载设置。

●针对上行初传和重传的下行ACK/NACK通过PHICH发送。

上行HARQ遵循如下原则：●无论HARQ反馈的内容是ACK还是NACK，UE只要正确接收了PDCCH信息就按照PDCCH的指示发送新数据或者进行重传（这种重传被称为自适应重传）。

●当UE没有检测到针对其C-RNTI的PDCCH，则UE按照HARQ反馈的指示进行重传。

具体如下：−如果HARQ反馈的是NACK，则UE进行非自适应重传，即在该进程中之前使用过的上行资源上进行重传。

−如果HARQ反馈的是ACK，则UE不进行任何上行初传或重传，而是把数据缓存在HARQ缓存区里，直到接收到PDCCH指示再进行重传。

也就是说，这种场景下，UE不采用非自适应重传。

LTEHARQ
HARQ协议根据时域灵活性分为同步和异步，也可以频域灵活性分为自适应和非自适应。

在LTE情况下，异步自适应HARQ用于下行链路。

上行链路则采用同步，非自适应的。

在上行方向HARQ给RLC层的报文也是乱序的，即某个PDU在物理层多次HARQ重传后才成功。

所以这段时间RLC不要发重传状态报告，否则MAC层成功解调的PDU又要重传了。

有两个解决方案：
1. 有个定时器，当定时器超时后RLC才可发送重传报告。

FDD是8个并行HARQ，这样假设HARQ重换次数为3，那么潮湿定时器应该是3 * 8 = 24ms.
2. mac HARQ发现某个PDU重传失败了，立刻通告RLC，由RLC重传。

LTE HARQARQ stands for Automatic Repeat Request,而且使用的是前向纠错FEC (forward error correction)。

HARQ实体的总体架构如下：在FDD模式下，HARQ进程有8个。

下行：1.任何同步进程中可用8个HARQ进程2.网络通过PDCCH给UE下发Process ID, RV上行：1.它必须在特定子帧中使用特定进程（同步进程）。

UE必须每8个子帧使用相同的HARQ进程号。

2.由于UE必须在特定子帧处使用特定的HARQ进程ID，所以接收器（eNode b）确切地知道当哪个HARQ进程出现时。

eNodeB还可以知道RV，因为eNodeB的UL Grant（DCI 0）可以使用MCS字段指定RV。

3.它有两种工作模式：自适应和非自适应HARQ（自适应）下面是自适应UL-HARQ过程的一个例子（关键思想是每个UL重传使用不同的RV，RV由DCI 0确定）。

（非自适应）下面是非自适应UL-HARQ过程的一个例子（关键思想是每个UL重传使用不同的RV，并且RV由TS36.321“5.4.2.2 HARQ过程”中指定的预定义序列确定）。

最后一个非常重要的问题是“UE如何知道它是否应该进行自适应重传和非自适应重传？”●如果在未切换NDI的情况下检测到DCI 0，则UE执行“自适应重传”。

（在这种情况下，UE不关心“HARQ反馈（PHICH）”，它基于DCI 0信息重新传输）。

●如果得到“HARQ反馈（PHICH=NACK）”，但没有得到DCI 0，则UE执行“非自适应重传”。

在这种情况下，UE在没有来自DCI 0的信息的情况下，在预定义的RV和MCS中重新传输PUSCH。

那么，有没有具体的规则（数学公式）来计算从SFN和子帧数HARQ进程ID？在LTE中，3GPP规范中没有定义具体的公式，但是以下是LTE情况下最简单的规则之一。

UL HARQ Process ID = (SFN x 10 + subframe) modulo 8接收者（在LTE情况下是eNodeB）需要知道确切的HARQ进程ID吗？不是真的。

LTE 中ACK,RI,CQI 相关技术介绍进入信道编码单元的控制信息包含信道质量信息（CQI 或PMI ）、HARQ-ACK 或秩指示（Rank Indication ）。

传输的不同编码符号数决定了控制信息的不同的码率。

当控制信息在PUSCH 上传输时，HARQ-ACK 、秩指示和信道质量指示的信道编码是独立于PUSCH 进行的。

一． ACK/NACK对于TDD ，高层配置支持两种ACK/NACK 反馈模式：- ACK/NACK 绑定（ACK/NACK bundling ），和- ACK/NACK 复用（ACK/NACK multiplexing ）对于TDD ACK/NACK 绑定模式，HARQ-ACK 由1 或2 个信息比特构成。

对TDD ACK/NAK 复用模式， HARQ-ACK 由2 或4 个信息比特构成。

当终端发送HARQ-ACK 比特时，相应的编码符号数Q ’由终端根据下式决定：⎪⎪⎪⎪⎪⎭⎫⎝⎛⋅⎥⎥⎥⎥⎥⎤⎢⎢⎢⎢⎢⎡⋅⋅⋅='∑-=--PUSCHsc C r r PUSCHoffset initial PUSCH symb initial PUSCH sc M K N M O Q 4,min 10β 其中：O 为ACK/NACK 比特数；PU SCHscM 为当前子幀发送的PUSCH 对应的传输块所占用的带宽，以子载波数目来表示；initial-PU SCH symb N 表示每个子幀中，上述传输块所承载的初始PUSCH 传输所占用的SC-FDMA 符号的数目，其表达式为()()SRS U Lsymb initial-PU SCH symb 12N N N --⋅=。

SRS N 可以分情况得出。

initialPUSCH sc M -，C ，r K 可以从该传输块的初始PDCCH 中得到。

ACK H ARQ offset PU SCH offset -=ββ，ACK H ARQ offset -β可确定。

5G和LTE中的HARQ协议LTE中有两种重传机制：MAC层的HARQ机制，以及RLC层的ARQ（只针对AM（aknowledgement mode确认模式）数据传输）机制。

HARQ：HARQ（HybridAutomatic Repeat reQuest混合⾃动重传请求），是⼀种结合FEC（ForwardError Correction，前向纠错）与ARQ（Automatic RepeatreQuest）⽅法的技术。

FEC通过添加冗余信息，使得接收端能够纠正⼀部分错误，从⽽减少重传的次数。

对于FEC⽆法纠正的错误，接收端会通过ARQ机制请求发送端重发数据。

接收端使⽤检错码，通常为CRC校验，来检测接收到的数据包是否出错。

如果⽆错，则接收端会发送⼀个肯定的确认（ACK）给发送端，发送端收到ACK后，会接着发送下⼀个数据包。

如果出错，则接收端会丢弃该数据包，并发送⼀个否定的确认（NACK）给发送端，发送端收到NACK后，会重发相同的数据。

前⾯介绍的ARQ机制采⽤丢弃数据包并请求重传的⽅式。

虽然这些数据包⽆法被正确解码，但其中还是包含了有⽤的信息，如果丢弃了，这些有⽤的信息就丢失了。

通过使⽤HARQ with softcombining（带软合并的HARQ），接收到的错误数据包会保存在⼀个HARQ buffer中，并与后续接收到的重传数据包进⾏合并，从⽽得到⼀个⽐单独解码更可靠的数据包（“软合并”的过程）。

然后对合并后的数据包进⾏解码，如果还是失败，则重复“请求重传，再进⾏软合并”的过程。

根据重传的⽐特信息与原始传输是否相同，HARQ with softcombining分为chase combining和incrementalredundancy（增量冗余）两类。

chase combining中重传的⽐特信息与原始传输相同；incremental redundancy 中重传的⽐特信息不需要与原始传输相同。

95 移动终端中的无线承载复用需要根据一定的规则来进行，相关的参数由来自eNodeB 的RRC 信令来配置。

每个无线承载都分配一个优先级和优先级比特速率。

移动终端将执行无线承载复用，这些无线承载按照优先级顺序进行服务，而且能获得的最大速率就是优先级比特速率。

如果满足优先级比特速率后仍然剩余资源，那么将按优先级顺序给其他无线承载使用。

为了协助上行调度器进行判决，移动终端可以将调度信息以MAC 消息的方式发送给eNode B 。

很明显，只有当一个移动终端被调度时它才能发送该信息。

当情况不是这样时，移动终端需要的上行资源指示符通过上行L1/L2控制信令结构的一部分来提供。

信道相关调度通常用于下行。

理论上也可以用于上行。

但是，对上行信道质量的估计不能像下行时那么直接。

下行信道条件可以通过小区内的所有移动终端测量获得，即终端测量eNode B 发射的导频信号来获得，并且为了信道质量估计，所有的终端可以共享相同的导频信号。

但是，上行信道质量的估计需要每个移动终端都能发送一个合理的导频信号，eNode B 可以通过该信号估计上行信道质量。

LTE 支持这样的导频信号，将在后面对此进行介绍，但是很明显，这样会增加系统开销。

所以，作为上行信道相关调度的补充或者代替方式，提供上行分集是非常重要的。

5.8.4 HARQ具有软合并的LTE HARQ 的作用与HSPA 的非常相似，都是为了提供传输的健壮性，从而克服传输差错。

它还可以用于增强容量。

由于HARQ 的重传很快，所以很多业务都允许一次或者多次重传，因此，这样就间接地实现了速率控制。

与HSPA 的HARQ 一样，它是MAC 层的一部分，而软合并操作由物理层来完成。

很明显，HARQ 不能应用到所有的业务类型。

例如，广播业务需要将相同的信息发送到多个用户，所以通常不会使用HARQ 机制。

因此，HARQ 仅支持DL-SCH 和UL-SCH 。

LTE 的HARQ 协议与HSPA 所使用的对应协议非常相似，都是多个并行的停－等进程。

LTE的HARQ进程数研究
根据协议213中的表9.1.2-1可以知道，当上行子帧n在PUSCH上发送上行信息，则在下行子帧在n+k PHICH中反馈ACK/NACK信息，如下表所示
Table 9.1.2-1: PHICH
k for TDD
根据协议213的授权或PHICH，则在n+k中发送上行PUSCH，如下表所示
Table 8-2 k for TDD configurations 0-6
将上面的2张表合到一块，用颜色标注上下行，如下图
ACK/NAK timing and RTT for TDD DL&UL
HARQ的进程数是由RTT的时间来决定的，以上行行子帧配比0的上行子帧3为例进行说明，RTT为11ms，在这11ms中将会有7个上行子帧，为了避免上行子帧的浪费，需要在这些上行子帧上发送其他进程的数据，因此需要设定上行行子帧配比0下HARQ的进程数为7，具体的如下图所示。

根据上面的分析方法，可以知道上行和下行的HARQ的进程数，具体如下表所示。

HARQ进程数。

HARQ详解HARQHARQ（Hybrid Automatic Repeat Request ）混合自动重传请求。

HARQ的关键词是存储、请求重传、合并解调。

接收方在解码失败的情况下，保存接收到的数据，并要求发送方重传数据，接收方将重传的数据和先前接收到的数据进行合并后再解码。

这里面就有一定的分集增益，减少了重传次数，进而减少了时延。

而传统的ARQ 技术简单地抛弃错误的数据，不做存储，也就不存在合并的过程，自然没有分集增益，往往需要过多地重传、过长时间地等待。

正如一个人喜欢吹牛说谎，但吹牛说谎前不打草稿，想怎么说就怎么说，对同一件事情的描述每次都不一样，全看心情。

如果你用ARQ的方式听他说话，你知道说得不靠谱，但真实的情况还是不知道；但是你用HARQ的方式听他说话，把几次说的话放在一起分析，真实的情况就跃然纸上了。

R99版本的ARQ中，数据包的重传工作由RNC完成；而HSDPA 的HARQ技术则主要由Node B完成数据包的选择重传，由终端完成重传数据的合并，这就大大提高了重传的速度。

混合自动重传技术可以高效地补偿由于采用链路适配所带来的误码，提高了数据传输速率，减小了数据传输时延。

数据通信最初是在有线网上发展起来的，通常要求较大的带宽和较高的传输质量。

对于有线连接，数据传输的可靠性是通过重传来实现的。

当前一次尝试传输失败时，就要求重传数据分组，这样的传输机制就称之为ARQ（自动请求重传）。

在无线传输环境下，信道噪声和由于移动性带来的衰落以及其他用户带来的干扰使得信道传输质量很差，所以应该对数据分组加以保护来抑制各种干扰。

这种保护主要是采用前向纠错编码（FEC），在分组中传输额外的比特。

然而，过多的前向纠错编码会使传输效率变低。

因此，一种混合方案HARQ，即ARQ和FEC相结合的方案被提出了。

HARQ是将FEC与ARQ结合起来使用，称为混合自动重传请求，即Hybird ARQ。

HARQ的基本原理如下：1. 在接收端使用FEC技术纠正所有错误中能够纠正的那一部分。