116
`
`LTE for UMTS — OFDMAand SC-FDMABased Radio Access
`
`Transmit Diversity
`Transmit diversity with two or four antenna ports using SFBC.
`Open-loop spatial multiplexing
`This is an open loop mode with the possibility to do rank adaptation based on the RI
`feedback. In the case of rank = | transmit diversity is applied similarly to transmission
`mode 2. With higher rank spatial multiplexing with up to four layers with large delay,
`CDDis used.
`Closed-loop spatial multiplexing
`This is a spatial multiplexing mode with pre-coding feedback supporting dynamic rank
`adaptation.
`Multi-user MIMO
`Transmission mode for downlink MU-MIMOoperation.
`Closed-loop Rank= 1 pre-coding
`Closed loop pre-coding similar to transmission mode 5 without the possibility of spatial
`multiplexing, i.e. the rank is fixed to one.
`Single-antenna port; port 5
`This mode can be used in a beam forming operation when UEspecific reference signals
`are in use.
`
`5.8.5. Physical Broadcast Channel (PBCH)
`
`The physical broadcast Channel (PBCH)carries the system information needed to access the
`system, such as RACHparameters, and as covered in more detail in Chapter 6. The channelis
`always provided with 1.08 MHz bandwidth, as shown in Figure 5.36, so the PBCHstructure
`is independent of the actual system bandwidth being used, similar to other channels/signals
`neededforinitial system access. The PBCHis convolutionally encoded as the data rate is not
`that high. As discussed in Chapter6, the broadcast informationis partly carried on the PBCH,
`
`10 ms = 10 subframes
`
`1.08 MHz {
`
`Synchronization
`Signals
`
`
`
`
`
`Figure 5.36
`
`PBCHlocationat the center frequency
`
`Samsung Ex. 1010
`251 of 1365
`
`

`

`Physical Layer
`
`117
`
`where the Master Information Block (MIB) is transmitted while the actual System Information
`Blocks (SIBs)are then on the PDSCH.The 600 sub-carriers in Figure 5.36 need only 9 MHz (50
`resource blocks) in the resource domain but the system bandwidth needed forsufficient attenu-
`ation for the adjacent operator increases the total bandwidth needed to 10 MHz,as discussed
`in Chapter 11. With a 1.4 MHz system bandwidth there are no resource blocks on either side
`of the PBCHin the frequency domain in use, so effectively only 6 resource blocks may be in
`use to meet the spectrum mask requirements.
`
`5.8.6 Synchronization Signal
`
`There are 504 Physical Cell Identities (PCIs) values in the LTE system, compared with the
`512 primary scrambling codes in WCDMA.The Primary Synchronization Signal (PSS) and
`the Secondary Synchronization Signals (SSS) are transmitted, similar to PBCH, always with
`the 1.08 MHz bandwidth, located in the end of Ist and 11th slots (slots O and 10) of the 1Oms
`frame, as shown in Figure 5.37.
`The PSSand SSSjointly point the space of 504 unique Physical-layer Cell Identities (PCIs).
`The PCIs form 168 PCI groups, each of them having 3 PCIs (thus a total of 504 PCIs). The
`location and structure of the PCIs meanthat taking a sample from the center frequency (with
`a bandwidth of 1.08 MHz) for a maximum of 5 ms contains the necessary information needed
`for cell identification.
`
`5.9 Physical Layer Procedures
`
`The key physical layer procedures in LTE are powercontrol, HARQ, timing advance and
`random access. Timing advance is based on the signaling in the Medium Access Control
`(MAC)layer (as shown in the MACsection in Chapter6), but as it is directly related to the
`physical layer, the timing advance details are covered in this chapter. The big contrast to
`WCDMA isthat there are no physical layer issues related to macro-diversity since the UE is
`only connected to one basestation at a time and hand handoveris applied. Also, a specific
`meansfor dealing with inter-system andinter-frequency measurements such as compressed
`
`_] sss
`
`ly PSS
`
`1.08 mz |
`
`PSS/SSS
`
`Figure 5.37 Synchronization signals in the frame
`
`Samsung Ex. 1010
`252 of 1365
`
`

`

`118
`
`LTE for UMTS — OFDMAand SC-FDMABased Radio Access
`
`modeis not needed in LTE, as LTE by naturehas a discontinuousoperation that will facilitate
`the measurements by scheduling.
`
`5.9.1 HARQ Procedure
`
`The HARQin LTE isbased onthe use of a stop-and-wait HARQ procedure. Once the packet
`is transmitted from the eNodeB, the UE will decode it and provide feedback in the PUCCH,
`as described in section 5.6. For negative acknowledgement (NACK) the eNodeB will send a
`retransmission. The UE will combinethe retransmission with the original transmission and
`will run the turbo decoding again. Upon successful decoding (based on CRC check) the UE
`will send positive acknowledgement (ACK)for the eNode. After that eNodeB will send a new
`packet for that HARQ process. Dueto the stop-and-wait way of operating, one needs to have
`multiple HARQprocesses to enable a continuous data flow. In LTE the numberof processes
`is fixed to 8 processesin both the uplink and downlink direction. An example of a single user
`continuous transmissionis illustrated in Figure 5.38. For multiple users, it is dependent on
`the eNodeB scheduler when a retransmissionis sent in the uplink or downlink direction, as a
`retransmission also requires that resourcesare allocated.
`The HARQoperation in LTE supports both soft combiningandthe use of incremental redun-
`dancy. The use of soft combining meansthat retransmission has exactly the same rate matching
`parameters as the original transmission and thus exactly the same symbols are transmitted. For
`incremental redundancy, the retransmission may have different rate matching parameters like
`the original transmission. The minimum delay between the end of a packet andthestart of a
`retransmission is 7 ms. The UE will send the ACK/NACKfora packetin frame7, in the uplink
`frame n+4. This leaves around 3 ms processing time for the UE, depending on the uplink/
`downlink timing offset controlled by the timing advance procedure. The downlink timing for
`a single transmitted downlink packet is shown in Figure 5.39. The retransmission instant in
`the downlink is subject to the scheduler in eNodeB and thusthe timing shown in Figure 5.39
`is the earliest momentfor a retransmission to occur.
`
`From scheduler buffer
`
`4
`
`5
`
`6
`
`7
`
`8
`
`1
`
`PUSCH/PDSCH
`
`2
`
`3
`
`2
`
`CRC CheckResult
`
`Fay
`
`ase
`
`RLClayer
`
`Figure 5.38 LTE HARQoperation with 8 processes
`
`Samsung Ex. 1010
`253 of 1365
`
`

`

`Physical Layer
`
`119
`
`PDSCH
`
`—_—’"
`lms
`
`PUCCHor PUSCH
`
`New Packet or
`Retransmission
`
`
`
`PDSCH
`
`ms UE processing time —————~ 3 ms eNodeB processing time
`lms
`
`Figure 5.39 LTE HARQtimingfor a single downlink packet
`
`5.9.2 Timing Advance
`
`The timing control procedure is needed so that the uplink transmissions from different users
`arrive at the eNodeBessentially within the cyclic prefix. Such uplink synchronization is
`needed to avoid interference between the users with uplink transmissions scheduled on the
`same subframe. The eNodeB continuously measures the timing of the UE uplink signal and
`adjusts the uplink transmission timing as shown in Figure 5.40. Timing advance commands
`are sent only when a timing adjustmentis actually needed. The resolution of a timing advance
`commandis 0.52 us, and timing advance is defined relative to the timing of the downlink radio
`frame received on UE.
`The timing advance value is measured from RACH transmission when the UE does not
`have a valid timing advance, i.e. the uplink for the UE is not synchronized. Suchsituations are
`system access, when the UE is in RRC_IDLEstate or when the UE has hadan inactivity period
`exceeding related timer, non-synchronized handover, and after radio link failure. Additionally,
`eNodeBcan assign to UE a dedicated (contention-free) preamble on RACHforuplink timing
`measurement when eNodeB wants to establish uplink synchronization. Such situations are
`faced with handover or when downlink data arrive for anon-synchronized UE. From the range
`defined for timing advance,cell sizes up to 100km would befacilitated, and even beyond by
`leaving someresources unused.
`
`5.9.3 Power Control
`
`For LTE, powercontrol is slow for the uplink direction. In the downlink direction there is no
`powercontrol. As the bandwidth varies due to data rate changes, the absolute transmission
`powerof the UE will also change. The powercontrol does not now actually control absolute
`powerbut rather the Power Spectral Density (PSD), power per Hz, for a particular device. What
`facilitates the use of a slowerrate for powercontrol is the use of orthogonal resources in the LTE
`uplink, which avoids the near—far problem that required fast power control in WCDMA.The key
`motivation for the powercontrolis to reduce terminal power consumption andalso to avoid an
`overly large dynamic rangein the eNodeBreceiver, rather than to mitigate interference. In the
`
`UplinkdataorRACH )
`
`————_—_————
`rr —
`Timing Advance (+n x 0.52us)=
`eNodeB
`
`UE
`
`Figure 5.40 Uplink timing control
`
`Samsung Ex. 1010
`254 of 1365
`
`

`

`120
`
`LTE for UMTS — OFDMAand SC-FDMABased Radio Access
`
`Power a
`
`Spectral
`Density
`
`Power per Hz unchanged
`
`2 X data
`rate
`
`Frequency
`
`TTIn
`
`Frequency
`TTI n+]
`
`Figure 5.41 LTE uplink powerwith data rate change
`
`receiver the PSDs of different users have to be reasonably close to each other so the receiver
`A/D converter has reasonable requirements and also the interference resulting from the non-
`ideal spectrum shape of the VE transmitter is kept under control. The LTE uplink powercontrol
`principle is illustrated in Figure 5.41 where at the change of data rate the PSD stays constant
`but the resulting total transmission poweris adjusted relative to the data rate change.
`The actual powercontrolis based onestimating the pathloss, taking into accountcell specific
`parameters and then applying the (accumulated) value of the correction factor received from
`the eNodeB. Dependingon the higher layer parametersettings, the power control command
`is either | dB up or downorthen the set of [-1dB, 0, +1dB, +3 dB] is used. The specifications
`also include powercontrolthat is absolute value based but, based on the textcase prioritization,
`it is not foreseen that this will be used in thefirst phase networks. The total dynamic range of
`powercontrolis slightly smaller than in WCDMA andthe devices now have a minimum power
`level of -41 dBm compared to -50dBm with WCDMA.
`
`5.9.4 Paging
`
`To enable paging, the UE is allocated a paging interval and a specific subframe within that
`interval where the paging message could be sent. The paging is provided in the PDSCH(with
`allocation information on the PDCCH). The key design criterion in pagingis to ensure a suf-
`ficient DRX cycle for devices to save powerandalso to ensure a fast enough response time for
`the incomingcall. The EEUTRAN mayparameterize the duration of the paging cycle to ensure
`sufficient paging capacity (covered in more detail in Chapter 6).
`
`5.9.5 Random Access Procedure
`
`The LTE Random Access (RACH)operation resembles that of WCDMA because both use
`preambles andsimilar ramping of preamble power. Theinitial power is based on the measured
`path loss in DL, and powerrampingis necessary because ofthe relatively coarse accuracy of
`the UE in path loss measurement and absolute powersetting, and to compensate for uplink
`fading. Although LTE PRACHresourcesare separate from PUSCH and PUCCH,powerramp-
`ing is useful for simultaneous detection of different preamble sequences and for minimizing the
`interference due to asynchronous PRACHtransmission at the adjacent PUCCH and PUSCH
`resources. The steps of the physical layer procedure are as follows:
`
`Samsung Ex. 1010
`255 of 1365
`
`

`

`Physical Layer
`
`121
`
`e Transmit a preamble using the PRACHresource, preamble sequence and powerselected by
`MAC.
`e@ Wait for the RACH response with matching preamble information (PRACHresource and
`preamble sequence). In addition to the preamble information, the response also contains
`the information on the uplink resource to use for further information exchange as well as
`the timing advance to be used. In the WCDMA RACHprocedure, after acknowledging a
`preamble, the UE continues with a 10 or 20msframe duration of data, or even longer, as
`described for Release 8 HSPA operation in Chapter 13. The fundamentaldifference in LTE is
`that the device will moveinstead directly to the use of UL-SCH onreception of the random
`access response, which has the necessary information.
`e Ifno matching random accessresponseis received, transmit the preamblein the next avail-
`able PRACHresource according to instructions of MAC,as shownin Figure 5.42.
`
`Although the LTE specification models that the physical layer just transmits preambles and
`detects responses underthe control of MAC, wedescribe below the complete procedure without
`focusing on the modeling ofthe specification.
`Two fundamentally different random access procedures have been specified for LTE. The
`contention based procedure is what we normally understand with random access: UEstrans-
`mit randomly selected preambles on a commonresourceto establish a network connection or
`request resourcesfor uplink transmission. The non-contention based random accessIs initiated
`by the network for synchronizing UE’suplink transmission, and the network can identify the
`UE from the very first uplink transmission. This procedure is nevertheless included underthe
`LTE random access because it uses PRACH resources. Both procedures are common for TDD
`and FDD systems.
`
`5.9.5.1 Contention and Non-contention Based Random Access
`
`The contention based procedure follows the signaling diagram in Figure 5.43(left half).
`In the first step, the UE transmits a preamble sequence on PRACH.Thedetails of PRACH
`and preamble sequencesare explained in section 5.7. For each cell a total of 64 sequences are
`reserved, and these are groupedfor the non-contention based and contention based procedures.
`The groupreserved for the contention based procedures1s divided further into two: by selecting
`the proper group, the UE sends onebit of information about the transport block size that the
`UE desires to send on PUSCH in Step3.
`In the second step, the UE receives a preamble response on DL-SCHresourcethat is assigned
`on PDCCH.The identity RA-RNTI that is used for this assignment is associated with the
`
` PRACH
`Onthe resources
`response
`PRACHresponse
`indicated by
`Downlink / eNodeB
`Not detected
`
`Uplink / UE
`
`Preamble
`
`Preamble
`
`UE specific data
`
`Figure 5.42 Powerramping in random access procedure
`
`Samsung Ex. 1010
`256 of 1365
`
`

`

`122
`
`LTE for UMTS — OFDMAand SC-FDMABased Radio Access
`
`Contention based RA
`
`Non-contention based RA
`
`
`
`
`
`1. Preamble on PRACH 1. Preamble on PRACH
`
`2. Preamble response on PDCCH+
`2. Preamble response on PDCCH +
`
`DL-SCH DL-SCH
`
`3. PUSCHtransmissionincluding
`contention resolution identity
`
`4. Contention resolution message
`
`Figure 5.43 The contention and non-contention based random access procedures
`
`frequency and time resource of the preamble. This permits bundling of the responsesthat are
`meant for preambles transmitted in the same PRACH frequency and time resource, whichis
`important for saving PDCCHresources. eNodeB transmits the response in a time window that
`can be configured up to 10 msin duration. The flexible window allows freedom for dimension-
`ing of the RACHreceiver and for scheduling of the responses.
`The response(signaling part of the MAClayer as covered in Chapter6) lists the sequence
`numbers of the observed preambles, and in addition the following information is given for
`each acknowledged preamble:
`
`e A grantfor the first transmission on PUSCH,including also information on the need for
`frequency hopping, power control commandfor uplink transmission and information on
`the need for CQI transmission and whether the PUSCH transmission needs to be delayed
`by one subframe from the nominalvalue.
`e A timing alignment command.
`e A temporary allocation for identity called temporary CRNTI, whichis used for addressing
`PUSCH grants and DL-SCH assignmentsin Steps 3 and 4 of the procedure.
`
`Thetypical probability of preamble collisions, meaning that two or more UEsare transmit-
`ting the same preamble sequence in the same frequency and time resource, is expected to be
`around 1%. These collisions are resolved in Steps 3 and 4: the UE includesits identity in the
`first message that it transmits on PUSCH in Step 3 and expects in Step 4 an acknowledge-
`ment that eNodeB hasreceived the identity. There are two forms of acknowledgement: it can
`be either (1) a PUSCH grant or DL-SCH assignment addressed with CRNTI if the UE had
`included CRNTIto the message of Step 3, or (2) the UE’s identity can be acknowledged with
`a messagethat is sent on a DL-SCHresource assigned with the temporary CRNTI. Thefirst
`form of acknowledgementis for RRC connected UEs while the second form is used when a
`UE tries to establish or re-establish RRC connection. HARQis used both in Step 3 and 4. In
`
`Samsung Ex. 1010
`257 of 1365
`
`

`

`Physical Layer
`
`123
`
`Step 3 there is no difference compared with normal HARQ, but in Step 4 the UE never sends
`NAK, and ACK is sent only by the UE that wins the contention resolution. No special actions
`are taken after a lost contention resolution but the UE simply retransmits a preamble just like
`after failing to receive the preamble response.
`The LTE network can control the RACH load rapidly. If the UE does not receive acknowl-
`edgement for its preamble in Step 2 or for its identity in Step 4, the UE retransmits the preamble
`with increased power if power ramp-up has been confi gured. Normally the retransmission can
`be done as soon as the UE is ready but the network can also confi gure a back-off parameter
`that forces the UE to add a random delay before the retransmission. When needed, the back-
`off parameter is included in the preamble response message, and the setting is obeyed by all
`the UE decoding the message. This allows much faster load control than in WCDMA where a
`similar load control parameter is in the broadcasted System Information.
`The non-contention based procedure, shown in Figure 5.43 (right half), is used for time align-
`ment during handover and when an RRC connected UE needs to be synchronized for downlink
`data arrival. The UE receives in the handover command or through PDCCH signaling an index
`of its dedicated preamble sequence, which it is allowed to transmit on PRACH. Besides the
`sequence index, some restrictions for the frequency and time resource can be signaled so that
`the same sequence can be simultaneously allocated for UEs that transmit on different PRACH
`subframes or, for TDD, at different PRACH frequencies. The preamble responses in the con-
`tention and non-contention based procedures are identical and they can thus be bundled to the
`same response message. As eNodeB knows the identity of the UE that has sent the dedicated
`preamble, the contention resolution with Steps 3 and 4 is not needed.
`The non-contention based procedure provides delay and capacity enhancements compared
`with the contention based procedure. As the preamble collisions are absent and the contention
`resolution is not needed, a shorter delay can be guaranteed, which is especially important for
`handover. The sequence resource is in effective use because it is assigned to the UE only when
`needed and can be released as soon as eNodeB detects that the UE has received the preamble
`response.
`An unsuccessful random access procedure ends based on preamble count or RRC timers.
`The preamble count is decisive only with two causes of random access: (a) an RRC connected
`UE, lacking scheduling request resources, asks resources because of uplink data arrival or (b)
`an RRC connected UE needs to be synchronized because of DL data arrival. If random access
`has been started because of RRC connection establishment or re-establishment or because of
`handover, the procedure continues until success or MAC reset in the expiry of the RRC timer
`corresponding to the cause of random access.
`
`5.9.6 Channel Feedback Reporting Procedure
`The purpose of the channel state feedback reporting is to provide the eNodeB with information
`about the downlink channel state in order to help optimize the packet scheduling decision. The
`principle of the channel state feedback reporting procedure is presented in Figure 5.44. The
`channel state is estimated by the UE based on the downlink transmissions (reference symbols,
`etc.) and reported to the eNodeB by using PUCCH or PUSCH. The channel state feedback
`reports contain information about the scheduling and link adaptation (MCS/TBS and MIMO)
`related parameters the UE can support in the data reception. The eNodeB can then take advan-
`tage of the feedback information in the scheduling decision in order to optimize the usage of
`the frequency resources.
`
`Samsung Ex. 1010
`258 of 1365
`
`

`

`124
`
`LTE for UMTS — OFDMAand SC-FDMABased Radio Access
`
` 4. eNodeB frequency
`domain downlink
`
`scheduling
`
`
`
`1. eNodeB
`
`
`
`transmission 2. UE CSI
`
`measurements
`
`“eNodeB
`
`Figure 5.44 Channel State Information (CSI) reporting procedure
`
`In general the channel feedback reported by the UE is just a recommendation and the eNodeB
`does not need to follow it in the downlink scheduling. In LTE the channel feedback reporting
`is always fully controlled by the eNodeB and the UE cannot send any channelstate feedback
`reports without eNodeB knowing it beforehand. The corresponding procedure for providing
`information about the uplink channelstate is called channel soundingandit is done using the
`Sounding Reference Signals (SRS)as presented in Section 5.6.4.3.
`The main difference of the LTE channelstate information feedback compared to WCDMA/
`HSDPAis the frequency selectivity of the reports, i.e. the information regarding the distribu-
`tion of channel state over the frequency domain can also be provided. This is an enabler for
`Frequency Domain Packet Scheduling (FDPS), a method that aimsto divide the radio resources
`in the frequency domainfor different users so that system performanceis optimized. In Figure
`5.45 the gain from the FDPSis illustrated. As the UE speedincreases, the CSI reports become
`moreinaccurate and get outdated faster leading to reduced gains in high mobility.
`
`Wideband CQI
`Best-M average CQI
`—— Uncompressed CQI
`
`(Mbps)
`
`
`Averagecellthroughput
`
`3
`
`30
`
`.
`UEvelocity (km/h)
`
`300
`
`Figure 5.45 Comparison ofthe average cell throughputs for different CQI schemesandvelocities
`
`Samsung Ex. 1010
`259 of 1365
`
`

`

`Physical Layer
`
`125
`
`5.9.6.1 Channel Feedback Report Types in LTE
`
`In LTE the UE can send three types of channel feedback information:
`• CQI – Channel Quality Indicator
`• RI – Rank Indicator
`• PMI – Pre-coding Matrix Indicator
`The most important part of channel information feedback is the Channel Quality Indicator
`(CQI). The CQI provides the eNodeB information about the link adaptation parameters the UE
`can support at the time (taking into account the transmission mode, the UE receiver type, number
`of antennas and interference situation experienced at the given time). The CQI is defi ned as a
`table containing 16 entries (Table 5.3) with Modulation and Coding Schemes (MCSs). The UE
`reports back to the eNodeB the highest CQI index corresponding to the MCS and TBS for which
`the estimated received DL transport block BLER shall not exceed 10%. The CQI operation has a
`high degree of similarity with HSDPA CQI use, as covered in [5]. Note that there are many more
`possibilities for MCS and TBS size values than only those 15 indicated by the CQI feedback.
`Rank Indicator (RI) is the UE’s recommendation for the number of layers, i.e. streams to
`be used in spatial multiplexing. RI is only reported when the UE is operating in MIMO modes
`with spatial multiplexing (transmission modes 3 and 4). In single antenna operation or TX
`diversity it is not reported. The RI can have values 1 or 2 with 2-by-2 antenna confi guration
`and from 1 up to 4 with 4-by-4 antenna confi guration. The RI is always associated with one
`or more CQI reports, meaning that the reported CQI is calculated assuming that particular RI
`value. Since the rank varies typically more slowly than the CQI it is normally reported less
`
`Table 5.3 CQI table
`
`Modulation
`
`CQI
`index
`
`Coding rate
`× 1024
`
`Bits per resource
`element
`
` 0
` 1
` 2
` 3
` 4
` 5
` 6
` 7
` 8
` 9
`10
`11
`12
`13
`14
`15
`
`out of range
`QPSK
`QPSK
`QPSK
`QPSK
`QPSK
`QPSK
`16QAM
`16QAM
`16QAM
`64QAM
`64QAM
`64QAM
`64QAM
`64QAM
`64QAM
`
` 78
`120
`193
`308
`449
`602
`378
`490
`616
`466
`567
`666
`772
`873
`948
`
`0.1523
`0.2344
`0.3770
`0.6016
`0.8770
`1.1758
`1.4766
`1.9141
`2.4063
`2.7305
`3.3223
`3.9023
`4.5234
`5.1152
`5.5547
`
`Samsung Ex. 1010
`260 of 1365
`
`

`

`126
`
`LTE for UMTS – OFDMA and SC-FDMA Based Radio Access
`
`often. RI always describes the rank on the whole system band, i.e. frequency selective RI
`reports are not possible.
`The PMI provides information about the preferred pre-coding matrix in codebook based
`pre-coding. Like RI, PMI is also relevant to MIMO operation only. MIMO operation with PMI
`feedback is called Closed Loop MIMO. The PMI feedback is limited to transmission modes
`4, 5, and 6. The number of pre-coding matrices in the codebook depends on the number of
`eNodeB antenna ports: in the case of two antenna ports there are altogether six matrices to
`choose from, while with four antenna ports the total number is up to 64 depending on the RI
`and the UE capability. PMI reporting can be either wideband or frequency selective depending
`on the CSI feedback mode.
`
`5.9.6.2 Periodic and Aperiodic Channel State Feedback Reporting
`
`Although in principle the UE has up-to-date information about the changes in channel state, a
`channel state feedback report initiated by the UE would raise several issues. First, to detect the
`reports blind decoding would need to be performed at the eNodeB, which is not desirable from
`the receiver implementation point of view. Secondly, as the eNodeB is anyway fully in charge
`of the scheduling decisions, UE initiated reports would often be unnecessary. Furthermore, the
`reports initiated by UE would complicate the uplink resource allocation considerably, leading
`to increased signaling overhead. Hence it was agreed that in the LTE standardization channel
`state feedback reporting is always fully controlled by the eNodeB, i.e. the UE cannot send
`any channel state feedback reports without eNodeB knowing beforehand.
`To fully exploit the gains from frequency selective packet scheduling, detailed CSI reporting
`is required. As the number of UEs reporting channel state feedback increases, however, the
`uplink signaling overhead becomes signifi cant. Furthermore the PUCCH, which is supposed
`to carry primarily the control information, is rather limited in capacity: payload sizes of only
`up to 11 bits/subframe can be supported. On the PUSCH there are no similar restrictions on
`the payload size, but since PUSCH is a dedicated resource only one user can be scheduled on
`a single part of the spectrum.
`To optimize the usage of the uplink resources while also allowing for detailed frequency
`selective CSI reports, a two-way channel state feedback reporting scheme has been adopted in
`LTE. Two main types of reports are supported: Periodic and Aperiodic. A comparison of the
`main features of the two reporting options is presented in Table 5.4.
`Periodic reporting using PUCCH is the baseline mode for channel information feedback
`reporting. The eNodeB confi gures the periodicity parameters and the PUCCH resources via
`higher layer signaling. The size of a single report is limited up to about 11 bits depending on
`the reporting mode, and the reports contain little or no information about the frequency domain
`behavior of the propagation channel. Periodic reports are normally transmitted on PUCCH. If
`the UE is scheduled in the uplink, however, the Periodic report moves to PUSCH. The reporting
`period of RI is a multiple of CQI/PMI reporting periodicity. RI reports use the same PUCCH
`resource (PRB, Cyclic shift) as the CQI/PMI reports – PUCCH format 2/2a/2b.
`When the eNodeB needs more precise channel state feedback information it can at any time
`request the UE to send an Aperiodic channel state feedback report on PUSCH. Aperiodic reports
`can be either piggybacked with data or sent alone on PUSCH. Using the PUSCH makes it pos-
`sible to transmit large and detailed reports. When the transmission of Periodic and Aperiodic
`reports from the same UE might collide, only the Aperiodic report is sent.
`
`Samsung Ex. 1010
`261 of 1365
`
`

`

`Physical Layer
`
`127
`
`Table 5.4 Comparison of Periodic and Aperiodic channel information feedback reporting
`
`
`
`Periodic reporting
`
`Aperiodic reporting
`
`When to send
`Where to send
`
`Periodically every 2–160 ms
`Normally on PUCCH, PUSCH used
`when multiplexed with data
`Payload size of the reports 4–11 bits
`Channel Coding
`Linear block codes
`CRC protection
`No
`Rank Indicator
`Sent in separate subframes at lower
`periodicity
`Only very limited amount of
`frequency information
`Only wideband PMI
`
`Frequency selectivity of
`the CQI
`Frequency selectivity of
`the PMI
`
`When requested by the eNodeB
`Always on PUSCH
`
`Up to 64 bits
`Tail biting convolutional codes
`Yes, 8 bit CRC
`Sent separately encoded in the same
`subframe
`Detailed frequency selective reports
`are possible
`Frequency selective PMI reports are
`possible
`
`The two modes can also be used to complement each other. The UE can be, for example,
`confi gured to send Aperiodic reports only when it is scheduled, while Periodic reports can
`provide coarse channel information on a regular basis.
`
`5.9.6.3 CQI Compression Schemes
`
`Compared to the WCDMA/HSPA, the main new feature in the channel feedback is the
`frequency selectivity of the report. This is an enabler for the Frequency Domain Packet
`Scheduling (FDPS). Since providing a full 4-bit CQI for all the PRBs would mean excessive
`uplink signaling overhead of hundreds of bits per subframe, some feedback compression
`schemes are used.
`To reduce feedback, the CQI is reporter per sub-band basis. The size of the sub-bands varies
`depending on the reporting mode and system bandwidth from two consecutive PRBs up to
`whole system bandwidth.
`The main CQI compression methods are:
`• wideband feedback
`• Best-M average (UE-selected sub-band feedback)
`• higher layer-confi gured sub-band feedback.
`Additionally, delta compression can be used in combination with the above options, e.g.
`when a closed loop MIMO CQI for the 2nd codeword can be signaled as a 3-bit delta relative
`to the CQI of the 1st codeword. When the number of sub-bands is large this leads to consider-
`able savings in signaling overhead.
`
`5.9.6.4 Wideband Feedback
`
`The simplest way to reduce the number of CQI bits is to use only wideband feedback. In
`wideband feedback only a single CQI value is fed back for the whole system band. Since
`
`Samsung Ex. 1010
`262 of 1365
`
`

`

`128
`
`LTE for UMTS — OFDMA and SC-FDMABased Radio Access
`
`no information about the frequency domain behavior of the channel is included, wideband
`feedback cannot be used in FDPS. Often, however, this is still sufficient, e.g. PDCCH link
`adaptation or TDM-like scheduling in low loaded cells do not benefit from frequency selective
`CQI reporting. Also, when the number of scheduled UEsgets high, the total uplink signaling
`overhead due to detailed CQI reports may become excessive and the wideband CQI reports
`are the only alternative.
`
`5.9.6.5 Best-M Average
`
`Best-M average is an effective compromise between the system performance and the uplink
`feedback signaling overhead. The principle of the Best-M average compression is shown in
`Figure 5.46. In Best-M average reporting the UEfirst estimates the channel quality for each
`sub-band. Then it selects the M best ones and reports back to the eNodeBasingle average
`CQI corresponding to the MCS/TBSthe UE could receive correctly assuming that the eNodeB
`schedules the UE on those M sub-bands. The parameter M depends on the system bandwidth
`and corresponds to roughly 20% of the whole system bandwidth.
`
`5.9.6.6 Higher Layer-configured Sub-band Feedback
`
`In higher layer-configured sub-band feedback a separate CQIis reported for each sub-band using
`delta compression. This will result in the best performance at the cost of feedback overhead:
`the payload size of the reports can be as large 64 bits. To keep the signaling on a manageable
`level, the sub-bandsizes with Full Feedback reporting are twice as large as with Best-M aver-
`age. This will limit the accuracy and performancein very frequency selective channels, where
`Best-M average maybe a better alternative.
`
`M =3best Subbands are selected and an average CQI value is reported PRBindex tTATAT ToT7 Ts] fio fe ff 3 [4J15|
`
`Figure 5.46 Theprinciple of the Best-M average compression. An average CQI value is reported for
`the M best sub-bands
`
`Samsung Ex. 1010
`263 of 1365
`
`

`

`Physical Layer
`
`129
`
`5.9.7 Multiple Input Multiple Output (MIMO) Antenna Technology
`
`In the Release 8 LTE specifi