`
`DJ
`
`LTE for UMTS — OFDMA and SC—FDMA Based 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 = 1 transmit diversity is applied similarly to transmission
`mode 2. With higher rank spatial multiplexing with up to four layers with large delay,
`CDD is 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—MIMO operation.
`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 UE specific reference signals
`are m use,
`
`5.8.5 Physical Broadcast Channel (PBCH)
`
`The physical broadcast Channel (PBCH) carries the system information needed to access the
`system, such as RACH parameters, and as covered in more detail in Chapter 6. The channel is
`always provided with 1.08 MHz bandwidth, as shown in Figure 5.36, so the PBCH structure
`is independent of the actual system bandwidth being used, similar to other channels/signals
`needed for initial system access. The PBCH is convolutionally encoded as the data rate is not
`that high. As discussed in Chapter 6. the broadcast information is partly carried on the PBCH,
`
`10 ms = 10 subframes
`
`
`
`Synchronization
`Signals
`
`
`
`
`
`Figure 5.36 PBCH location at 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 (5135) are then on the PDSCH. The 600 sub—carriers in Figure 5.36 need only 9MHZ (50
`resource blocks) in the resource domain but the system bandwidth needed for sufficient attenu—
`ation for the adjacent operator increases the total bandwidth needed to lOMHz, as discussed
`in Chapter I 1. With a 1.4 MHz system bandwidth there are no resource blocks on either side
`of the PBCH in 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 lst and 1 1th slots (slots 0 and 10) of the lOms
`frame, as shown in Figure 5.37.
`The PSS and SSS jointly 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 PC15 mean that 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 power control, HARQ, timing advance and
`random access. Timing advance is based on the signaling in the Medium Access Control
`(MAC) layer (as shown in the MAC section in Chapter 6), but as it is directly related to the
`physical layer, the timing advance details are covered in this chapter. The big contrast to
`WCDMA is that there are no physical layer issues related to macro-diversity since the UE is
`only connected to one base station at a time and hand handover is applied. Also, a specific
`means for dealing with inter-system and inter-frequency measurements such as compressed
`
` r
`
`\
`
`\
`
`v
`
`,—’
`
`‘\ #16 ms =10 subframes : 20 slots —
`1.08 MHZI
`
`\ PSS/SSS
`
`Figure 5.37 Synchronization signals in the frame
`
`Samsung Ex. 1010
`252 of 1365
`
`
`
`118
`
`LTE for UMTS — OFDMA and SC—FDMA Based Radio Access
`
`mode is not needed in LTE. as LTE by nature has a discontinuous operation that will facilitate
`the measurements by scheduling.
`
`5. 9. 1 HARQ Procedure
`
`The HARQ in LTE is based on the 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 combine the 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. Due to the stop-and-wait way of operating, one needs to have
`multiple HARQ processes to enable a continuous data flow. In LTE the number of processes
`is fixed to 8 processes in both the uplink and downlink direction. An example of a single user
`continuous transmission is illustrated in Figure 5.38. For multiple users. it is dependent on
`the eNodeB scheduler when a retransmission is sent in the uplink or downlink direction, as a
`retransmission also requires that resources are allocated.
`The HARQ operation in LTE supports both soft combining and the use of incremental redun-
`dancy, The use of soft combining means that 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 and the start of a
`retransmission is 7 ms. The UE will send the ACK/NACK for a packet in frame n, in the uplink
`frame n+4. This leaves around 3ms 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 thus the timing shown in Figure 5.39
`is the earliest moment for a retransmission to occur.
`
`From scheduler bufl'er
`
`PUSCH/PDSCH 1
`
`2
`
`3
`
`1"TX
`
`2 1
`
`4
`
`5
`
`5
`
`7
`
`8
`
`1
`
`“ TX (new packet)
`
`CRC CheckResult
`
`Fa“
`
`pass
`
`RLC layer
`
`Figure 5.38 LTE HARQ operation with 8 processes
`
`Samsung Ex. 1010
`253 of 1365
`
`
`
`Physical Layer
`
`119
`
`PDSCH
`
`PACKET
`1 ms
`
` New Packet or
`
`Retransmission
`
`PUCCH or PUSCH
`
`ACK/NAC .
`3 ms U}: processing time W 3 ms eNodeB processing time
`1 ms
`
`Figure 5.39 LTE HARQ timing for a single downlink packet
`
`PDSCH
`
`_
`
`5. 9.2 Timing Advance
`
`The timing control procedure is needed so that the uplink transmissions from different users
`arrive at the eNodeB essentially 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 adjustment is actually needed. The resolution of a timing advance
`command is 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. Such situations are
`system access, when the UE is in RRC_IDLE state or when the UE has had an inactivity period
`exceeding related timer. non-synchronized handover. and after radio link failure. Additionally.
`eNodeB can assign to UE a dedicated (contention-free) preamble on RACH for uplink timing
`measurement when eNodeB wants to establish uplink synchronization. Such situations are
`faced with handover or when downlink data arrive for a non—synchronized UE. From the range
`defined for timing advance, cell sizes up to 100km would be facilitated, and even beyond by
`leaving some resources unused.
`
`5. 9. 3 Power Control
`
`For LTE, power control is slow for the uplink direction. In the downlink direction there is no
`power control. As the bandwidth varies due to data rate changes. the absolute transmission
`power of the UE will also change. The power control does not now actually control absolute
`power but rather the Power Spectral Density (PSD), power per Hz. for a particular device. What
`facilitates the use of a slower rate for power control 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 power control is to reduce terminal power consumption and also to avoid an
`overly large dynamic range in the eNodeB receiver, rather than to mitigate interference. In the
`
`Uplink data or RACH (0‘ ’0)—
`
`— _
`Timing Advance (in x 0.52us) é_ k :—
`eNodeB
`
`UE
`
`Figure 5.40 Uplink timing control
`
`Samsung Ex. 1010
`254 of 1365
`
`
`
`120
`
`LTE for UMTS — OFDMA and SC—FDMA Based Radio Access
`
`Power per Hz unchanged
`
`Spectral
`Density
`
`2 X data
`rate
`
`TTI n
`
`Frequency
`
`Frequency
`TTI n+1
`
`Figure 5.41 LTE uplink power with 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 UE transmitter is kept under control. The LTE uplink power control
`principle is illustrated in Figure 5.41 where at the change of data rate the PSD stays constant
`but the resulting total transmission power is adjusted relative to the data rate change.
`The actual power control is based on estimating the path loss, taking into account cell specific
`parameters and then applying the (accumulated) value of the correction factor received from
`the eNodeB. Depending on the higher layer parameter settings, the power control command
`is either 1 dB up or down or then the set of [—ldB, 0, +ldB, +3 dB] is used. The specifications
`also include power control that is absolute value based but, based on the text case prioritization,
`it is not foreseen that this will be used in the first phase networks. The total dynamic range of
`power control is slightly smaller than in WCDMA and the 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 paging is to ensure a suf-
`ficient DRX cycle for devices to save power and also to ensure a fast enough response time for
`the incoming call. The E-UTRAN may parameterize 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 and similar ramping of preamble power. The initial power is based on the measured
`path loss in DL, and power ramping is necessary because of the relatively coarse accuracy of
`the UE in path loss measurement and absolute power setting, and to compensate for uplink
`fading. Although LTE PRACH resources are separate from PUSCH and PUCCH, power ramp-
`ing is useful for simultaneous detection of different preamble sequences and for minimizing the
`interference due to asynchronous PRACH transmission 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
`
`o Transmit a preamble using the PRACH resource, preamble sequence and power selected by
`MAC.
`
`0 Wait for the RACH response with matching preamble information (PRACH resource 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 RACH procedure, after acknowledging a
`preamble, the UE continues with a 10 or 20ms frame duration of data, or even longer, as
`described for Release 8 HSPA operation in Chapter 13. The fundamental difference in LTE is
`that the device will move instead directly to the use of UL-SCH on reception of the random
`access response, which has the necessary information.
`o If no matching random access response is received, transmit the preamble in the next avail—
`able PRACH resource according to instructions of MAC, as shown in Figure 5.42.
`
`Although the LTE specification models that the physical layer just transmits preambles and
`detects responses under the control of MAC, we describe below the complete procedure without
`focusing on the modeling of the specification.
`Two fundamentally different random access procedures have been specified for LTE. The
`contention based procedure is what we normally understand with random access: UEs trans—
`mit randomly selected preambles on a common resource to establish a network connection or
`request resources for uplink transmission. The non-contention based random access is initiated
`by the network for synchronizing UE’s uplink transmission, and the network can identify the
`UE from the very first uplink transmission. This procedure is nevertheless included under the
`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. The details of PRACH
`and preamble sequences are explained in section 5.7. For each cell a total of 64 sequences are
`reserved, and these are grouped for the non-contention based and contention based procedures.
`The group reserved for the contention based procedures is divided further into two: by selecting
`the proper group, the UE sends one bit of information about the transport block size that the
`UE desires to send on PUSCH in Step 3.
`In the second step, the UE receives a preamble response on DL—SCH resource that is assigned
`on PDCCH. The identity RA-RNTI that is used for this assignment is associated with the
`
`Downlink / eNodeB
`Not detected
`
`PRAC H
`On the resources
`
`response
`indicated by
`
`PRACH response
`
`
`
`resource \
`
`Uplink I U E
`
`Preamble
`
`Preamble
`
`UE specific data
`
`Figure 5.42 Power ramping in random access procedure
`
`Samsung Ex. 1010
`256 of 1365
`
`
`
`122
`
`LTE for UMTS — OFDMA and SC—FDMA Based 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. PUSCH transmission including
`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 responses that are
`meant for preambles transmitted in the same PRACH frequency and time resource, which is
`important for saving PDCCH resources. eNodeB transmits the response in a time window that
`can be configured up to lOms in duration. The flexible window allows freedom for dimension-
`ing of the RACH receiver and for scheduling of the responses.
`The response (signaling part of the MAC layer as covered in Chapter 6) lists the sequence
`numbers of the observed preambles, and in addition the following information is given for
`each acknowledged preamble:
`
`o A grant for the first transmission on PUSCH, including also information on the need for
`frequency hopping, power control command for uplink transmission and information on
`the need for CQI transmission and whether the PUSCH transmission needs to be delayed
`by one subframe from the nominal value.
`0 A timing alignment command.
`0 A temporary allocation for identity called temporary CRNTI, which is used for addressing
`PUSCH grants and DL-SCH assignments in Steps 3 and 4 of the procedure.
`
`The typical probability of preamble collisions, meaning that two or more UEs are 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 includes its identity in the
`first message that it transmits on PUSCH in Step 3 and expects in Step 4 an acknowledge-
`ment that eNodeB has received 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 CRNTI to the message of Step 3, or (2) the UE’s identity can be acknowledged with
`a message that is sent on a DL—SCH resource assigned with the temporary CRNTI. The first
`form of acknowledgement is for RC connected UEs while the second form is used when a
`UE tries to establish or re-establish RRC connection. HARQ is 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 — OFDMA and SC—FDMA Based Radio Access
`
`4. eNodeB frequency
`domain downlink
`
`
`
`Figure 5.44 Channel State Information (CSI) reporting procedure
`
`In general the channel feedback reported by the UE isjust 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 channel state feedback
`reports without eNodeB knowing it beforehand. The corresponding procedure for providing
`information about the uplink channel state is called channel sounding and it is done using the
`Sounding Reference Signals (SRS) as presented in Section 5.6.4.3.
`The main difference of the LTE channel state information feedback compared to WCDMA/
`HSDPA is 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 aims to divide the radio resources
`in the frequency domain for different users so that system performance is optimized. In Figure
`5.45 the gain from the FDPS is illustrated. As the UE speed increases, the CSI reports become
`more inaccurate and get outdated faster leading to reduced gains in high mobility.
`
`Mdeband CQI
`Best-M average CQI
`— Uncompressed CQI
`
`(Mbps)
`
`
`Averagecellthroughput
`
`3
`
`30
`
`_
`UE velocnty (km/h)
`
`300
`
`Figure 5.45 Comparison of the average cell throughputs for different CQI schemes and velocities
`
`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-FDMA Based 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 UEs gets 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 UE first estimates the channel quality for each
`sub—band. Then it selects the M best ones and reports back to the eNodeB a single average
`CQI corresponding to the MCS/TBS the 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 CQI is 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-band sizes with Full Feedback reporting are twice as large as with Best—M aver-
`age. This will limit the accuracy and performance in very frequency selective channels, where
`Best—M average may be a better alternative.
`
`M = 3 best Subbands are selected and an avera_e C0 I value is r . -rted
`
`
`
`Subbandindex
`I-————I-l-m
`
`mnnnnmm
`
`
`
`
`Figure