`
`Performance 245
`
`
`
` bps/Hz/cell
`
`
`
`HSPA R6 HSPA R6 + HSPA R7 HSPA R8/R9
`
`LTE R8
`
`UE equalizer MIMO
`
`DC-HSDPA
`
`Figure 9.25
`
`Spectral efficiency of HSPA and LTE
`
`applications are, for example, voice, real time gaming and other interactive applications. The
`latency can be measured by the time it takes for a small IP packet to travel from the terminal
`through the network to the internet server. and back. That measure is called round trip time
`and is illustrated in Figure 9.26.
`The end—to—end delay budget is calculated in Table 9.21 and illustrated in Figure 9.27. The
`l-ms frame size allows a very low transmission time. On average, the packet needs to wait for
`0.5 ms for the start of the next frame. The retransmissions take 8ms at best and the assumed
`
`retransmission probability is 10%. The average delay for sending the scheduling request is
`
`UE
`
`eNodeB
`
`SAE GW
`
`Server
`
`Figure 9.26 Round trip time measurement
`
`Table 9.21 Latency components
`
`Delay component
`
`Delay value
`
`Transmission time uplink + downlink
`Buffering time (0.5 x transmission time)
`Retransmissions 10%
`
`2 ms
`2 x 0.5 x 1 ms = 1 ms
`2 x 0.1 x 8 ms = 1.6 ms
`
`Uplink scheduling request
`Uplink scheduling grant
`UE delay estimated
`eNodeB delay estimated
`Core network
`
`0.5 x 5 ms = 2.5 ms
`4 ms
`4 ms
`4ms
`1 ms
`
`Total delay with pre-allocated resources
`Total delay with scheduling
`
`13.6 ms
`20.1 ms
`
`Samsung Ex. 1010
`378 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 387
`
`
`
`246
`
`LTE for UMTS — OFDMA and SC—FDMA Based Radio Access
`
`
`
`U Core
`I BTS
`
`Transmission time
`
`I Uplink scheduling grant
`El Uplink scheduling request
`El Retransmissions
`
`I Buffering time
`
`LTE round trip time
`
`Figure 9.27 End-to-end round trip time including scheduling latency
`
`2.5 ms and the scheduling grant 4ms. We further assume a UE processing delay of 4ms, an
`eNodeB processing delay of 4ms and a core network delay of 1 ms.
`The average round trip including retransmission can be clearly below 15ms if there are
`pre—allocated resources. If the scheduling delay is included, the delay round trip time will be
`approximately 20 ms. These round trip time values are low enough even for applications with
`very tough delay requirements. The practical round trip time in the field may be higher if the
`transport delay is longer, or if the server is far away from the core network. Often the end-to-
`end round trip time can be dominated by non-radio delays, e.g. by the distance and by the other
`elements in the intemet. The propagation time of 5000 km is more than 20 ms.
`
`9.8 LTE Refarming to GSM Spectrum
`
`LTE will be deployed in the existing GSM spectrum like 900MHz or 1800Ml-Iz. The flex-
`ible LTE bandwidth makes refarming easier than with WCDMA because LTE can start with
`1.4MHz or 3.0 MHZ bandwidths and then grow later when the GSM traffic has decreased. The
`required separation of the LTE carrier to the closest GSM carrier is shown in Table 9.22. The
`required total spectrum for LTE can be calculated based on the can’ier spacing. The coordinated
`
`Table 9.22 Spectrum requirements for LTE refarming
`
`5MHz LTE (25 RBs)
`3MHz LTE (125 RBs)
`1.4 MHz LTE (6 RES)
`
`LTE total spectrum requirement
`LTE—GSM carrier spacing
`Coordinated
`Uncoordinated Coordinated
`Uncoordinated
`2.5 MHz
`2.7 MHz
`4.8 MHz
`5.2 MHz
`1.6MHz
`1.7 MHz
`3.0MHz
`3.2 MHz
`0.8 MHz
`0.9 MHz
`1.4MHz
`1.6MHz
`
`Samsung Ex. 1010
`379 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 388
`
`
`
`Performance
`
`247
`
`LTE-GSM carrier
`spacing 2.5 MHz
`
`LTE total spectrum requirement
`4.8 MHz
`
`
`
`Figure 9.28 LTE 5-MHz refarming example
`
`10 MHz
`
`LTE 1.4 MHz
`
`LTE 3.0
`MHz
`
`LTE 5.0 MHz
`
`GSM
`frequencies
`
`LTE
`bandwidth
`
`49
`
`42
`
`34
`
`25
`
`0
`
`1.4 MHz
`
`3.0 MHz
`
`5.0 MHz
`
`
`
`Figure 9.29 LTE refarming to GSM spectrum
`
`case assumes that LTE and GSM use the same sites while the uncoordinated case assumes that
`different sites are used for LTE and GSM. The uncoordinated case causes larger power differ-
`ences between the systems and leads to a larger guard band requirement. The coordinated case
`values are based on the GSM UE emissions and the uncoordinated values on LTE UE block-
`ing requirements. It may be possible to push the LTE spectrum requirements down further for
`coordinated deployment depending on the GSM UE power levels and the allowed LTE uplink
`interference levels. The limiting factor is the maximum allowed interference to the PUCCH
`RBs that are located at the edge of the carrier.
`The carrier spacing defi nition is illustrated in Figure 9.28. Figure 9.29 shows the expansion
`of the LTE carrier bandwidth when the GSM traffi c decreases. Only seven GSM carriers need
`to be removed to make room for LTE 1.4 MHz and 15 GSM carriers for LTE 3.0 MHz.
`
`9.9 Dimensioning
`This section presents examples on how to convert the cell throughput values to the maximum
`number of broadband subscribers. Figure 9.30 shows two methods: a traffi c volume based
`approach and a data rate based approach. The traffi c volume based approach estimates the
`maximum traffi c volume in gigabytes that can be carried by LTE 20 MHz 1 + 1 + 1 confi gura-
`tion. The spectral effi ciency is assumed to be 1.74 bps/Hz/cell using 2 × 2 MIMO. The busy
`
`IPR2022-00457
`Apple EX1010 Page 389
`
`
`
`248
`
`LTE for UMTS – OFDMA and SC-FDMA Based Radio Access
`
`Traffic volume based dimensioning
`20 MHz x 1.74
`bps/Hz/cell
`/ 8192
`
`Convert Mbps to GBytes
`
`Cell capacity 35 Mbps
`
`3600 seconds per hour
`
`x 3600
`
`Busy hour average loading
`50%
`Busy hour carries 15% of
`daily traffic
`
`30 days per month
`
`3 sectors per site
`
`5 GB traffic per user
`
`x 50%
`
`/ 15%
`
`x 30
`
`x 3
`⇒ 4600 GB/site/month
`/ 5 GB
`
`Total
`
`920 subs/site
`
`Data rate based dimensioning
`
`Cell capacity 35 Mbps
`
`From simulations
`
`Busy hour average loading
`50%
`
`x 50%
`
`Required user data rate
`
`/1 Mbps
`
`Overbooking factor
`
`/20
`
`Average busy hour data
`rate per sub
`
`= 50 kbps
`
`3 sectors per site
`
`x 3
`
`Total
`
`1050 subs/site
`
`Figure 9.30 LTE dimensioning example for 1 + 1 + 1 at 20 MHz
`
`hour is assumed to carry 15% of the daily traffi c according to Figure 9.30 and the busy hour
`average loading is 50%. The loading depends on the targeted data rates during the busy hour:
`the higher the loading, the lower are the data rates. The maximum loading also depends on the
`applied QoS differentiation strategy: QoS differentiation pushes the loading closer to 100%
`while still maintaining the data rates for more important connections.
`The calculation shows that the total site throughput per month is 4600 GB. To offer 5 GB
`data for every subscriber per month, the number of subscribers per site will be 920.
`Another approach assumes a target of 1 Mbps per subscriber. Since only some of the sub-
`scribers are downloading data simultaneously, we can apply an overbooking factor, for example
`20. This essentially means that the average busy hour data rate is 50 kbps per subscriber. The
`number of subscribers per site using this approach is 1050.
`The calculation illustrates that LTE has the capability to support a large number of broad-
`band data subscribers.
`Figure 9.31 illustrates the technology and spectrum limits for the traffi c growth assuming
`that HSPA and LTE use the existing GSM sites. The starting point is voice only traffi c in a GSM
`network with 12 + 12 + 12 confi guration, which corresponds to a high capacity GSM site found
`in busy urban areas. This corresponds to 12 × 8 × 0.016 = 1.5 Mbps sector throughput assuming
`that each time slot carries 16 kbps voice rate. The voice traffi c was the dominant part of the
`network traffi c before fl at rate HSDPA was launched. The data traffi c has already exceeded
`the voice traffi c in many networks in data volume. The second scenario assumes that the total
`traffi c has increased 10 times compared to the voice only case. The sector throughput would
`then be 15 Mbps, which can be carried with three HSPA carriers using a 15 MHz spectrum.
`The last scenario assumes 50 times more traffi c compared to voice only, which leads to
`75 Mbps and can be carried with two LTE carriers each of 20 MHz with a total 40 MHz of
`spectrum. The site throughput will be beyond 200 Mbps, setting corresponding requirements
`for the transport network capacity also.
`
`IPR2022-00457
`Apple EX1010 Page 390
`
`
`
`Performance
`
`249
`
`Starting point
`voice only
`
`10x scenario
`
`50x scenario
`
`Sector throughput
`
`Required spectrum
`
`Site throughput
`(3 sectors)
`
`BTS configuration
`
`4.5 Mbps
`
`Figure 9.31 Traffic growth scenarios with 10 times and 50 times more traffic
`
`LTE ZOMI-Iz
`
`HSPA ISMI-Iz
`
`LTE ZOMI-Iz
`
`GSM5MHz+HSPA5MHz
`
`[E LTE lOMHz
`
`Figure 9.32 Example of a European operator with good spectrum resources
`
`An example of the availability of the spectrum resources by a European operator within a
`few years is illustrated in Figure 9.32: GSM with 5MHz, HSPA with 201Vle and LTE with
`SOMHZ. Such an amount of spectrum would allow the traffic to increase more than 50 times
`compared to the voice only scenario. There are many operators in Asia and Latin America
`with less spectrum resources, which makes it more difficult to provide the high broadband
`wireless capacities.
`
`9.10 Capacity Management Examples from HSPA Networks
`
`In this section some of the HSDPA traffic analysis in RNC and at the cell level is shown and the
`implications discussed. It is expected that the analysis from broadband HSDPA networks will
`also be useful for the dimensioning of broadband LTE networks. All the analysis is based on the
`statistics of a single RNC and up to 200 NodeBs. The NodeBs were equipped with a 5—code round
`robin type shared HSDPA scheduler where five codes of HSDPA are shared among three cells
`and the transport solution was 2*E1 per NodeB. The maximum power for HSDPA was limited
`to 6W. This area was selected to be analyzed as in this RNC the RF capability and transport
`capability are in line with each other, i.e. the transport solution can deliver the same throughput
`as the shared 5-code HSDPA scheduler. First it is necessary to evaluate the cell level data volume
`
`Samsung Ex. 1010
`382 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 391
`
`
`
`
`
`250 LTE for UMTS — OFDMA and SC—FDMA Based Radio Access
`
`fluctuations and contributions to RNC level total data volume so that the RNC throughput capacity
`can be checked. Then the cell level data volume and throughput limits are evaluated for when
`the new throughput improving features (proportional fair scheduler, cell dedicated scheduler,
`more codes, code multiplexing and so on) for the radio interface are introduced.
`
`9.10.] Data Volume Analysis
`
`Figure 9.33 shows the data volume distribution over 24 h for the RNC on a typical working day.
`It can be seen that the single hour data volume share is a maximum of 6% from the total daily
`traffic volume and the fluctuation is just 3% to 6%. Also the hourly data volume share from
`the busy hour data volume share is 50% during the early moming hours and steadily increases
`towards the busiest hours from 8 pm to 1 am. The 3 hours from 9 pm to midnight are the busi-
`est hours during the day. The usage increases heavily after about 6 pm, which indicates that as
`the working day ends then is the time for internet usage.
`Looking into the individual cell contribution to the total RNC level data volume in Figure
`9.34, it can be seen that during the night when the data volume is low and mobility is low, the
`traffic is also heavily concentrated on certain areas (cells) and during the day the share of cells
`contributing to the total data volume also increases.
`As can be seen from Figure 9.34 during the early morning hours, 10% of the cells con-
`tribute 70—85% of the total RNC level data volume whereas during the busiest hours the
`same 70—85% data volume is contributed by 19—25% of the cells. During the early moming
`hours the data volume is very concentrated on just a couple of cells, which means that cells
`covering the residential areas should have very high data volume during the night time and
`early morning and due to low mobility the channel reservation times should be extremely
`long. This is shown in Figure 9.35, which indicates that during the early morning hours the
`
`
`
`
`
`DataVolumePerHourFromMax
`
`|
`
`100%
`95%
`90%
`85%
`80%
`75%
`70%
`65%
`60%
`55%
`50%
`45%
`40%
`35%
`30%
`25%
`20%
`15%
`10%
`5%
`0%
`
`I
`
`10%
`
`9%
`
`5%
`
`4%
`
`3%
`
`2%
`
`1%
`
`0%
`
`8888888888888888838888 .
`OFNNVIO‘DFQUJOFNQVIDIDFQUJ
`a
`
`Figure 9.33 Daily RNC level hourly data volume deviation
`
`
`
`DataVolumeShara%FromDayTrafficVqumo
`
`
`
`
`
`Samsung Ex. 1010
`383 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 392
`
`
`
`251
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Performance
`
`1 00%
`
`95%
`
`%%%%%%%%5050505088776655
`
`3.8
`
`
`
`hoanEu38.xanSana—.2502m.53.oScam
`
`.$m
`
`$0
`
`$5
`
`$9
`
`$3
`
`o\aNN
`
`Qanw
`
`o\..5
`
`38
`
`$8
`$3
`o\oov
`.xbv
`$3.
`Amount of Cells
`
`"Kano
`
`{one
`
`.x.I.
`
`icon
`
`$om
`
`£33
`
`£60
`
`..\omm
`
`o\oma
`
`Figure 9.34 Cells’ data volume contribution to total RNC data volume
`
`
`
`100%
`
`95%
`
`90%
`
`
`75% 70%
`
` 0:00*0"50
`
`80%
`
`mvxvo
`
`mvavow
`
`ouvxvou
`
`8&98
`
`nvvxvov
`
`mmvxvon
`
`mavxvow
`
`mhvaON
`
`mwvxvom
`
`mavxvom
`
`ONwvxvoov
`
`8va03
`
`Swvxvowp
`
`OVNVXVONN
`
`8~vxvocu
`
`oumvxvoom
`
`convxvovn
`
`8vvxvown
`
`ovvvxvoNv
`
`omvvxvomv
`
`Sovxvoom
`
`8wvxvoou
`
`Suwnx
`
`Sowvxvooa
`
`HS-DSCH Allocation Duration per HS-DSCH Allocation [s]
`
`Figure 9.35 Average HS
`
`DSCH allocation duration CDF of cells
`
`Samsung Ex. 1010
`384 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 393
`
`
`
`252
`
`LTE for UMTS – OFDMA and SC-FDMA Based Radio Access
`
`average HS-DSCH allocation duration under the cell is a lot longer than during the highest
`usage. The lack of mobility and potentially some fi le sharing dominating usage during night
`time means that certain cells have extremely high data volumes at certain hours only and some
`cells have a fairly high usage and low data volume variation throughout the whole day. This
`makes effi cient cell dimensioning a challenging task as if capacity is given to a cell according
`to the busy hour needs, then some cells are totally empty during most times of the day.
`In Figure 9.36, the data volume fl uctuation per hour is shown as the cumulative distribution
`of cells having a certain percentage of data volume from the highest data volume during the
`specifi c hour. From the graphs it can be seen that during the early morning and low data volume
`times there is also the highest number of cells having the lowest data volume. This indicates a
`very high fl uctuation of data volume between all the cells.
`Figure 9.37 shows the busy hour share of the total daily traffi c on the cell level. The cells
`are sorted according to the data volume: the cells on the left have the highest traffi c volume.
`The busy hour carries 10–20% of the daily traffi c in busy cells. Figure 9.38 shows the dis-
`tribution of the cells depending on the percentage of the data carried by the busy hour. The
`most typical values are 10–15% and 15–20%. We need to note that the busy hour share on
`the RNC level was only 6%. The difference between the cell and RNC level traffi c distribu-
`tion is explained by the trunking gain provided by RNC since the traffi c is averaged over
`hundreds of cell.
`
`9.10.2 Cell Performance Analysis
`The cell performance analysis is carried out using the key indicators below:
`
` Active HS-DSCH Throughput – typically given in this analysis as kbps, it is the throughput
`under a cell when data have been sent in TTIs. Put simply it is the amount of data (kbit)/
`number of active TTIs (s) averaged over a 1 h measurement period.
` HSDPA Data Volume – typically given in this analysis as Mbyte, kbit or Mbit, it is the
`amount of data sent per cell during the 1 h measurement period.
` Average number of simultaneous HSDPA users, during HSDPA usage – the amount of
`simultaneous users during the active TTIs, i.e. how many users are being scheduled during
`active TTIs per cell (the maximum amount depends on operator purchased feature set).
`When taking the used application into account, the average number of simultaneous users
`during HSDPA usage needs to be replaced by the number of active users who have data in
`the NodeB buffers. Averaged over the 1 h measurement period.
` Allocated TTI share – the average number of active TTIs during the measurement period
`(i.e. when there are data to send, the TTI is active) over all the possible TTIs during the 1 h
`measurement period.
` Average Throughput per User – typically given as kbps, it is the Active HS-DSCH Throughput/
`Average number of simultaneous HSDPA users adjusted by the allocated TTI share; it is the
`average throughput that one single user experiences under a certain cell. When taking into
`account the used application, the Active HS-DSCH throughput needs to be divided by the
`number of active users who have data in the NodeB buffer. Averaged over the 1 h measure-
`ment period.
` Average reported Channel Quality Indicator (CQI) – every UE with HS-DSCH allocated
`measures and reports the CQI value back to the BTS. The average reported CQI is the
`
`(cid:129)
`(cid:129)
`(cid:129)
`(cid:129)
`(cid:129)
`(cid:129)
`IPR2022-00457
`Apple EX1010 Page 394
`
`
`
`
`
`Performance 253
`
`' ' ' 3:00 #4:“! —.—5:00 —I—6:00 —I—7fl0
`1:00 — ' 2:00
`
`
`9:00
`10:00
`11:00
`12:00
`I II 13:00
`14:!”
`I 0- 15:00
`
`
`
`
`22:00 +230020:00
`- t - 17:00
`
`
`
`*
`!.-:‘:*:-—.--
`-
`- ‘
`
`
`
`
`
`
`
` CDFofCells
`
`
`
`60%
`
`55%
`
`'
`
`X
`if
`6
`
`a?
`A
`9
`76
`g
`
`a?
`A
`0
`1’6
`§
`
`A
`§
`16
`a?
`9
`
`A
`5
`76
`a?
`m
`
`3'3
`32
`a?
`a?
`a?
`A
`A
`A
`A
`A
`A
`A
`A
`A
`A
`0
`§
`0
`§
`12
`§
`0
`E
`0
`§
`1’6
`1’6
`76
`)6
`1’6
`16
`1’6
`)6
`1’6
`1’6
`a?
`a?
`a?
`a?
`3?
`g
`3
`§
`3
`§
`2
`§
`0
`§
`0
`% Data Volume from Highest Data Volume Cell
`
`a?
`A
`:13
`1’6
`§
`
`A
`5
`1’6
`a?
`2
`
`A
`§
`76
`a?
`0
`
`a?
`A
`:5;
`1’6
`§
`
`3‘3
`F
`8
`{2
`A
`§
`
`Figure 9.36 Percentage data volume per cell from highest data volume cell per hour
`
`
`— Daily Data Volume per Cell
`—BH Share
`
`100%
`
`90%
`
`80%
`
`E
`70% if
`§
`>
`60% q
`3
`z-
`
`50% g
`E
`40% g
`g
`30% g
`20%
`
`10%
`
`0%
`
`Samsung Ex. 1010
`386 of 1365
`
`Cells
`
`Figure 9.37 Busy hour share of cell level data volume
`
`100%
`95%
`90%
`35%
`30%
`
`g 75%
`._
`E
`70%
`3 65%
`‘5
`
`g 60%(I)
`i 55%
`o
`
`i 50%
`g
`45%
`7’,
`40%
`35%
`3‘
`30%
`g 25%20%
`15%
`
`> g
`
`10% ,
`5%
`0%
`
`IPR2022-00457
`Apple EX1010 Page 395
`
`
`
`254
`
`24%
`
`22%
`
`20%
`18%
`
`16%
`
`g14%
`312%
`§10%
`
`8%
`
`6%4%
`
`2%
`
`LTE for UMTS — OFDMA and SC—FDMA Based Radio Access
`
`
`
`I
`I
`I
`
`I
`I
`as
`at
`as
`at
`as
`a?
`as
`as
`a?
`at
`g§2§a§3§9§3§8§3§8§8§
`gA
`a
`g.
`9‘
`3
`a
`g.
`a
`g,
`Q
`g
`s:
`a:
`g:
`a:
`g
`g
`9,
`g
`g
`§§2§§§§§9§2§3§n§8§§
`BusyHourShamofCells
`
`Figure 9.38 CDF and PDF of data volume busy hour share
`
`100%
`
`90%
`
`80%
`
`70%
`
`60%%
`50%§
`40%8
`
`30%
`
`20%
`
`10%
`
`CQI value reported by all the UEs under the cell averaged over the 1 h measurement
`period.
`
`The end user throughput depends then on the average active HS-DSCH throughput and the
`average number of simultaneous HSDPA users. Figure 9.39 shows the average throughput as a
`function of the average number of simultaneous users. The average active HS—DSCH through-
`put is approximately 1.2 Mbps. When the allocated TTI share is low, the end user throughput
`approaches the HS-DSCH throughput. When the allocated TTI share increases, the end user
`throughput is reduced. This calculation is, however, pessimistic since it assumes that all users
`are downloading all the time.
`When the used application activity is taken into account, i.e. the active HS-DSCH throughput
`is divided by the number of users who have data in the NodeB buffer, the average throughput
`per user is quite different from the formula used above. Figure 9.40 shows the comparison of
`the two different average throughput per user formulas and the Active HS—DSCH throughput.
`It should be noted that Figure 9.39 and Figure 9.40 are not from exactly the same time. The
`end user throughput is 400—800 kbps when only those users are considered that have data
`in the buffer. This can be explained when analyzing the difference between the number of
`simultaneous users and the number of simultaneous users who have data in the NodeB buffer
`
`as shown in Figure 9.41.
`Figure 9.4] shows that the used application plays a significant role in the end user throughput
`evaluation and it should not be ignored. Therefore, the average user throughput may be low
`because the application simply does not need a high average data rate. The network performance
`analysis needs to separate the data rate limitations caused by the radio network and the actual
`used application data rate.
`
`Samsung Ex. 1010
`387 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 396
`
`
`
`Performance
`
`255
`
`+Active HS—DSCH throughput [kbps]
`
`Average Throughput per User [kbps] - .- Allocated TTI Share'100
`
`1400
`
`1300
`
`
`
`
`
`AverageThroughputperHSDPAUser
`
`
`
`[kbps],AverageActiveHS-DSCH
`
`
`
`
`
`1200
`
`1100
`
`1000
`
`97h010'1Non888888§
`
`
`
`Throughput[kbps]
`
`an088
`
`14
`
`13
`
`12
`
`11
`
`10
`
`9
`
`8
`
`7
`
`6
`
`5
`
`4
`
`3
`
`
`
`
`
`AllocatedTTIShare'100
`
`Average Number of Simultaneous HSDPA Users
`
`Figure 9.39 Average throughput per user
`
`
`- o - Average Throughput per User [kbps]
`Average Throughput per User (users with data in BTS buffer) [kbps]
`—~—Active HS-DSCH Throughput [kbps]
`
`
`
`
`
`
`
`§§§§
`
`O
`
`Figure 9.40 Average throughput per user, comparisons of different formulas
`
`Time [hour]
`
`Samsung Ex. 1010
`388 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 397
`
`
`
`256
`
`LTE for UMTS – OFDMA and SC-FDMA Based Radio Access
`
`Number of Simultaneous Users
`
`Number of Simultaneous Users, Data in BTS buffer
`
`22
`
`20
`
`18
`
`16
`
`14
`
`12
`
`10
`
`02468
`
`Number of Simultaneous Users
`
`Figure 9.41 Number of simultaneous users with and without taking the used application into account
`
`Time [hour]
`
`9.11 Summary
`3GPP LTE Release 8 enables a peak bit rate of 150 Mbps in downlink and 50 Mbps in uplink
`with 2 × 2 MIMO antenna confi guration in downlink and 16QAM modulation in uplink. The
`bit rates can be pushed to 300 Mbps in downlink with 4 × 4 MIMO and 75 Mbps in uplink with
`64QAM. It is expected that the fi rst LTE deployments will provide bit rates up to 150 Mbps.
`The LTE link level performance can be modeled with the theoretical Shannon limit when
`suitable correction factors are included. LTE link level performance was shown to be robust
`with high mobile speeds, and the uplink LTE performance can be optimized by using adaptive
`transmission bandwidth. The link level simulations are used in the link budget calculations,
`which indicate that the LTE link budget is similar to the HSPA link budget with the same data
`rate and same spectrum.
`The LTE system performance is optimized by orthogonal transmission schemes, by MIMO
`transmission and by frequency domain scheduling. The spectral effi ciency can be further
`enhanced with multi-antenna transmission and higher order sectorization. The high effi ciency
`can be maintained for different LTE bandwidths between 5 and 20 MHz, while the spectral
`effi ciency is slightly lower with the narrowband LTE carriers 1.4 and 3.0 MHz. It was shown
`that LTE provides higher spectral effi ciency compared to HSPA and HSPA evolution especially
`due to the frequency domain packet scheduling.
`The user plane latency in LTE can be as low as 10–20 ms. The low latency is relevant for
`improving the end user performance since many applications and protocols benefi t from low
`latency. The low latency is enabled by the short sub-frame size of 1 ms.
`
`IPR2022-00457
`Apple EX1010 Page 398
`
`
`
`Performance
`
`257
`
`The fl exible refarming of LTE to the GSM spectrum is enabled by the narrowband options:
`the refarming can be started with 1.4 or 3.0 MHz and later expanded when the GSM traffi c has
`decreased. All UEs need to support all bandwidths between 1.4 and 20 MHz.
`The dimensioning of the broadband wireless networks is different from voice networks.
`The examples from HSPA networks illustrate that the traffi c distribution over the day, over the
`geographical area and the user mobility need to be considered.
`
`References
` [1] 3GPP Technical Specifi cation 36.213 ‘Physical layer procedures’, v. 8.3.0.
` [2] 3GPP Technical Specifi cation 36.306 v8.2.0: ‘User Equipment (UE) radio access capabilities’, August 2008.
` [3] P. Mogensen et al. ‘LTE Capacity compared to the Shannon Bound’, IEEE Proc. Vehicular Technology Conference,
`pp. 699–703, April 2007.
` [4] 3GPP Technical Specifi cation 25.942 ‘Radio Frequency (RF) system scenarios’, v. 7.0.0.
` [5] 3GPP Technical Report 25.996 ‘Spatial channel model for Multiple Input Multiple Output (MIMO) simulations’,
`V.7.0.0.
` [6] I.Z. Kovács et al , ‘Effects of Non-Ideal Channel Feedback on Dual-Stream MIMO OFDMA System Performance’,
`IEEE Proc. Veh. Technol. Conf., Oct. 2007
` [7] I.Z. Kovács et al., ‘Performance of MIMO Aware RRM in Downlink OFDMA’, IEEE Proc. Veh. Technol. Conf.,
`pp. 1171–1175, May 2008.
` [8] S. Kumar, et al., ‘Performance Evaluation of 6-Sector-Site Deployment for Downlink UTRAN Long Term
`Evolution’, IEEE Proc. Vehicular Technology Conference, September 2008.
` [9] B. Hagerman, D. Imbeni, J. Barta, A. Pollard, R. Wohlmuth, P. Cosimini, ‘WCDMA 6 Sector Deployment-Case
`study of a real installed UMTS-FDD Network’, Vehicular Technology Conference, Vol. 2, pp. 703–707, Spring
`2006.
`[10] K.I. Pedersen, P.E. Mogensen, B. Fleury, ‘Spatial Channel Characteristics in Outdoor Environments and Their
`Impact on BS Antenna System Performance’, IEEE Proc. Vehicular Technology Conference, pp. 719–723, May
`1998.
`[11] 3GPP TSG RAN R1-072444 ‘Summary of Downlink Performance Evaluation’, May 2007.
`[12] 3GPP TSG RAN R1-072261 ‘LTE Performance Evaluation – Uplink Summary’, May 2007.
`
`IPR2022-00457
`Apple EX1010 Page 399
`
`
`
`10
`Voice over IP (VoIP)
`
`Harri Holma, Juha Kallio, Markku Kuusela, Petteri Lundén, Esa Malkamäki,
`Jussi Ojala and Haiming Wang
`
`10.1 Introduction
`While the data traffi c and the data revenues are increasing, the voice service still makes the
`majority of operators’ revenue. Therefore, LTE is designed to support not only data services
`effi ciently, but also good quality voice service with high effi ciency. As LTE radio only supports
`packet services, the voice service will also be Voice over IP (VoIP), not Circuit Switched (CS)
`voice. This chapter presents the LTE voice solution including voice delay, system performance,
`coverage and inter-working with the CS networks. First, the general requirements for voice and
`the typical voice codecs are introduced. The LTE voice delay budget calculation is presented.
`The packet scheduling options are presented and the resulting system capacities are discussed.
`The voice uplink coverage challenges and solutions are also presented. Finally, the LTE VoIP
`inter-working with the existing CS networks is presented.
`
`10.2 VoIP Codecs
`GSM networks started with Full rate (FR) speech codec and evolved to Enhanced Full Rate
`(EFR). The Adaptive Multi-Rate (AMR) codec was added to 3GPP Release 98 for GSM to
`enable codec rate adaptation to the radio conditions. AMR data rates range from 4.75 kbps to
`12.2 kbps. The highest AMR rate is equal to the EFR. AMR uses a sampling rate of 8 kHz,
`which provides 300–3400 Hz audio bandwidth. The same AMR codec was included also for
`WCDMA in Release 99 and is also used for running the voice service on top of HSPA. The
`AMR codec can also be used in LTE.
`The AMR-Wideband (AMR-WB) codec was added to 3GPP Release 5. AMR-WB uses a
`sampling rate of 16 kHz, which provides 50–7000 Hz audio bandwidth and substantially better
`voice quality and mean opinion score (MOS). As the sampling rate of AMR-WB is double the
`sampling rate of AMR, AMR is often referred to as AMR-NB (narrowband). AMR-WB data
`rates range from 6.6 kbps to 23.85 kbps. The typical rate is 12.65 kbps, which is similar to the
`
`LTE for UMTS: OFDMA and SC-FDMA Based Radio Access Edited by Harri Holma and Antti Toskala
`© 2009 John Wiley & Sons, Ltd. ISBN: 978-0-470-99401-6
`
`IPR2022-00457
`Apple EX1010 Page 400
`
`
`
`260
`
`LTE for UMTS - OFDMA and SC-FDMA Based Radio Access
`
`normal AMR of 12.2 kbps. AMR-WB offers clearly better voice quality than AMR-NB with the
`same data rate and can be called wideband audio with narrowband radio transmission. The radio
`
`bandwidth is illustrated in Figure 10.1 and the audio bandwidth in Figure 10.2. The smallest bit
`rates, 1.8 and 1.75 kbps, are used for the transmission of Silence Indicator Frames (SID).
`This chapter considers AMR codec rates of 12.2, 7.95 and 5.9 kbps. The resulting capacity
`of 12.2kbps would also be approximately valid for AMR—WB 12.65 kbps.
`When calling outside mobile networks, voice transcoding is typically required to 64 kbps
`Pulse Code Modulation (PCM) in links using ITU G.711 coding. For UE-to-UE calls, the
`transcoding can be omitted with transcoder free or tandem free operation [1]. Transcoding
`generally degrades the voice quality and is not desirable within the network.
`
`AMR—NB
`
`AMR-\VB
`
`23.35 kbps
`
`19.85 kbps
`
`
`Figure 10.1 Adaptive Multirate (AMR) Voice Codec radio bandwidth
`
`Human ear
`20-20000 Hz
`
`
`Wideband AIvIR
`50—7000 Hz
`
`
`
`
`Narrowband
`
`AMR
`300-3400 Hz
`
`Figure 10.2 Adaptive Multirate (AMR) Voice Codec audio bandwidth
`
`Samsung Ex. 1010
`392 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 401
`
`
`
`Voice over IP (VoIP)
`
`261
`
`There are also a number of other voice codecs that are generally used for VoIP. A few
`examples are G.729, using an 8kbps coding rate, and Internet Low Bit Rate Codec (iLBC),
`using 13 kbps which is used, for example, in Skype and in Googletalk.
`
`10.3 VoIP Requirements
`
`There is a high requirement set for the radio network to provide a reliable and good quality
`voice service. Some of the main requirements are considered below.
`The impact of the mouth—to-ear latency on user satisfaction is illustrated in Figure 10.3. The
`delay preferably should be below 200 ms, which is similar to the delay in GSM or WCDMA
`voice calls. The maximum allowed delay for a satisfactory voice service is 280 ms.
`1P Multimedia Subsystem (IMS) can be deployed to control VoIP. IMS is described in Chapter
`3. IMS provides the information about the required Quality of Service (QoS) to the radio network
`by using 3GPP standardized Policy and Charging Control (PCC) [3]. The radio network must
`be able to have the algorithms to offer the required QoS better than Best Effort. QoS includes
`mainly delay, error rate and bandwidth requirements. Q08 in LTE is described in Chapter 8.
`The voice call drop rates are very low in the optimized GSM/WCDMA networks today — in
`the best case below 0.3%. VoIP in LTE must offer similar retainability including smooth inter—
`working between GSM/WCDMA circuit switched (CS) voice calls. The handover functionality
`from VoIP in LTE to GSM/WCDMA CS voice is called Single radio Voice Call Continuity
`(SR-VCC) and described in detail in section 10.10.
`The AMR 12.2kbps packet size is 31 bytes while the [P header is 40—60 bytes. [P header
`compression is a mandatory requirement for an efficient VoIP solution. IP header compres—
`
`
`
`
`
`
`
`
`
`
`
`||
`
`
`
`||||||||||||||||
`
`
`
`\
`
`:5 ¢///¢W%¢W/%a
`W//////<<%//////m
`
`
`
`
`
`
`Ln;
`
`3:21.233?
`
`“$7
`
`LIIO
`
`\\
`///// /// ////////
`\
`.— OO
`
`N8
`
`300
`§
`Mouth to ear delay (ms)
`
`§
`
`
`
`$5?
`
` Nearly all
`
`/ users
`///
`A
`
`
`
`
`
`A
`G.l l4_F01
`
`Figure 10.3 Voice mouth-to-ear delay requirements [2]
`
`Samsung Ex. 1010
`393 of 1365
`
`IPR2022-00457
`Apple EX1010 Page 402
`
`
`
`262
`
`LTE for UMTS – OFDMA and SC-FDMA Based Radio Access
`
`sion is required both in UE and in eNodeB. All the VoIP simulations in this chapter assume
`IP header compression.
`The IP connectivity requires keep alive messages when the UE does not have a phone call run-
`ning. The frequency of the keep alive messages depends on the VoIP solution: operator IMS VoIP
`can use fairly infrequent keep alive messages since IMS is within the operator’s own network and
`no fi rewalls or Network Address Tables (NAT) are required in between. The internet VoIP requires
`very frequent keep alive message to keep the connectivity open through fi rewalls and NATs. The
`frequent keep alive message can affect UE power consumption and network effi ciency.
`VoIP roaming cases need further attention especially if there are some LTE networks designed
`for VoIP and data, while some networks are designed for data only transmission without the
`required voice features. VoIP roaming also requires IMS and GPRS roaming agreements and the
`use of visited GGSN/MME model. One option is to use circuit switched (CS) calls whenever
`roaming with CS Fallback for LTE procedures. Similarly CS calls can also be used for emer-
`gency calls since 3GPP Release 8 LTE specifi cations do not provide the priority information
`from the radio to the core network nor a specifi c emergency bearer.
`
`10.4 Delay Budget
`The end-to-end delay budget for LTE VoIP is considered here. The delay should preferably be
`below 200 ms, which is the value typically achieved in the CS network today. We use the fol-
`lowing assumptions in the delay budget calculations. The voice encoding delay is assumed to
`be 30 ms including a 20 ms frame size, 5 ms look-ahead and 5 ms processing time. The receiv-
`ing end assumes a 5 ms processing time for the decoding. The capacity simulations assume a