TD-LTE WiMAX Comparison

TD-LTE AND WIMAX TECHNOLOGY AND PERFORMANCE COMPARISON

October 2009

1 (65)

© Nokia Siemens Networks Document provided under NDA 1 Confidential

TD-LTE AND WIMAX

TECHNOLOGY AND PERFORMANCE COMPARISON


October 2009

2 (65)


The information in this document is subject to change without notice and describes only the product defined in the introduction of this documentation. This document is not an official customer document and Nokia Siemens Networks does not take responsibility for any errors or omissions in this document. This document is intended for the use of Nokia Siemens Networks customers only for the purposes of the agreement under which the document is submitted. No part of this documentation may be used, reproduced, modified or transmitted in any form or means without the prior written permission of Nokia Siemens Networks. The documentation has been prepared to be used by professional and properly trained personnel, and the customer assumes full responsibility when using it. Nokia Siemens Networks welcomes customer comments as part of the process of continuous development and improvement of the documentation. The information or statements given in this documentation concerning the suitability, capacity or performance of the mentioned hardware or software products are given “as is” and all liability arising in connection with such hardware or software products shall be defined conclusively and finally in a separate agreement between Nokia Siemens Networks and the customer. IN NO EVENT WILL Nokia Siemens Networks BE LIABLE FOR ERRORS IN THIS DOCUMENTATION OR FOR ANY DAMAGES, INCLUDING BUT NOT LIMITED TO SPECIAL, DIRECT, INDIRECT, INCIDENTAL OR CONSEQUENTIAL OR ANY LOSSES SUCH AS BUT NOT LIMITED TO LOSS OF PROFIT, REVENUE, BUSINESS INTERRUPTION, BUSINESS OPPORTUNITY OR DATA, that might arise from the use of this document or the information in it. THE CONTENTS OF THIS DOCUMENT ARE PROVIDED "AS IS". EXCEPT AS REQUIRED BY APPLICABLE MANDATORY LAW, NO WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT, ARE MADE IN RELATION TO THE ACCURACY, RELIABILITY OR CONTENTS OF THIS DOCUMENT. NOKIA SIEMENS NETWORKS RESERVES THE RIGHT TO REVISE THIS DOCUMENT OR WITHDRAW IT AT ANY TIME WITHOUT PRIOR NOTICE. This document and the product it describes are considered protected by copyrights and other intellectual property rights according to the applicable laws. The wave logo is a trademark of Nokia Siemens Networks Oy. Nokia is a registered trademark of Nokia Corporation. Siemens is a registered trademark of Siemens AG. Other product names mentioned in this document may be trademarks of their respective owners, and they are mentioned for identification purposes only. Copyright © Nokia Siemens Networks 2009. All rights reserved.


October 2009

3 (65)


1. Introduction....................................................................................... 4 2. Radio Access Network........................................................................ 5

2.1. Coverage Related Differentiators ......................................................... 5 2.2. Capacity Related Differentiators .......................................................... 8 2.3. Channel Coding Differentiators............................................................ 9 2.4. TX Diversity and Simultaneous MIMO with FDPS...................................11 2.5. Overhead Comparisons.....................................................................13

2.5.1. L2 Overhead.............................................................................14 2.5.2. DL Air Interface Resource Allocation.............................................15 2.5.3. Localized resource allocation .......................................................16 2.5.4. Channel quality reporting ...........................................................18 2.5.5. User and Control Plane Latencies .................................................19

2.6. 802.16m and LTE Release 8 Comparison .............................................21 2.6.1. MIMO and Beamforming Differentiators ........................................22

2.7. Interference Mitigation......................................................................23 2.8. Other Approaches ............................................................................25

3. Core Network ................................................................................... 27 4. Terminal specific aspects ................................................................. 30

4.1. Chipset specific aspects ....................................................................30 4.2. SIMs ..............................................................................................31 4.3. Industry Ecosystem..........................................................................31 4.4. Addressable market available to 16m compared to LTE .........................32

5. Network planning aspects................................................................ 33 5.1. Comparison of frequency reuse 1 and reuse 3......................................33

5.1.1. SINR distribution.......................................................................33 5.1.2. Cell Capacity ............................................................................35 5.1.3. Reuse 1 in TD-LTE and 16m as compared to 16e today...................37

5.2. Performance of 20 MHz channel vs. running 2x10 MHz ..........................38 6. Comparison of advanced MIMO/BF feature sets............................... 40 7. Uplink OFDMA vs SC-FDMA............................................................... 41 8. Standards Environment.................................................................... 51

8.1. Pros and cons contrast of the organizational structure, process, and funding 51 8.2. Detailed timeline of 3GPP path...........................................................51 8.3. Summary of Standards and Expected Deployment Timelines ..................57

9. Operator Migration/Co-existence..................................................... 58 9.1. Practical pros/cons of operator's optimized migration path from 16e >16m vs. 16e TD-LTE (Rel9 or 10).......................................................................58 9.2. 16m migration using backwards compatibility mode..............................58 9.3. Migration from 16e to TD-LTE as an overlay.........................................60


October 2009

4 (65)


1. Introduction This document frames a set of topics on key technology and performance pros/cons between TD-LTE 3GPP Release 8 and WiMAX 802.16e. Where applicable, evolution with 3GPP Release9 and Rel10 and IEEE 802.16m are noted.

There are 4 main topics in this document – technology, ecosystem, standards, and migration. Based on our analysis, it will become clear that there is little technical difference between 16m and LTE. Both have the potential to perform very well in wireless networks. There are more similarities than differences. When comparing 16m and LTE there were, for different reasons, trade offs made when developing the two technologies. The result, however, is relatively minor differences in overall performance – the largest difference coming from the SC-FDMA on the UL which provides better coverage, cheaper devices with lower power consumption and hence, better device battery life.

The most significant difference between the technologies is the ecosystem and the standards development. The overriding question about 16m v. TD-LTE is akin to the long time battle between 3GPP and 3GPP2. Both technology roadmaps have at one time or another leap frogged each other, but it has been the 3GPP standard that has won out in the end, mostly due to the ecosystem and holistic approach to system level standardization. In our view, the 3GPP LTE ecosystem will be vastly superior to the 16m ecosystem in terms of volumes, variety, and functionality. Additionally, the standards organization for 3GPP has a long history of successful development and deployment of wireless and mobile systems. Much of the struggles that plague the widespread commercial adoption of WiMAX, including mobility and compatibility have already been anticipated in LTE because of its 3GPP heritage. The device ecosystem for LTE will develop quickly and the second wave of LTE chipsets in the industry will include support for TD-LTE, in addition to LTE FDD.

From a business perspective, we must acknowledge the importance of the device portfolio on operator’s technology strategy. It is likely that we will find more suppliers committing to develop WiMAX/16m devices than suppliers committing to develop WiMAX/TD-LTE devices. These multi-mode devices significantly facilitate the migration of subscribers from one technology to another.

All of this points to a final conclusion that TD-LTE is the logical choice for an operator’s pre-4G technology.


October 2009

5 (65)


2. Radio Access Network

2.1. Coverage Related Differentiators Although WiMAX and LTE share a common foundation in their reliance on OFDMA there are differentiators that allow LTE to perform better in terms of coverage. The key differentiators are:

• Support of Single Carrier Frequency Division Multiple Access (SC-FDMA) • More Advanced TX Diversity

The following figure shows that the downlink performance of LTE and WiMAX are nearly identical due to their common OFDMA foundation.


October 2009

6 (65)


The following figure shows a comparison in the link level performance of WiMAX’s implementation of OFDMA and LTE’s implementation of SC-FDMA. Note that there is no change to UL OFDMA in 802.16m so this comparison is still valid.

For low bit rate Modulation and Coding Schemes (MCS), i.e. QPSK, there is virtually no difference between SC-FDMA and OFDMA. However, at higher MCS, SC-FDMA performs 0.5-1.0dB better. With Broadband Wireless’ increasing demands for higher throughput, the use of higher modulation and coding schemes is required and it is clear that SC-FDMA provides not only the same MCS levels as OFDMA, but also increased SINR performance. One of the reasons for the same or better performance of SC-FDMA performance in comparison to the OFDMA receiver is the turbo equalizer. The downside to SC-FDMA is that the MIMO receiver, which is located at the BTS, is more complicated in comparison to an OFDMA MIMO receiver. However, because the incremental computation resources that are required to realize the more complicated receiver reside on the BTS, this is generally not considered an issue. What is noted elsewhere in this response is the benefits for the hand held device that are associated with the use of SC-FDMA as the uplink access technology. Because of the lower Peak to Average Power Ratio required, battery life for the terminal is much better in comparison to devices that have to utilize OFDMA.


October 2009

7 (65)


Another key differentiator for coverage in LTE is the use of Space-Frequency Block Codes and Frequency-Shift Time Diversity TX diversity schemes on the control channels as well as the higher order TX diversity that is available. While WiMAX does utilize Cyclic Delay Diversity to enhance the coverage reach of the downlink, SFBC and FSTD provide some gains over CSTD in terms of common control performance. Finally, with respect to coverage differentiators, LTE does not suffer from the uplink limiting issues associated with WiMAX’s initial network entry SBC-REQ message. The following two figures show the coverage differences between TD-LTE and WiMAX, both with and without taking into account the impact due to the SBC-REQ message as well as the assumptions behind the simulations that were performed. It is not known at this time if the 802.16m specification has resolved this issue.

Result summary MAPL & Cell Range:TD-LTE vs. WiMAX 2300 MHz (urban)

Urban Urban (SBC-REQ)

ConclusionDelta between max. allowable pathloss values:3 dB benefit of TD-LTEwhen SBC-REQ msg. not considered0.1 km benefit of TD-LTE when SBC-REQ msg. not considered

128.9 dB

WiMAX

TD-LTE

0.56 km

124.2 dB 132.0 dB

WiMAX

TD-LTE

0.44 km132.0 dB0.66 km

Delta between max. allowable pathloss values:7.84 dB benefit of TD-LTE

when SBC-REQ msg. considered0.22 km benefit of TD-LTE

when SBC-REQ msg. considered

0.66 km


October 2009

8 (65)


Link budget assumptions

• Frame Type 2 (TDD), Normal CP• DL/UL configuration 1 (3DL, 2UL)• System overhead according to 3GPP 36.211

• Noise Figures: UE: 7 dB, eNB: 2.2dB• Diversity: 2Tx-2Rx (DL),1Tx-2Rx (UL)• Capacity calculated with 2x2 MIMO SM

LTE TDD• WiMAX TDD• DL:UL ratio 26:21• MAP overhead: 5 symbols (DL), 3 symbols • MAC Overhead: 4% (DL), 6% (UL)• Noise Figures: MS: 7 dB, BTS: 3dB• Diversity: 2Rx STC (DL and UL)• Capacity calculated with 2x2 STC

WiMAX

General and common assumptions• Outdoor macro cell layout (cloverleaf)• Operating band 2300 MHz• bandwith 10 MHz• TMA/MHA disabled• Cell edge throughput:

256 kbps (DL), 64 kbps (UL) • MCS: optimized for highest MAPL/cell range of the

limiting link (UL), DL MCS adapted to match UL MAPL

• Equipment parameters:– TMA/MHA disabled – UE/SS Tx Power 23 dBm– eNB/BTS Tx Power 44.8 dBm (30W)– Antenna Gain 18 dBi (UL), 0dBi (DL)– Feeder Loss 0.5 dB

• Cell Area Probability = 90%

• Channel model: Enhanced Pedestrian A 5 Hz• Cell load: DL 50% / UL 50%• Frequency reuse 1• Interference margin: 1dB (at 50% load)• Multiple server gain = 2dB• HARQ enabled• BLER on first transmission = 10%• Propagation model:

– COST 231 Hata 2-slope propagation model with– Penetration loss (du,u,su,r): 22, 17, 12, 10 [dB]– Clutter correction factor (du,u,su,r): 3, 0, -12, -28 [dB]– Stand. deviation outdoor (du,u,su,r): 9, 8, 8, 7[dB]– Antenna height NB (du,u,su,r): 30 m– Antenna height UE/MS: 1.5 m

• Traffic load model = ‘full buffer’

2.2. Capacity Related Differentiators There are four primary capacity differentiators for LTE in comparison to WiMAX. First is the impact due to the differences in Channel Coding. Second is more advanced TX diversity available on the downlink control channels. Thirdly, LTE has more advanced Multi-stream MIMO. Finally, LTE supports simultaneous MIMO and FDPS. While 802.16m does implement more streams of MIMO and 802.16x, in general supports MIMO and AMC and FDPS, the issues with IOT with 802.16e cast a shadow of doubt on whether the benefits of these techniques will be realized in the 802.16x ecosystem. Individually, the improvements brought by LTE provide small incremental improvements in comparison to the techniques used in WiMAX. Taken separately, they would not appear to give LTE an advantage over WiMAX. However, the sum of the improvements is enough to provide clear capacity benefits. It should be noted, though, that the performance benefits of these differentiators are not without their own tradeoffs. Where applicable, the tradeoffs are noted.


October 2009

9 (65)


2.3. Channel Coding Differentiators There are two primary improvements to Channel Coding that improve the capacity of LTE in comparison to WiMAX. Both higher maximum block size and support for incremental redundancy improve the spectral efficiency of the uplink and downlink. LTE and WiMAX utilize different turbo coding schemes. LTE utilizes parallel concatenated convolutional code with QPP interleaver while WiMAX utilizes a duo-binary circular recursive systematic convolutional code. As can be seen in the following figure, WiMAX’s duo-binary coding does provide a marginal gain in comparison to LTE’s implementation.

While a higher maximum code block length increases the coding complexity, simulations show that there is up to 0.7dB gain associated with a larger code block length. The following figure shows the differences between a code block length of 480bits and 5000 bits. For reference, the maximum code block sizes for the technologies 6144 and 480 bits for LTE and WiMAX, respectively.


October 2009

10 (65)


Note that the actual code block size and the associated turbo coding gain of LTE depends on the frequency domain scheduling rate and the channel bandwidth. It is widely known that Hybrid Automatic Repeat Request (HARQ) provides improvements to the air interface in comparison to other error control techniques. Incremental Redundancy (IR) has been specified in 3GPP since EDGE and is in widespread use in HSDPA. Gains from IR are MCS dependent, there is no gain for rates below 0.3-0.4 for example. However, the following figure shows that up to 1.6 dB gain due to IR for initial transmissions with a coding rate of 2/3 in comparison to Chase Combining.


October 2009

11 (65)


There is 1.6 dB gain for the initial

code rate of 2/3

As the figure above shows, there is a clear link layer gain for the higher modulation and coding schemes due to incremental redundancy. For BWA networks, it is increasingly important to enhance the performance of especially the higher MCS. It should be noted that IR does increase the terminal complexity with respect to Chase Combining. However, as noted above, due to it’s widespread use in HSDPA, the issues with terminal complexity have effectively been mitigated. Also worth noting is that the 802.16x standard does specify incremental redundancy, but it has yet to be made part of the WiMAX Forum Profiles.

2.4. TX Diversity and Simultaneous MIMO with FDPS In general Release 8 supports more advanced multi-antenna technologies compared to 802.16e. The multi-antenna and MIMO capabilities of WiMAX are improved in 802.16m, but the standard specification from IEEE leaves the integration with the upper radio layers, core network, and IOT certification to the WiMAX Forum. Because of the lack of a holistic approach to the solution, implementation of these


October 2009

12 (65)


and other advanced features such as beam forming have taken an in-ordinate amount of time. The following table shows the various multi-antenna schemes/options that are available with Release 8 and compares them with 802.16e.

SupportedSupportedUplink MU-MIMO

Not supportedSupportedMIMO + FDPS

Not supportedSupportedDownlink MU-MIMO

Up to 2x2Up to 4x4Antenna setup

Simple 2x2 vertical encoding

Codebook basedlinear precoding with optional CDD

DL spatialmultiplexing

SupportedSupported’Classical’ downlinkbeamforming

One codewordUp to 2 codewordsMaximum number of codewords

SFBC (2 TX ant)SFBC/FSTD (4 TX ant)


LTE-R8

Alamouti

CSTD

WiMAX-R1

TX diversity for DL data channels

TX diversity for DL control channels

SupportedSupportedUplink MU-MIMO

Not supportedSupportedMIMO + FDPS

Not supportedSupportedDownlink MU-MIMO

Up to 2x2Up to 4x4Antenna setup




SupportedSupported’Classical’ downlinkbeamforming

One codewordUp to 2 codewordsMaximum number of codewords



LTE-R8

Alamouti

CSTD

WiMAX-R1



Of particular note is the use of Space-Frequency Block Codes and Frequency-Shift Time Diversity TX diversity schemes for both 2TX and 4TX. Release 8 specifies that both SFBC and FSTD can be used for control channels and data channels. Further, Release 8 allows Downlink Multi-User MIMO, which under the right SINR conditions, allows an increase in the performance of the cell. Lastly, Release 8 LTE has a lower pilot density in the UL in comparison to the use of PUSC in WiMAX. This is remedied with the use of AMC in WiMAX, but LTE Release 8 was built from the ground up to utilize AMC and MIMO.


October 2009

13 (65)


2.5. Overhead Comparisons The following paragraphs discuss the distinct differences that WiMAX and LTE Release 8 have. In particular, the differences in protocol, architecture, L2 overhead, air interface allocations, CQI Reporting, and Control and User Plane latencies. From an organizational perspective there is quite a bit of overhead associated with the protocol architecture of 802.16x in comparison to 3GPP. For example, note that the logical radio protocol differences in the following figure. Unless noted otherwise, the differences are between 802.16e and LTE Release 8.

CRLC

MAC-u SAP

RRC SAP

PHY SAP

RLC-u SAP

NAS_SEC

IP

RRC

RLC

PHY

MAC

RLC-c SAP

MAC-c SAP

CPHY

NAS

NAS_SEC-c SAP

PDCP SAP

CMAC

PDCP

CPDCP

LTE

The logicalarchitecture of LTE is defined in terms

of service accesspoints and

(informative) primitives

Primary scope of this study

There is no clear definition of the logical radio protocol

architecture in WiMAX

Primary scope of this study

With 802.16x, there is no clear definition of the logical radio protocol architecture. Specifications for some features may span several different un-related sections/chapters. In stark contrast, the logical architecture of LTE is strictly defined and in terms of definition style and layout is similar to the other 3GPP Releases. While this may not be considered a source for “overhead” in the technical sense, the lack of clarity in the 802.16x specification places a considerable overhead on the understanding, implementation, and interoperability testing of its features. In addition, as the figure below indicates, the 802.16x does not have a formal syntax for its message definitions which requires a “hand written” parser for each message. Further, extension bits in the message are manually placed which could possibly “dead end” the message or render its parser unreasonably complex.


October 2009

14 (65)


Example: WiMAX downlink channel descriptor message (DCD)

The extension bits aremanually included(there are possible

dead ends in protocolextendibility)

No formal syntax is used in the message definitions(each message needs a

handwritten parser)

2.5.1. L2 Overhead LTE and WiMAX support similar mechanisms for the construction of MAC PDUs. Header compression, segmentation, concatenation, multiplexing, and padding are available in both systems. The figure below shows the main difference lies in the overhead needed to support these operations. As noted in the figure below, because of 3GPP’s CS Voice Heritage, a highly optimized unack mode for VoIP is utilized. As a result of these optimizations, LTE’s L2 overhead is approximately half of WiMAX.


October 2009

15 (65)


L2 header overhead

0 %

2 %

4 %

6 %

8 %

10 %

12 %

14 %

0 5 10 15 20 25 30 35 40 45 50

Scheduling rate

Ove

rhea

d

LTE (acknowledged mode)

LTE (unacknowledged mode)

WiMAX (fragementation)

WiMAX (no fragmentation)

LTE supports highlyoptimized unack

mode for VoIP

VoIP

2.5.2. DL Air Interface Resource Allocation The function of distributed transmission is somewhat different in LTE and WiMAX. While WiMAX utilizes PUSC to pseudo-randomly scatters clusters of subcarriers across the channel BW, LTE scatters localized allocations, also referred to as Resource Blocks, of contiguous carriers throughout the BW utilizing Frequency Domain Packet Scheduling (FDPS). While WiMAX implements a form of localized allocation in the form of AMC, 802.16e is unable to use AMC and MIMO simultaneously. It should be noted that the capability of performing simultaneous AMC and MIMO are expected in a future release, assuming the historical IOT issues can be overcome. LTE has been optimized from the beginning towards localized transmission, MIMO and FDPS. In LTE, the distributed allocations are primarily needed for the support of semi-persistent scheduling and high-speed users. The use of semi-persistent scheduling in LTE allows the overhead associated with resource allocation to be significantly reduced. It should be noted that the use of semi-persistent scheduling does impair the flexibility of the air interface scheduler as it is unable to utilize those resources once they have been allocated.


October 2009

16 (65)


0 %

5 %

10 %

15 %

20 %

25 %

30 %

0 5 10 15 20 25 30 35 40 45 50

Scheduling rate

Dow

nlin

k ov

erhe

ad

LTE (Distributed, semi-persistent)

LTE (Distributed, dynamic)

WiMAX (PUSC)

VoIP

Only HARQ retransmissions are

dynamically scheduledwhen using semi-

persistent scheduling

The advantages LTE over WiMAX in terms of semi-persistent scheduling could be reduced by it’s implementation in 802.16m.

2.5.3. Localized resource allocation From the beginning, LTE utilizes localized resource allocation in the UL, whereas the use of PUSC is widespread with WiMAX. The following table illustrates the primary differences with respect to localized resource allocation of the UL air interface. As noted in the table below, the capability of the UE to support UL MIMO is expected in a later release.


October 2009

17 (65)


CompatibleNot compatibleCompatibleMIMO-compatibility

5 ms5 ms1 msMaximum allocation

SupportedSupportedNot supportedAdaptive bitloading

SupportedSupportedNot supportedFrequency-scatteredUL allocations

SupportedSupportedSupportedFrequency-scatteredDL allocations (*)

WiMAX R1xWiMAX R1LTE-R8 Issue

CompatibleNot compatibleCompatibleMIMO-compatibility

5 ms5 ms1 msMaximum allocation

SupportedSupportedNot supportedAdaptive bitloading

SupportedSupportedNot supportedFrequency-scatteredUL allocations

SupportedSupportedSupportedFrequency-scatteredDL allocations (*)

WiMAX R1xWiMAX R1LTE-R8 Issue

(*) Frequency-scattering refers to a method where the data of one user is mapped over N non-adjacent resource blocks, where N is the scattering order The following figure shows the overhead associate with UL resource allocation. Note that LTE control channels are allocated in single symbol increments with a max of 3 symbols.

0 %

5 %

10 %

15 %

20 %

25 %

30 %

35 %

0 5 10 15 20 25 30 35 40 45 50

Scheduling rate (*)

Ove

rhea

d

WiMAX LTE

The exact slope of the WiMAX curve depends on how many

HARQ regions are used in relation to subbursts

(*) LTE: the number of allocated UL/DL users per 5 ms, WiMAX: the number of allocated UL/DL time/frequency chunks per 5ms

The control channels of LTE are occupied at the granularity of 1 OFDM

symbol (max 3 symbols)


October 2009

18 (65)


2.5.4. Channel quality reporting LTE and WiMAX support similar techniques for the wideband CQI reporting. However, as can be seen in the table below, there are some significant differences between the systems in the area of frequency selective reporting.

Main differences in periodic frequency selective reportingIssue

788 kHz788 kHz360 or 720 kHzFrequencyresolution (10 MHz)

Sounding / reciprocSounding / reciprocSoundingUplink CQI

Relative CQI for 5 best bands

Relative CQI for 5 bestbands

Best-M average for a cycled BW part (*)

Reporting scheme

WiMAX R1xWiMAX R1LTE-R8

Max ±1 dB per 5 msMax ±1 dB per 5 msAbsolute CQIChannel trackingcapability

5 ms5 ms1 msTime resolution

> 40 bits> 40 bits~5 bitsBand indication OH

Via MAC messageVia MAC messageWith every reportBand indication

Main differences in periodic frequency selective reportingIssue

788 kHz788 kHz360 or 720 kHzFrequencyresolution (10 MHz)

Sounding / reciprocSounding / reciprocSoundingUplink CQI

Relative CQI for 5 best bands

Relative CQI for 5 bestbands

Best-M average for a cycled BW part (*)

Reporting scheme


Max ±1 dB per 5 msMax ±1 dB per 5 msAbsolute CQIChannel trackingcapability

5 ms5 ms1 msTime resolution

> 40 bits> 40 bits~5 bitsBand indication OH

Via MAC messageVia MAC messageWith every reportBand indication

The channel tracking capability of WiMAX is limited by its relative reporting mechanism. Secondly, the CQI has a higher band indication OH in comparison to LTE. For WiMAX the overhead is >40 bits whereas with LTE it is 5 bits. Because of this, as more users are added to a sector, they need to be multiplexed amongst the frames. Because of the higher overhead and subsequent impact to uplink subframe capacity and the desire to maintain performance in mobile situations, the number of supported users is limited.


October 2009

19 (65)


Uplink control overhead (CQI+ACK/NACK)

0 %

10 %

20 %

30 %

40 %

50 %

60 %

0 20 40 60 80 100 120 140 160 180 200

Number of users in the cell

Ove

rhea

d

LTE (5 ms reporting period)

WiMAX (5 ms reporting period)

LTE (20 ms reporting period)

WiMAX (20 ms reporting period)

The highest reportingrates (5 ms for WiMAX, 1-2 ms for LTE) are notapplicable for a large

number of UEs

VoIP

The uplink controlchannels can eat up a considerable amount

of VoIP capacity

As a result of it’s faster tracking capability (minimum 1ms) and absolute tracking capability as well as it’s lower band indication overhead, LTE-R8 is able to track rapidly changing channel conditions without significant impact to uplink capacity.

2.5.5. User and Control Plane Latencies While both WiMAX and LTE are expected to fulfill 3GPP’s requirement for control latency, as shown in the figure below, LTE has a lower user plane latency due to it’s shorter frame length.


October 2009

20 (65)


0

20

40

60

80

100

120

140

160

2 3 4 5 6 7 8 9 10 11 12 13 14 15

BS <-> MME/ASN-GW delay [ms]

Tota

l C-p

lane

del

ay [m

s]

LTE

WiMAX

3GPP requirement

Both systems arelikely to achieve the

3GPP requirement for control plane latency

LTE has lowersensitivity on the BS <-> MME delay

User-plane latencyHARQ probability

9.5 ms9.5 ms3.5 ms0%


12.512.55.0 ms30 %

User-plane latencyHARQ probability

9.5 ms9.5 ms3.5 ms0%


12.512.55.0 ms30 %


October 2009

21 (65)


2.6. 802.16m and LTE Release 8 Comparison The key differentiators for overhead may be summarized as follows:

• Both 802.16 and LTE have similar OH in the time domain. 802.16m has longer symbol length than LTE which results in less over all symbols per frame.

• 802.16m has a longer cyclic prefix but a lower RTG/TTG than LTE • While the pilot density of 802.16e is higher than LTE, that has been fixed in

802.16m. • 802.16m outperforms LTE in the frequency domain. • 802.16m and LTE actually have similar basic resources in the same 5ms

WiMAX frame or LTE subframe period As can be seen from the following tables, the overheads are very similar for WiMAX 802.16m and LTE Release 8.

Overhead Comparison(2):

5.2% - 17.5%

0.17% -0.51%

0.27% - 2.7%0.27%

4.1% -13.6%*

0.34%1.09%

802.16m

0.11% - 1.15%CQI Report (CQICH/PUCCH)**

6.7% - 17.8%Total Controls

0.14% - 0.44%ACK (ACKCH/PUCCH)***

0.69%BW Req./RACH

4.9% - 14.8%A-MAP/PDCCH0.39%BCH0.39%SCH

LTE TDDOverhead Component for Control Channel

* Assume 3-10 PRBs for A-MAP** For CQI, assume 3-30 reports every frame for LTE and 16m *** For ACK, assume 10-30 allocations/frame for LTE and 16m

PDC C H in LTE takes at leastone symbol, thus higher

overhead


October 2009

22 (65)


While the control channel overhead is very similar for 802.16m and LTE, the overall overhead of 802.16m is marginally better than TD-LTE Rel 8.

12.4%13.8%Time Domain

8.6%10.5%Pilots (RS) DL

21.3%16%Frequency Domain

12.9%10.5%Pilots (RS) UL

33% - 44%31% - 40%Total Overhead6.7% - 17.8%5.2% - 17.5%Controls Overhead

10.5%5.5%Guard band

6.7%10.5%Cyclic prefix5.7%3.3%RTG/TTG

LTE TDD802.16m Overhead Component

12.4%13.8%Time Domain

8.6%10.5%Pilots (RS) DL

21.3%16%Frequency Domain

12.9%10.5%Pilots (RS) UL

33% - 44%31% - 40%Total Overhead6.7% - 17.8%5.2% - 17.5%Controls Overhead

10.5%5.5%Guard band

6.7%10.5%Cyclic prefix5.7%3.3%RTG/TTG

LTE TDD802.16m Overhead Component

2.6.1. MIMO and Beamforming Differentiators In general Release 8 LTE has better support for multiple antenna technology than 802.16e. Both 802.16m and Release 9 are considering similar technologies for enhancing performance such as:

• Aggregating multiple channels in more than one frequency band within the scope of a single MAC protocol instance.

• Multi-hop stations and Relay Nodes 3GPP Release 9 Study Items (SI) preparation is concentrating on the following:

• Relaying Nodes • Aggregated multiple channels in more than one frequency band within the

scope of single MAC protocol instance • Nomadic/Local Area concept

– 1Gbit/s in 100MHz – Flexible spectrum usage

• hNB – FSU

• Local IP Breakout • Automated configuration

– (could potentially be integrated part of other SI, e.g HNB, Relay Nodes, Macro BS, Nomadic/Local Area Concept)

• Higher number of antennas.


October 2009

23 (65)


– Dual stream SU-MIMO UL ? – Beamforming (rel. 8) + Dual stream beamforming

Release 10 or LTE-Advanced are concentrating on enhancing performance

• Enhanced Multi-Antenna Features and techniques to increase the performance of single user and multi-user MIMO

• Coordinated Beamforming • Coordinated Scheduling • Coordinated Multi-point Transmission • Joint Processing • Carrier Aggregation • Relay Nodes • Heterogeneous Networks • Uplink Multi-cluster scheduling

802.16x work groups are focusing on the following improvements

• MIMO: up to 4x4 DL an 2x4 UL • Scalable BW: 5-20MHz (if sufficient for IMT-A minimum requirements for

cellular layer) • Multi-hop relay stations • Flexible frequency reuse and interference mitigation (potential spectrum

sharing with FSS)

2.7. Interference Mitigation A lot of effort has been expended on Interference Coordination as a method of improving the performance of cellular data networks. In low load situations, CQI Assisted scheduling allows the eNode-B to base its packet scheduling decisions on the terminal’s CQI feedback.


October 2009

24 (65)


Cell edge user

equipment Frequency Domain Resource Blocks

Transmitted power

Favourable Resource Blocks for the cell edge user equipment

Scheduling decisions converge towards having neighboring cells avoid scheduling on the same PRBs (if possible), assuming the CQI feedback delay is shorter than the change of used PRBs. Therefore, CQI assisted scheduling under fractional load provides an implicit mechanism for pseudo coordination without 2x signaling. Under high load conditions where there is sufficient traffic to utilize all PHY layer resources, yet not block users due to congestion, the preferred mode of operation is to use flat power spectrum with a QoS aware packet scheduler. Single Frequency Reuse (SFR) has been investigated for situations where there is high offered traffic and the potential for blocking users is possible. Soft frequency reuse (SFR) is mainly for interference limited scenarios. As can be seen in the following figure, SFT is realized by assigning a power v. spectrum “pattern” to individual cells. These “Cell Types” mimic traditional frequency reuse schemes. The figure below illustrates an N=3 reuse of power patterns.


October 2009

25 (65)


Cell type 1

Cell type 2

Cell type 3

13

2

f

P(f)

f

P(f)

f

P(f)

SFR has to be applied in a cluster of neighboring cells to have maximum effect. The coordination of the SFT power pattern needs to be coordinated such that the neighboring cells that are highly coupled do not apply high (or low) transmit power in the exact same part of the frequency band. This, obviously, calls for coordinated actions amongst the neighboring cells. While, SFR is applicable for uni-cast transmission on the downlink shared channel, it is not applicable for BCH, SCH, PDCCH resource allocation table, etc. While the literature and some vendors show that there are considerable gains to be had for an unrealistic 19 site hexagonal grid, it is a non-trivial problem to select the “optimal” power pattern for a cluster of irregular cells with different traffic mixes (users with different QoS, different activity, different locations, etc…..). When compared against a simple time-frequency proportional fair scheduler with flat power spectrum, we can obtain trade-offs between cell throughput and user throughput at the cell edge (95% coverage) for SFR and QoS-aware PS with flat power spectrum. However, initial simulations show that the performance of SFR is marginal in comparison to QoS aware Packet Scheduling with flat power spectrum. Given the complexity of the messaging for the X2 interface and the decision making required to select the optimal power pattern, the potential for SFR at this time is marginal. However, because the potential for performance improvement exists, NSN continues to perform research in this area.

2.8. Other Approaches Other approaches to interference mitigation have been considered. In particular, PRBs being to different reuse classes, and scheduler would consider on UE basis if adjacent cell(s) are interfering or not. Such schemes have been considered, and algorithms where the UE’s are sorted in terms of cell-edge and cell-centre, and then schedule them on different sets of PRBs to create larger isolation. However, this is


October 2009

26 (65)


problematic in reality when also considering the effect of shadow fading, real antenna patterns, etc…. In most of NSN’s studies we have found that plain re-use one is the most attractive solution, when combined with QoS and radio channel aware (meaning based on CQI and UL sounding) scheduling. Users in "difficult" conditions would then automatically get more resources in terms of number of scheduler PRBs and/or power, or in terms of how often they are scheduled.


October 2009

27 (65)


3. Core Network In general, the core network for WiMAX and LTE are very similar. Both are all IP based and flat architectures. LTE provides tight interworking for interoperability with other 3GPP networks. Loose interworking, based on Mobile IP can be deployed for interoperability with WiMAX. The following slides provide a view of the core network architectures for a combined WiMAX and LTE radio access. The ability of NSN’s core network to support multiple radios is a key strategy of NSN and is an area of differentiation. However, this does not imply that there will be multi-mode devices for combined 16m and LTE service. NSN does not currently see an ecosystem developing for such devices.

Confidential1 © Nokia Siemens Networks February 2009

WiMax evolution to LTE, target architecture

ASNGW

NSN WiMAXBase Station

3GPPHSS

AAA

PDN-GW

MME

Internet

CorporateIntranet

Evolved Packet Core

Content and Services(including VoIP and IMS)

S2a

S-GW

• Interworking based on 3GPP release 8 architecture• Proxy Mobile IPv6, S2a-interface on the PDN-GW element• Network-based mobility (as opposed to client MIP)

• Allows handovers between WiMax and 3GPP release 8 Evolved Packet Core• Same interworking solution (with optimizations) has been specified by 3GPP and 3GPP2 for CDMA-interworking• Solution allows mobility between WiMax, CDMA and LTE networks


October 2009

28 (65)


8 © Nokia Siemens Networks LTE EPC Solution Summary / RTi / 22.06.2009Company Confidential

EPC

2G

3G

PCRF

Gb

Iu S4

S1-MMES1-U S11

SGi

HSS/HLR/AAA

S5

User planeControl plane

SAE Gateway

LTE

S3SGSN

BSC

RNC

SGSN

S10

Radio Access NetworkOther access networks

eNode-B

S12

MME

ServingGW

PDNGW

S6aVoLTE

Services in Packet Data Network

Internet

Operator services

Company intranets

LTE EPC Solution

OSSLIG

OBS

ASN-GWWiMAX

5 © Nokia Siemens Networks Presentation / Author / Date

Nokia Siemens Networks offers end-to-end solutions for LTE/SAE

Voice call connectivity

Business Support System

Packet Core evolution2G CS/PS

3G CS/PS

Content and servicenetworks

4G Systems

Flat all-IPControlGateways

ChargingSubscriber data managementNetwork Management

Transport solutions

Once operator evolves to TD-LTE, operator will have the added benefit of access to interworking with legacy 2G/3G systems and devices. The next slide shows the VCC capabilities between these systems. It should be noted that VCC with much more difficult to achieve with WiMAX and would require dual mode WiMAX/LTE or


October 2009

29 (65)


WiMAX/WCDMA/GSM devices. NSN do not expect these dual mode devices to become a major part of the ecosystem.

9 © Nokia Siemens Networks Presentation / Author / Date

Voice over CS for 2G/3G Access

Voice over IMS for LTE Access

IMS 7 – Ready for mobile LTE accessVoice Call Continuity between LTE and 2G/3G

SignalingUser planeVoIPVoCS

LTE

BSC

UE

UE

eNb SAE-GW

MGCF

MGW

MSC-S(Incl. S-IWF)

MMEHSS

2G/3

G

Cx

S11

SIP

H.248

MAP MGWHLR

S6a

BGF

SIP

VAS

PSTN/PLMN

RNC

IMS (incl. VCC+DTF)

• S6a Interface support for in the HSS• 3GPP PCRF for LTE/SAE• Single Radio Voice Call Continuity support: LTE -> 2G/3G CS

SR-V

CC

UE

S3-CS

H.248

IP Network


October 2009

30 (65)


4. Terminal specific aspects

4.1. Chipset specific aspects LTE FDD and TD-LTE are very well harmonized in 3GPP, see the table below. Therefore, TD-LTE implementation can take full benefit of LTE FDD products. TD-LTE products will follow soon after FDD products. NSN sees that many chipset vendors plan to implement common TDD-FDD chips. Layer 3GPP Specifications General TDD-

specific

36.211, PHY & Structure

88% 12%

36.212 Channel coding

100% 0% Just some flags

36.213 PHY procedures

85% 15%

L1

36.214 Measurements

100% 0%

36.300 E-UTRAN Architecture

100% 0%

36.306 UE capabilities

95% 5% Multi-mode

36.321 MAC 100% 0%

L2/L3

36.331 RRC 99% 1% TDD specific SIB parameters

36.41x S1 100% 0% RAN 3 36.42x X2 100% 0% 36.101 UE RF requirements

80% 20% TDD RF

36.104 eNB RF requirements

95% 5% TDD RF

RAN 4

36.141 eNB 95% 5% TDD RF


October 2009

31 (65)


conformance test

4.2. SIMs

LTE is a SIM-based technology, following the successful leads of the global 3GPP technologies, GSM and WCDMA/HSPA. According to the current 3GPP specifications, SIMs will be required for LTE. The SIM provides authentication of the end-user on the network and ensures the security of one’s data. With the SIM it is also possible to store applications for both the end-user and operator. The SIM also makes it a very simple matter for the end-user to transfer their contacts and service preferences from one device to another. The use of the SIM also affords the end-user a maximum degree of flexibility in terms of being able to easily change from one piece of mobile equipment to the other, without having to go through the operator’s customer care or provisioning department to make changes in mobile equipment. The SIM is always operator-specific and when considering a retail model, any SIM can be placed into any purchased device. The device ecosystem for LTE will be developed using the SIM concept as we know it today, thus providing continuation of the ecosystem and the interoperability aspect that is a key aspect of the 3GPP Standards.

4.3. Industry Ecosystem More than 85% of mobile subscribers are using networks based on 3GPP standards. This represents the largest and most robust ecosystem in the mobile industry providing significant cost advantages to operators and end users. The advantages of being a part of this ecosystem include leveraging the huge variety of terminals, from simple and cheap voice only terminals up to real multimedia terminals. Another advantage is the cost benefit for terminals and network infrastructure products due to the large, diversified supplier base and huge volumes of products manufactured. The wireless industry expects that the 3GPP ecosystem will continue to be leveraged for LTE/SAE as most terminals will be UMTS/LTE or GSM/UMTS/LTE multimode terminals offering cost advantages derived from a common terminal platform and mass volume production quantities. The following graph (from Informa) is a good illustration of the strength of the 3GPP ecosystem and marketplace.


October 2009

32 (65)


Subscribers per technology

0

1000

2000

3000

4000

5000

6000

7000

2008 2009 2010 2011 2012 2013 2014

Mill

ion

WiMAXLTEHSPAGSMCDMA

4.4. Addressable market available to 16m compared to LTE

NSN sees LTE as the most significant and widely deployed broadband wireless technology after HSPA. NSN believes that LTE has the potential to achieve a huge scale as it is truly a technology of convergence as major operators from both the GSM/WCDMA (3GPP) and CDMA (3GPP2) worlds have chosen LTE as their 4G technology. As of this point in time, NSN does not see a combined 16m/TD-LTE ecosystem as the future 4G TDD standard.


October 2009

33 (65)


5. Network planning aspects

5.1. Comparison of frequency reuse 1 and reuse 3 NSN has conducted a simulation campaign for FDD-LTE Reuse 1 and Reuse 3 scenarios. The conditions of the simulation campaign are provided in the Appendix 1. The simulations were done with 5MHz N=3, 5MHz N=1, and 15MHz N=1, and the results are illustrative of the tradeoffs between N=1 and N=3 frequency reuse.

5.1.1. SINR distribution The key takeaways from the figure below is that 5MHz with N=3 reuse provides better SINR distribution, as expected. 5MHz with N=3 reuse performs

• 5 dB better than 5 MHz Reuse 1, reference case (50 percentile) • 1 dB better than 15 MHz Reuse 1, reference case (50 percentile) • 7 dB better than 5 MHz Reuse 1, large case (50 percentile) • 7 dB better than 15 MHz Reuse 1, large case (50 percentile)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

-10 -5 0 5 10 15 20 25 30 35

SINR(dB)

CDF-15MHz_R1_ISD-Short_Center CDF-5MHz_R1_ISD-Short_Center CDF-5MHz_R3_ISD-Short_Center CDF-15MHz_R1_ISD-Short_Neighbor CDF-5MHz_R1_ISD-Short_Neighbor CDF-5MHz_R3_ISD-Short_Neighbor

5 dB

1 dB

When comparing the impacts to cell size larger than the reference, as expected in Large Scenario the SINR is degraded in all the cases with the biggest impact in 15MHz Reuse 1. As expected the 5Mhz Reuse 3 case had the least impact.


October 2009

34 (65)


0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

-10 -5 0 5 10 15 20 25 30 35

SINR(dB)

CDF-15MHz_R1_ISD-Short_Center CDF-15MHz_R1_ISD-Large_Center

15MHz R1

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

-10 -5 0 5 10 15 20 25 30 35

SINR(dB)

CDF-5MHz_R3_ISD-Short_Center CDF-5MHz_R3_ISD-Large_Center

5MHz R3


October 2009

35 (65)


5.1.2. Cell Capacity The table below compares the capacity and spectral efficiency of the different simulation cases. Key observations from the results:

• 5 MHz Reuse 3 provides a capacity gain relative to 5 MHz Reuse 1 • 15 MHz Reuse 1 provides a capacity gain relative to 5 MHz Reuse 3

Downlink Capacity

(Mbps)15MHz Reuse 1 -

Center 5MHz Reuse 1 -

Center 5MHz Reuse 3 -

Center Reference Case 25.08 7.08 9.00

Large Case 17.85 5.81 8.53Delta (%) 28.81% 17.92% 5.22%

Delta (Mbps) 7.22 1.27 0.47SE (bps/Hz/sector)

Reference Case 1.7 1.4 0.6

SE (bps/Hz/sector) Large Case 1.2 1.2 0.6

Not surprisingly, 15 MHz Reuse 1 capacity is impacted most when comparing the reference case to the larger cell case. The capacity for 15MHz Reuse 1 is reduced by 28.8% in comparison to 5MHz Reuse 3’s 5.2% case when comparing reference case to the larger cell case. Of note though, is that even with the 28.8% reduction in capacity, it is still higher than 5MHz Reuse 3. Another thing to note is that these simulations do not reflect the use of a Frequency Selective Scheduler which should improve the overall spectral efficiency of Reuse 1.


October 2009

36 (65)


0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

20.00

22.00

24.00

26.00

28.00

15MHz Reuse 1 -Center



BW, Reuse

Mbp

s

Reference Case Large Case

5 MHz Reuse 1/3 has 1.3X to 1.5X timesmore capacity when

compared with 5 MHz reuse 1 because of the better SINR distribution

(less interference).

15 MHz Reuse 1 has 2.7X to 2X more capacity when

compared with 5 MHz reuse 1/3


October 2009

37 (65)


5.1.3. Reuse 1 in TD-LTE and 16m as compared to 16e today This section outlines how reuse 1 works better in TD-LTE and 16m, as compared to WiMAXR1.0/16e system today. From simple SINR simulation + Shannon theory it can be seen that in hexagonal deployments reuse 1 gives better performance than reuse 3 in both UL and DL. As can be seen in the following figure, N=3 reuse nearly doubles the performance of the system due to lower interference which is seen primarily by users at the cell edge.

But this comes at the expense of reducing the mean spectral efficiency by 1/3.


October 2009

38 (65)


Reuse 1 was baseline even for WCDMA. The key to support of reuse 1 is robustness because SINR can drop below 0 dB due to higher interference so more advanced channel estimation/reference signals and channel coding is needed and this can be difficult, especially for initial control messages.

5.2. Performance of 20 MHz channel vs. running 2x10 MHz

This section compares the performance of 20 MHz channel vs. running 2x10 MHz (stacked 10 MHz channels in multi-carrier radio head). There are four main issues to consider:

1) Higher Sector and User Peak Rate: with 20 MHz channel one user can be allocated all 20 MHz bandwidth this increases peak data rate (mainly feasible in DL).

2) Frequency Diversity: from frequency diversity point of view there is no big difference between 10 and 20 MHz, at 10 MHz frequency diversity is already at max and we get optional performance from features such as frequency selective scheduling.

3) Load Balancing and Mobility: with 2x10 MHz load balancing between carriers will typically be based on the interfrequency handover. This needs measurement gaps and will give some negative impact to UE performance.

4) Trunking Gain: Some gain especially related to QoS responsiveness can be obtained from 20 MHz as full BW is handled by one fast scheduling unit versus two scheduling units which need to coordinate. Also, 1 20MHz channel


October 2009

39 (65)


will require less overhead than 2 10MHz channels. Finally, in some cases guard bands could be saved.

Given the above benefits and that the terminal ecosystem will support 20MHz channels, it would be advisable for an operator to consider TD-LTE 20MHz channels from the start.


October 2009

40 (65)


6. Comparison of advanced MIMO/BF feature sets

Although 802.16m MIMO capabilities are enhanced with respect to 802.16e, LTE Release 8 supports more advanced multi-antenna technologies.

SupportedSupportedSupportedUplink MU-MIMO

SupportedNot supportedSupportedMIMO + FDPS

Not supported?Not supportedSupportedDownlink MU-MIMO

Up to 4x2?Up to 2x2Up to 4x4Antenna setup

Codebook basedprecoding?




SupportedSupportedSupported’Classical’ downlinkbeamforming

Up to two CWs?One codewordUp to 2 codewordsMaximum number of codewords



LTE-R8

Alamouti

CSTD

WiMAX-R1

Something betterfor 4x2?

CSTD

WiMAX-R1x



SupportedSupportedSupportedUplink MU-MIMO

SupportedNot supportedSupportedMIMO + FDPS

Not supported?Not supportedSupportedDownlink MU-MIMO

Up to 4x2?Up to 2x2Up to 4x4Antenna setup

Codebook basedprecoding?




SupportedSupportedSupported’Classical’ downlinkbeamforming

Up to two CWs?One codewordUp to 2 codewordsMaximum number of codewords



LTE-R8

Alamouti

CSTD

WiMAX-R1

Something betterfor 4x2?

CSTD

WiMAX-R1x



As shown in LTE-WiMAX comparison LTE R8 has some advantage over WiMAX 16e from MIMO point of view, 16m and LTE-A seem more similar. Main aspect to highlight here is related to ecosystem: LTE will be deployed on a large scale with both MIMO and beamforming configuration, and these features are maturing not only from standards but also from deployment perspective e.g. up to 8 pipe RF modules available in the market place, remote radio heads, compact dual polarized antenna arrays, active antenna (RRH+Antenna in one box), TDD channel reciprocity exploited with online base station RF calibration etc.


October 2009

41 (65)


7. Uplink OFDMA vs SC-FDMA

All of the negatives of a high PAPR associated with OFDM led NSN, Nokia and other 3GPP participating organizations to investigate other options for UL transmission schemes. SC-FDMA was chosen because it provides the best of both single carrier and OFDM systems. It brings together a low PAPR similar to GSM and CDMA systems and good multi-path resistance and flexible spectrum allocation of an OFDM system. From a design perspective, SC-FDMA and OFDMA have about the same complexity in terms of implementation of the transmitter. On the receiver side, the SC-FDMA MIMO implementation is a little more difficult; however it is not considered an issue. See our response to 1a for more details. With SC-FDMA, the MS does not require as highly linear power amplifier as OFDMA. There are three benefits to this:

- Cheaper RF Amplifiers can be used in the terminal - Lower peak powers are required to achieve the requisite average power output power - Because of the lower power requirements, batteries and devices can be made smaller, or battery life can be extended for handheld devices.

Because of the benefits SC-FDMA brings in terms of lower cost and better power consumption performance, the terminal ecosystem will be more likely to flourish.

• Pros/cons – from MS/Tx implementation side (complexity/cost/performance tradeoffs)

SC-FDMA reduces PAPR significantly compared to OFDMA (appx 2dB). The additional DFT needed for SC-FDMA increases slightly the complexity, but not by an amount that would impact the device.

• PAPR gains vs. diversity gains

PAPR gains of SC-FDMA over OFDMA is on the order of 2dB with single TX, not only is link budget significantly improved but also power consumption reduced dramatically resulting in longer battery life. OFDMA allows for more diversity in frequency but typically gains are limited as there many other sources of diversity.

• Pros/cons from BS/Rx implementation side (complexity/cost/performance tradeoffs); MLD receivers, interference cancellation, MU-MIMO (SDMA), frequency based scheduling.

Additional DFT needed for SC-FDMA increasing slightly the receiver complexity but typically this is only very small part of the overall receiver complexity.


October 2009

42 (65)


Frequency dependent channel aware scheduling Here OFDMA has some advantage but once the limitation on resource allocation imposed by control signaling is taken into account the gains are limited (<10%). Moreover such gains are conditioned on slowly moving UE in very frequency selective channel. If the channel is less frequency selective, gains disappear. For LTE-A frequency dependent scheduling will be supported with introduction of clustered SC-FDMA. MU-MIMO MU-MIMO is supported LTE and even with current implementation we can achieve good gains especially in cases where 4 RX are available at the base station. Some restrictions from SC-FDMA but these will be removed in LTE-A will the introduction of clustered SC-FDMA. The slide below with UL MIMO evolution for LTE-A not considering clustered SC-FDMA (results under development).

Soc Class ification level Presentation / Author / Date7 © Nokia Siemens Networks

Uplink single-cell MIMO techniquesResponsible Person(s) for these results:

Kari Pajukoski and Klaus Pedersen

1 2 3 4 51

1.5

2

2.5

3

Spe

ctra

l effi

cien

cy [b

ps/H

z/ce

ll] Average Cell Spectral Efficiency

1 2 3 4 5

0.06

0.08

0.1

Spe

ctra

l effi

cien

cy [b

ps/H

z/us

er]

Coverage Spec tral Efficiency

1x4no MIMO

1x4MU-MIMO

2x4MU-MIMOSingle-S

2x4MU-MIMO

Dual-S

2x4SU-MIMO

Dual-S

The results below are from UL UPRISE 10 MHz simulations using fairly realistic assumptions and 3GPP agreed code-books for 2-Tx UE cases. Thus, results status is GREEN.

Current main Conclusions:• Compared with 1x4 and no

MU-MIMO, we gain ~50%in average cell throughput from using 2-Tx UEs and MU-MIMO.

• The use of 2-Tx UE with single stream results in improved coverage.

• Detailed system level simswith 4-Tx UEs are under preparation. Estimate of 4-Tx UE gain over 2-Tx UEis ~15% on average cell performance.

The following slides provide some basic reasoning to support the 3GPP choice of SC-FDMA over OFDMA for the UL transmission scheme.


October 2009

43 (65)


Soc Classification level Presentation / Author / Date2 © Nokia Siemens Networks

1. SC-FDMA BASICThe basic SC-FDMA transmission scheme is based on low-PAPR single-carrier transmission with cyclic prefix to achieve uplink inter-user orthogonal and to enable efficient frequency-domain equalization at the receiver side:• Each user transmit localized in frequency domain• Low PAPR by single carrier transmission• Frequency domain generation enabling zero roll-off and spectrum

efficiency up to 90 %:• Variable transmission bandwidth• Channel dependent scheduling in frequency domain

(or frequency hopping)• Robust to Power control and synchronization errors

FFT

Adjacent sub-carriers

Symbol source

IFFT

su

CP

Interference fromOther users localizedin frequency


October 2009

44 (65)



2. Why 3GPP selected SC-FDMA (over OFDMA) for ULOFDMA is spectrum efficient and allows for simple 1 tap Equalizer, but suffers from high peak to average power ratio (PAPR) which decreases power efficiency of transmitter -> This is very critical for mobile terminals with low form factor (bothfrom a battery drain and heat disipation point of view.• Potential reducing of PAR for OFDMA clearly needs

complicated solutions in the UE, which are seen infeasible• Single carrier FDMA methods with cyclic prefix provides

performance closed to OFDMA, but with a much reduced PAPR which significantly improves cell coverage.

Hence OFDMA is the best selection for DL and SC-FDMA the best selection for UL


October 2009

45 (65)



2.1 UL Link Performance comparison: SC-FDMA/OFDMA

TU 3 km/h BW 5 Mhz

0.01

0.1

1

-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

SNR [dB]

BLE

R

SC-FDMA FDE OFDMA SC-FDMA TE+FDE

BPSK 1/3

QPSK 1/3

QPSK 1/2

QPSK 3/4

16QAM 1/2

16QAM 2/3

16QAM 3/4

• SC-FDMA and OFDMA have similar link performance • Hence the reduced OBO for SC-FDMA can be mapped directly into an improved UL coverage

FDE (Frequency Domain Equalizer)•BPSK: SC-FDMA and OFDMA performs almost equally. •QPSK: OFDMA is slightly better or equal to SC-FDMA•16QAM: OFDMA 0.3-0.7 dB better than

SC-FDMA

Turbo Equalizer•For low channel coding rates SC-FDMA with turbo equalizer has same performance as OFDM.•For high channel coding rates of, .e.g., rate ¾, the Turbo Decoder performance suffers in case of frequency selective OFDM• Hence SC-FDMA performs approx 0.5-1 dB better than OFDM for high coding rates


2.2 PAPR/OBO analysis

The higher the PAPR, the higher output power back-off (OBO) is needed to avoid clipping leading to higher ACLR and increased EVM

The OBO needed to fulfill ACLR requirements depends on the PA implementation and the signal characteristics

Cubic Metric represents a better match to the OBO than PAPR

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

pi/2 BPSK QPSK 16QAMOB

O d

iffer

ence

bet

wee

n O

FDM

A a

nd S

C-F

DM

A [d

B]

CMPA1PA2

Roll–off=0

Note: pi/2 BPSK not included anymore in LTE


October 2009

46 (65)



2.3 Coverage of higher bit rates

100*(Psc-Pofdma)/Psc,P=90 % availability

Bit Rates > 7 Mbits/s (16QAM)•SC-FDMA with turbo equalizer gives 20-30% high availability compared to OFDMA, •SC-FDMA with FDE gives about 10% high availability of peak data rates compared to OFDMABit Rates < 7 Mbits/s (QPSK)•SC-FDMA with turbo equalizer gives 35-50% high availability compared to OFDMA•SC-FDMA with FDE gives about 30% high availability compared to OFDMA

0

10

20

30

40

50

60

QPSK ECR=1/3

2.36 Mbit/s

QPSK ECR=1/2

3.56 Mbit/s

QPSK ECR=3/4

5.37 Mbit/s

16 QAM ECR=1/2

7.16 Mbit/s

16 QAM ECR=2/3

9.56 Mbit/s

16 QAM ECR=3/4

10.7 Mbit/s

Avai

labi

lity

gain

com

pare

d to

OFD

MA

[%]

FDEFDE+TDE


2.4 Conclusion

• Uplink transmission scheme of LTE to be based on low-PAPR single-carrier transmission (SC-FDMA)with cyclic prefix to achieve uplink inter-user orthogonalityand to enable efficient frequency-domain equalization at the receiver side.• Benefit over OFDMA is improved user data-rates availability of 20-50% in macro-cells.• SC-FDMA gives superior performance compared to other systems using OFDMA as UL MA technology


October 2009

47 (65)



3. Performance and Complexity of MIMO Receivers

• This chapter compares the MIMO performance and complexity of SC-FDMA and OFDM :

• Chapter 2.1 discuss different receiver structure for OFDMA and shown that Interefrence cancelattiontype of receiver is the most promising one

• Chapter 2.2 comparing complexity of interference cancellation type of receiver between OFDMA and SC-FDMA

• Perfomance comparision is shown is chapter 2.3 and conclusion in chapter 2.4


3.1 OFDMA MIMO Receiver• MLD compares the received signal with the full set of possible symbols in search for most likely transmitted signal• In OFDMA, candidate symbol set is limited to a single sub-carrier – MLD

has lower complexity in OFDMA than in SC-FDMA• However, the as shown in Figure 1, simple and practical Turbo SIC

receivers provide better performance than ML receiver. This is due to the fact the turbo SIC receivers can utilize the channel coding, while an ML receiver exploiting channel coding would not be feasible from implementation point of view .

10 12 14 16 18 20 22

10-2

10-1

100

SNR[dB]

BLE

R

OFDM MMSEOFDM SIC 1itOFDM SIC 2itOFDM SIC 4itOFDM SIC 6itOFDM ML

OFDMA 2x2 SCM-C 16QAM 2/3• The performance of MMSE is not adequate even with OFDMA• The performance of the ML receivers operating on symbol level is worse than what can be achieved with a simple SIC receiver

Hence, Turbo SIC is the preferred receiver for SU-MIMO uplink in the LTE-Advanced regardless of the multiple access scheme.


October 2009

48 (65)



3.2 OFDMA/SC-FDMA SIC Complexity

• Only difference between OFDMA and SC-FDMA is an additional DFT/IDFT.• The contribution of DFT/IDFT to baseband receiver complexity is about 15 %.• The complexity of SIC receiver is similar for both OFDMA and SC-FDMA

• By reducing the number of turbo decoding iterations for each SIC iteration, the overall increase in receiver complexity can be kept low

• Turbo equalizer / SIC performs multistream equalization by utilizing an iterative loop between decoding and equalization.

1,kY

2,kY

:),1(,ˆ

kH

:),2(,ˆ

kH

)(1,

~ ikX

)(2,

~ ikX


October 2009

49 (65)



3.3 Perfromance comparision between SC-FDMA and OFDMA•Figure shows the performance of OFDMA and DFT-S-OFDMA with Turbo-SIC and MMSE receivers in SCM channel considering MCSs from QPSK to 16/64 QAM [1]. • It can be seen that there are no differences in performance between the two UL multiple access schemes with 2-by-4 antenna configuration; with 2-by-2 antennas OFDMA performs marginally better with some coding rates of 16/64 QAM. With QPSK the performance of OFDMA and DFT-S-OFDMA is the same.


3.4 Conclusion regarding to perfromance & complexity of MIMO Receiver• It has been shown that Turbo SIC type of receiver provides the best performance regardless of the UL MA scheme.

– The implementation of MLD type is more feasible for OFDMA but it can not provide any value over SIC type of receiver

• Regarding the performance, considering SIC type of receivers, there are no substantial differences between SC-FDMA and OFDMA •Furthermore, non-contiguous resource allocations, if seen as beneficial, can also be realized with clustered DFT-S-OFDM.


October 2009

50 (65)



4. On the Bandwith flexibility of LTE Uplink

LTE Release 10 enables transmission of multiple non-adjacent sub-carriers in the same time• Spectrum usage similar than with OFDMA system• Better utilization of the frequency selectivity of the channelIn LTE, the non-contigous transmission in frequency is still form of serial transmission having essentially lower PAR than OFDMA

DFTSub-car-rie

Map-ping

IFFT

CPInsertionModulatorChannel

Coding

F-D packet scheduler allocation

User #1

User #2

RP index

frequency


October 2009

51 (65)


8. Standards Environment In this section, differences between IEEE/WF spec/cert process compared to 3GPP are discussed, including viability of practical feature development, certification process, and timelines.

8.1. Pros and cons contrast of the organizational structure, process, and funding

There are several key differences between the WiMAX and 3GPP ecosystems. In the WiMAX ecosystem, PHY and MAC development as well as ITU-R support occurs in IEEE where membership belongs to individual (not companies). Thus there is participation and input from academia and small companies in addition to the "big telco" companies. The process makes frequent use of voting for decision process, and includes guidelines and oversight from IEEE Standards Association for the entire standardization process. Standardization in IEEE 802 is self-funding, with registration fees from all members of IEEE 802 funding the Plenary meetings, and registration fees from members of IEEE 802.16 & 802.21 funding the interim meetings. Development of the upper layers of the Radio Access Network, the Core Network, WiMAX Forum Certification, WiMAX Roaming, and also marketing and regulatory support occurs in the WiMAX Forum. The WiMAX Forum is structured as a LLC, with Board of Directors consisting of representatives from key WiMAX companies. The structure of a Corporation led by a Board of Directors allows the standardization and other activities to be directed to support the commercial interests of the majority of members in the WiMAX Forum. Funding comes from Company membership fees, Certification fees including logo licensing fees, meeting attendance fees, and WiMAX Forum sponsored trade shows. 3GPP LTE is supported by all major operators and vendors, the ecosystem is extremely versatile and diverse and not driven by the interest of a single company. The benefits of the 3GPP standards process include consensus building instead of voting, impressive track record in building implementable standards with good trade off between performance and complexity and long term commitment to provide smooth evolution.

8.2. Detailed timeline of 3GPP path

General 3GPP schedules and status are provided in the slides below. Details on the 3GPP specific milestones are provided after this introduction.


October 2009

52 (65)


4 © Nokia Siemens Networks Presentation / Author / Date Nokia Siemens Networks Confidential

Release 8LTE (FDD/TDD)

SAE “All-IP”CDMA Inter-working

Third Generation Partnership Project “3GPP”Release Timeline

Release 6HSUPAI-WLAN

Release 7HSPA+/I-HSPA

2000 2005 2007

Release 4TD-SCDMAMSC-Server

2001

Release 5HSDPA

IMS

2002

Release 99WCDMA

MSC

Release 9LTE-LCS, E911SON, HSPA+++

H(e)NB

Func. Freeze 12/09ASN.1 03/10

Release 10LTE-Advanced

TBD

March 2009 ~ 18 months post Rel-9 completion


Detail of Rel-9 and Rel-10 Timing

1Q2009 2Q2009 3Q2009 4Q2009

REL-8 finalization (RRC ASN.1 freezing)Rel-9 starts in RAN1

Rel-9 WID work starts in RAN2, RAN3 & RAN4

Completion of Rel-9 functionality including RAN2 and RAN4 core requirements (at least for main parts)

1Q2010

Rel-9 finalization (e.g. RRC ASN.1 freezing)

2Q2010 3Q2010 4Q2010 . . .

1Q2010

First Rel-10 WIs start in RAN1 (expected)

2Q2010 3Q2010 4Q2010

Rel-10 WIs starting in RAN2-RAN4 (expected)

1Q2011 2Q2011 3Q2011 4Q2011

Rel-10 Content Freeze (expected) ASN.1 (expected)Rel-10 stage 3

created (expected)


October 2009

53 (65)



RAN 5 Status Rel-8 (p1)(note Rel-9 work in RAN5 will start in 2010)• COMMON LTE SPECIFICATIONS

• TS 36.508 (estimated now 100% complete): • On the SIG side, waiting feedback from TTCN verification.

• TS 36.509 (estimated now 95% complete): • Covers all the required functions, the main remaining question is to address

the security of control messages.• LTE SIGNALLING

• Current overall WI status for LTE SIG conformance test specifications is estimated now as 90% complete expected to be completed 03/’10

• TS 36.523-1 (Protocol test cases), estimated now 55% complete (90% for P1 and P2 cases):

• 452 test cases to be written (covering protocol layers & combined procedures)• 164 test cases are 100% complete (ready for TTCN development/verification)• 96 test cases are 80% complete (i.e. ready for TTCN development)• The development of TS 36.523-2 is aligned with 36.523-1.• TS 36.523-3 (TTCN) approved as v2.0.0 at RAN Plenary


RAN 5 Status Rel-8 (p2)

• LTE RF• General: Meeting recorded over 150 RF/RRM documents. Overall completeness of LTE

RF WI is now 86%, where TDD has reached the same level as FDD.• WI Completion date postponed to Dec 2009:

• Except for CSI and PHICH tests, most RF test cases are now 90-100% complete with only a few in 80% due to later arrival from RAN4.

• TS 36.521-1• Transmit Power

• Maximum Output Power clarified with 1.5dB corner frequency relaxations• MPR agreed as conditional with SEM and ACLR• A-MPR further defined, including NS_07 & NS_08 additions• Configured UE transmitted output power extended with 1 test point,

• Output Power Dynamics• Minimum Output Power completed• Transmit OFF Power agreed to be covered by time mask• ON/OFF Time Mask procedures agreed• PRACH and SRS Time Mask procedures agreed• Absolute Power Tolerance procedures agreed, missing confirmation 1.5dB relaxation details• Relative Power Tolerance procedures conditionally agreed, waiting for RAN4 guidance• Aggregate Power Control Tolerance completed

• TS 36.521-2 ICS• Routine updates on TCs’ applicability

• TS 36.521-3 RRM• Mostly routine updates already agreed resubmissions on Idle Mode, Connected State &

Mobility, Timing, and Measurements were updated based on core requirements.


October 2009

54 (65)



RAN 5 Status Rel-8 (TDD Specific)

• TS 36.508: A/N mode update: The default A/N mode for TDD in 36.508 is changed to use multiplexing instead of bundling, to align with RAN4’s simulation assumption.

• TS 36.508: Add 5MHz channel bandwidth option for Operating Band 40 (2300-2400 MHz).

• TS 36.508: TDD special subframe pattern update: The ssp6 with two symbols for UpPTS is used as default in 36.508.

• TS 34.114: LCR-TDD UE over the Air conformance testing methodology is agreed. Therefore this WI is completed.

• LTE SIG: Multi-cell interference reduction discussion for TDD (R5-094376): Next step, the details are to be included in TS36.523-3

• LTE SIG MAC: Updates on UL SPS and DL SPS is agreed. For UL SPS, the TDD specific feature of twoIntervalConfig is included.

• RAN/EPC Stage-3 specs complete (RAN1/2/3 + SA2)

The RAN 1/2/3 and SA2 LTE/SAE REL8 are complete and frozen. The final specification freezing, including ASN.1 freezing, took place in March 2009. No RRC ASN.1 non-backwards compatible CRs have been approved since then, indicating that the specifications are very stable. The LTE FDD and TDD specifications are fully aligned and they were completed and frozen at the same time. Also, the RAN4 LTE REL8 requirements specifications both for FDD and TDD were frozen in March 2009. The RAN4 specifications include LTE FDD and LTE band variants including Band 38. Following this, RAN4 has worked on some requirement details but primarily on test cases which are based on the core requirements developed by RAN4 and frozen in March 2009. Current targets are for R9 to be frozen in Dec.2009 and R10 to be frozen in March 2011

• RAN 5 conformance specs complete

Completion levels after the last RAN5 meeting in August 2009 are:

1. TS 36.523-1 (Protocol test cases), estimated now 55% complete (but 90% for GCF P1 and P2 cases)

2. TS 36.521-1 (RF) estimated 90% complete 3. TS 36.521-3 (RRM) estimated 82% complete

In general TDD test implementations are at the same level as FDD.


October 2009

55 (65)


The RAN5 target is to complete all GCF priority 1 and 2 signaling and also priority 1 and 2 RF/RRM test cases at the last meeting of 2009 at end of October. This would mean that the LTE work items would be changed to maintenance mode. It is worth noting that RAN5 will still continue completing the priority 3 and 4 signaling test cases and also some issues in the RF/RRM area.

• RAN 5 test suites complete MCC160 has concentrated on implementing the GCF priority 1 and 2 cases in TTCN. At the end of August 2009 there were already 163 (out of 217) GCF priority 1 and 2 test cases implemented in TTCN and the numbers have increased since. The last delivery was in mid September 2009 (week 38). The target is to align priority 1 and 2 cases with the December’09 prose specs so that all the CRs from the October 2009 RAN5 meeting would be implemented at the end of 2009. So all priority 1 and 2 cases should be implemented in TTCN and ready for verification at the end of 2009 and will be moved to maintenance mode for 2010. TTCN verification for the FDD cases has already started. At least one SS vendor (Anite) has submitted verification results to the RAN5 reflector. They have submitted 6 test cases (1 RLC, 1 EMM, 3 RRC and 1 MAC) for verification. The target date for the verification approval of the 1st submitted case is the 26th of October. So far, no participants have commented on their verification logs but comments are expected. FDD and TDD TTCN implementations should be aligned but the SS vendors plans for the TDD testers are not so clear and also one TDD terminal will be needed for the TDD TTCN branch verification (and 2 terminals for GCF validation).

• Test equipment availability (radio/protocol)

Radio: As we have worked towards starting the verification and test development work first with FDD, the SS vendors have also planned their test equipment roadmaps accordingly. It seems that other UE vendors are doing the same, since no SS vendor has stated specific plans as to when the TDD equipment will be available. Nevertheless, all major SS vendors have stated their intentions for TDD development, in which capability will be in place later in 2010, based on Nokia’s current understanding. FDD test equipment is ready for verification during this Autumn (2009). It is Nokia’s understanding that TDD test equipment availability will not lag much behind FDD if there is business interest in getting TDD test equipment as the TDD test cases are developed together with FDD and there is a good alignment between the FDD and TDD test cases. Protocol: Anite, AT4Wireless and R&S indicated in the previous RAN5 meeting that they are expecting that first signaling TTCN-3 test cases (FDD) to be verified using their SS platform in Sep 2009. So far only Anite has submitted test cases for verification. Anritsu indicated delays in their plans and they are expecting the first test cases are verified late this year. All these SS vendors are committed to developing an LTE TDD


October 2009

56 (65)


test platform, but availability of this is expected later (1H/2010) than FDD (depends also on UE availability). Again, if commercial interest in the TDD test equipment is indicated, it is Nokia’s understanding that TDD test equipment availability will not lag much behind FDD.

• LSTI IOT and IODT sessions TDD IODT work has started Q1/09 with PHY layer test vector verification (simulation environment verification, Uplink and Downlink) based on Dec/08 3GPP standards initially. An update to Mar/09 3GPP standards level took place during Q2/09 and LSTI IODT concluded the test vector verification work Q3/09 based on the fact that at least four vendors within LSTI have successfully passed all test cases Lab-based TDD IODT/IOT verification of real UE and NW equipment (pre-commercial) is currently ongoing, based on Mar/2009 standards. The first testing results were reported in Sep/09. The current official schedules to finalize the TDD IODT/IOT verification is Apr/2010.

• GCF certification start The GCF target is to start GCF certification at the end of 2010 (Q4/2010). One of the GCF certification entry conditions is that there is at least one commercial LTE network in operation in the region of interest so a Band 38 network is needed. Additionally, GCF has begun defining Field Trial requirements for LTE. The GCF Field Trial requirements will refer to the GSMA DG.11 specification as a guideline as was also done with GERAN and UTRA GCF FT requirements. Work is ongoing in the GSMA Device Group to draft LTE Field Trial test scenarios. It is expected that there will be ~30 GCF LTE FT requirements (this is based on the current definition work in GSMA DG LTE) and the GSMA DG LTE drafting work is scheduled to be completed by Q2 2010 LTE Field Trials will be conducted in GCF "Field Trial Qualified Operator" (FTQO) networks using a combination of infrastructure. The introduction of GCF LTE Field Trials will be synchronized with the introduction LTE conformance test cases and availability for GCF Field Trial Qualified Networks.


October 2009

57 (65)


8.3. Summary of Standards and Expected Deployment Timelines

802.16m - Specification likely complete Q4 2010 or Q1 2011 - Rel 2 certification begins Q1 2012 - First commercial deployments likely late 2012 LTE - R8 specification complete today - R8 commercial deployments expected in 2010 and widespread in 2011 - R9 specification expected to be frozen Q4 2009 - R9 commercial deployments expected in 2011 - R10 (LTE-Advanced) specification expected to be frozen Q1 2011 - R10 (LTE-Advanced) commercial deployments expected in 2012 As you can see, 16m, which NSN feels is roughly on par in terms of performance with R8 TD-LTE, will be available for commercial deployments about 2 years after R8 TD-LTE. Also, R9 TD-LTE which brings improvements to R8 and is expected to have better performance than 16m, will be available for commercial deployments up to a year before 16m. Finally, R10 LTE-Advanced will be available in about the same time frame as 16m. From a commercial deployment schedule perspective, TD-LTE has a 1 to 2 year time to market advantage compared to 16m.


October 2009

58 (65)


9. Operator Migration/Co-existence

9.1. Practical pros/cons of operator's optimized migration path from 16e >16m vs. 16e TD-LTE (Rel9 or 10)

NSN does not see migration from 16e to 16m as the optimum 4G evolution path for operator considering the questions and scope of this document. NSN expects the LTE ecosystem to be much broader with far more devices and a greater global reach. In general, 16m and TD-LTE will have very similar performance with various trade-offs that have been discussed in this document. It would be difficult, if not impossible to make a strategic decision about the better technology between 16m and LTE. NSN’s opinion is that the best migration for operator is not a question of technology, but of ecosystem and global compatibility. In that respect, migrating to TD-LTE is the best path forward. That being said, NSN will be supportive of whatever technology migration decision is made and willing to help operator in reaching it’s goals of a robust, high capacity, 4G network. TD-LTE solutions will be available from NSN and WiMAX solutions through various partners.

9.2. 16m migration using backwards compatibility mode

In this section, a practical example of an operator moving from a mature 16e market to 16m is provided, using backwards compatibility mode. Estimates of system performance impact and operational complexities associated with such a transition are provided. In practice, migrating from 16e to 16m will require the deployment of a separate 10 or 20 MHz carrier per sector for 16m. This migration path will require an extra 30 to 60 MHz per site, new HW at each site, and a tower climb. The assumption for the existing network at the time of migration includes sites that are already at the max capacity for the existing RRH – 2x10 MHz per sector for 16e. This is the capacity limit of the existing 16e RRHs. Adding 16m capacity with the backward compatibility aspects of the 16m specification on existing dual carrier RRHs, would impact the existing network performance by stealing symbols from one technology to the other. The next slide shows that even though there is a compatibility mode, the result will steal capacity from the existing carrier already designated for 16e.


October 2009

59 (65)



16e and 16m UL Backward Compatibility• In the Uplink, 802.16e and 802.16m can be multiplexed either in time or

in frequency.• A special UL physical structure compatible with the 802.16e UL-PUSC

permutation facilitates frequency multiplexing.

At the highest level, the performance impact would be no room for growth in the existing network, and even a degradation of existing customer’s performance. For simplicity, let’s assume that adding 16m to the existing 16e frame adds no extra overhead and provides a 1 to 1 trade of 16e capacity for 16m capacity. If the existing network already requires 2x10 carriers, then stealing capacity from 16e is not an option. In practice, the addition of 16m to existing 16e frames will not be without additional overhead and will not provide a 1 to 1 capacity trade. Since NSN is not doing development work for 16m products internally, we do not have lower level details around the performance impacts. We will be happy to discuss with our partners if case further insight on this impact is requested. From the spectrum perspective, the assumption that each site is using 60MHz – 3x2x10MHz, means that adding 16m to the network would require an additional 30 MHz per site if 10MHz per sector was deployed. If the 20MHz option was to be deployed, each site would require an additional 60MHz, bringing the total to 120MHz per site. Likely this is not an option in most markets. Therefore, the best possible migration would be deploying 16m as an overlay as follows:

- Add extra HW per site – including at least RRHs and tower climbs - Add a minimum of 30MHz spectrum per site – a total of 90MHz per site - Add external combiners or new antennas for the new 16m RRHs - As 16e customers decrease, turn off 16e carriers and turn on 16m carriers


October 2009

60 (65)


It is not likely that the 16e network will be able to be turned completely off for many years. Other similar network migrations have shown that the legacy networks must be kept up for many, many years (10 years for large established networks is not unusual) or the carrier must pay to migrate those legacy networks with new customer CPE/devices – a costly option.

9.3. Migration from 16e to TD-LTE as an overlay NSN would recommend an overlay strategy for initial TD-LTE deployments with a spectrum migration plan for WiMAX to LTE evolution. As with any technology evolution, as users migrate to LTE, WiMAX spectrum can be reduced and LTE carrier BW increased to efficiently utilize the spectrum assets. Another option is to plan for operating 2 separate networks - maintaining the baseline 16e network for certain user profiles and use cases and a separate TD-LTE network for full mobility and interoperability with other technologies. This would allow operator access to markets that are currently closed to them and allow more potential MVNO partnerships. In practice, migrating from 16e to TD-LTE will require the deployment of a separate 20 MHz carrier for TD-LTE. This migration path will require an extra 20 MHz per site, new HW at each site, and a tower climb. Basically the same site requirements as 16m migration except that the spectrum requirement would be smaller due to LTE’s reuse 1 scenario. Performance impacts would be minimal as this migration would involve two separate systems each capable of increasing or decreasing carriers as the network evolved and grew. NSN believes that the two networks would be compatible and able to co-exist side by side. There would need to be some adjustment made to the WiMAX UL/DL ratio, but other than that, operational impacts should be minimal. The slides below provide a very basic look at co-existence between WiMAX and TD-LTE.


October 2009

61 (65)


TD-LTE and WiMAX in same band

• WiMAX and TD-LTE can be used in the same coverage area in the adjacent carrier frequencies with the following conditions:

– Systems are synchronized– DL and UL division is agreed– The two networks’ radio configuration parameters are set properly to avoid

DL<->UL overlapping in time between the systems• These listed conditions are similar also in WiMAX only or TD-LTE only.

There are combinations of LTE and WiMAX (DL,UL) divisions which have no capacity loss, i.e. no parts of the frames need to be muted, e.g.

– LTE (2:2) DDSUU <-> WiMAX (DL,UL)=(26,21) – LTE (3:1) DDDSU <-> WiMAX (DL,UL)=(35,12)

• Nokia Siemens Networks supports both above configurations already in the first TD-LTE release


October 2009

62 (65)



Support for WiMAX TDD and LTE TDD in the Same Band on Different Carriers - General• Deploying two TDD cellular systems in the same coverage area on adjacent

carriers requires synchronization of the networks and alignment of the DL and UL parts of the transmitted radio frames– This is to avoid BS-to-BS and MS-to-MS interference– The BS-to-MS and MS-to-BS interference from adjacent carrier is controlled by the spectrum masks if

both systems have been designed to enable operation on the adjacent carriers, which is true in the case of LTE and WiMAX

UL

ULUL DL

DLDL UL

DLSystem 1on carrier f1

System 2on carrier f2

UL

ULUL DL

DLDL UL

DL


OK

BS-to-BS interferenceMS-to-MS interferenceNOK

Transmit-to-receive and receive-to-transmittransition gaps (TTG and RTG)


time

time

Information provided under NDA.


WiMAX TDD Frame Structure

• WiMAX frame is 5 ms long consisting of a number of DL OFDMA symbols, a number of UL OFDMA symbols, a transmit-to-receive transition gap (TTG) and a receive-to-transmit transition gap (RTG) (IEEE 802.16e-2005)

• WiMAX Forum has defined a list of mandatory (DL,UL) divisions ofsymbols, which depends on the system profile. The most common one is for the 5 MHz and 10 MHz channel BW, where (DL,UL) = (26,21),…,(35,12), OFDMA symbol length ≅ 102.9 μs, TTG ≅ 105.7 μs and RTG = 60 μs

• Within a (DL,UL) division, parts of DL and UL subframes can be “muted”by the scheduler function of the BS

0 us 2674.3… us

TTG = 105.7... us

RTG = 60 us

OFDMA symbol = 102.9... us

2780 us 4940 us 5000 us

Example: DL/UL = 26/21

DL Preamble DL Control

DL Data UL Control

UL Data



October 2009

63 (65)



LTE TDD Frame Structure (3GPP TS 36.211 V8.2.0, 2008-03)• LTE Frame length is 10 ms consisting of 10 subframes of length 1ms each, consisting of

2 slots length of 0.5 ms each.• In TDD the subframes are allocated either to DL (D) or to UL (U) and in addition there is a

special subframe S which contains a DL part (DwPTS) a guard period (GP) and an UL part (UpPTS). The guard period is used for UE timing advance to compensate the propagation delay (like TTG in case of WiMAX).– There are several parameter settings defined for shorter and longer guard periods between the DL

and UL, trading-off between the cell capacity and cell range• Seven configurations have been specified with different (DL,UL) divisions and either 5 ms

or 10 ms DL/UL switching period• Each normal 1ms subframe is either allocated to DL or UL but there is not much flexibility

within a subframe (e.g. muting part of a subframe)

DUUSDUUUSD10 ms6DDDDDDDUSD10 ms5DDDDDDUUSD10 ms4DDDDDUUUSD10 ms3DDUSDDDUSD5 ms2DUUSDDUUSD5 ms1UUUSDUUUSD5 ms09876543210

Subframe numberSwitch-point periodicity

(DL,UL) configuration S

Dw

PTS

GP

UpP

TS



October 2009

64 (65)



Matching the WiMAX and LTE TDD Frames• The LTE TDD configurations with 5 ms switching point periodicity can be matched with

WiMAX 5 ms frames, by matching the GP of LTE’s special subframe to WiMAX TTG• The figure shows this as an example for the LTE configuration 1 synchronized with WiMAX

(DL,UL)=(26,21), such that muting parts of frames might not needed– Similarly the LTE configuration 2 (DDDSU) can be matched for example with WiMAX (DL,UL)=(35,12)

WiMAX (26,21)UL 21DL 26 UL 21DL 26 UL 21DL 26

LTE (DDSUU)

5 ms

Dw

PTS

GP

UpP

TS

time from WiMAX frame start (μs)

(LTE cofiguration with 214us GP and one UL sounding reference symbol)

TTGDL UL

D4

D5

S6

U7

U8

D9

DUUSDDUUSD9876543210

D4

D5

S6

U7

U8

D9

DUUSDDUUSD9876543210

Information provided under NDA. From the spectrum perspective, the assumption that each site is using 60MHz – 3x2x10MHz, means that adding TD-LTE to the network would require an additional 20 MHz per site as opposed to the 60 MHz per site in the 16m case. This would bring the total spectrum need per site up by only 33% to 80MHz. This is due to the basic design for a reuse 1 deployment in LTE. Therefore, the TD-LTE migration path would be as follows:

- Add extra HW per site – including a BTS and tower climbs - Add 20MHz spectrum – to a total of 80MHz per site - Add external combiners or new antennas for TD-LTE - As 16e customers decrease, turn off 16e carriers and turn on more TD-LTE

carriers

In case an operator has a limited spectrum allocation, e.g. 20MHz spectrum in total for both WiMAX 16e and TD-LTE systems, then the migration from 16e to TD-LTE will have a high impact on 16e performance. Introduction of TD-LTE will require the deployment of a separate 10MHz TD-LTE carrier, thus limiting the available spectrum for 16e to 10MHz. With 3-sectorized sites and PUSC-1/3 frequency reuse within the 10MHz carrier bandwidth, the RF carrier is subdivided into 3 non-overlapping sets, allocated to the sectors, i.e. each sector uses effectively 10/3 MHz spectrum, as shown in the figure below. Having only 10/3 MHz per sector would severely reduce the per cell capacity, maximum throughput per subscriber and hence degrade the overall user experience. Likewise, the capacity and maximum throughput of the TD-


October 2009

65 (65)


LTE network would be lower compared to an operator using full 20MHz TD-LTE carrier bandwidth.

f1

f2

f3

(a) WiMAX 16e: 10/3 MHz per sector

WiMAX 16e 10MHz and TD-LTE 10MHz deployed at the same Base Station site

Each sector uses 10/3 MHz(PUSC-1/3 frequency reuse)

f4

f4

f4

(b) TD-LTE carrier added: 10MHz per sector

Each sector uses 10 MHz(frequency reuse 1)

It is not likely that the 16e network will be able to be turned completely off for many years. Other similar network migrations have shown that the legacy networks must be kept up for many, many years (10 years for large established networks is not unusual) or the carrier must pay to migrate those legacy networks with new customer CPE/devices – a costly option. Therefore the 16e BTSs would not be able to be “SW upgraded” to TD-LTE any time in the foreseeable future.

In conclusion, the migration path to 16m and TD-LTE are very similar, both requiring HW, site work, and spectrum. In the 16m case, the amount of HW required may be smaller, but only if existing vendors’ additional LTE RRH could use the same BB modules and they were sized to handle the extra capacity. Otherwise a new BB module would be needed, the same as the TD-LTE migration case. In the TD-LTE case, the spectrum requirements would be smaller for even much higher level of capacity – 20MHz total with a reuse 1 for LTE as opposed to 30MHz total providing only 10MHz per sector. The site work would be roughly the same. So, migrating to 16m does not bring any real benefit from the site/cost/spectrum perspective.

TD-LTE WiMAX Comparison

Documents

official customer document

performance comparison

wimax technology

confidential tdlte

customer comments

mentioned hardware

prior notice

software products