03 - LTE Dimensioning Guidelines - Outdoor Link Budget - FDD - Ed2.9 - Internal (1)

INTERNAL LTE Dimensioning Guidelines – Outdoor Link Budget - FDD February 2011

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET>

Copyright © 2011 by Alcatel-Lucent. All Rights Reserved. About Alcatel-Lucent Alcatel-Lucent (Euronext Paris and NYSE: ALU) provides solutions that enable service providers, enterprises and governments worldwide, to deliver voice, data and video communication services to end-users. As a leader in fixed, mobile and converged broadband networking, IP technologies, applications, and services, Alcatel-Lucent offers the end-toend solutions that enable compelling communications services for people at home, at work and on the move. For more information, visit Alcatel-Lucent on the Internet.

Notice The information contained in this document is subject to change without notice. At the time of publication, it reflects the latest information on Alcatel-Lucent’s offer, however, our policy of continuing development may result in improvement or change to the specifications described.

Trademarks Alcatel, Lucent Technologies, Alcatel-Lucent and the Alcatel-Lucent logo are trademarks of Alcatel-Lucent. All other trademarks are the property of their respective owners. AlcatelLucent assumes no responsibility for inaccuracies contained herein.

Alcatel-Lucent – Proprietary Version 2.9

2 / 61

INTERNAL


Copyright © 2011 by Alcatel-Lucent. All Rights Reserved. About Alcatel-Lucent Alcatel-Lucent (Euronext Paris and NYSE: ALU) provides solutions that enable service providers, enterprises and governments worldwide, to deliver voice, data and video communication services to end-users. As a leader in fixed, mobile and converged broadband networking, IP technologies, applications, and services, Alcatel-Lucent offers the end-toend solutions that enable compelling communications services for people at home, at work and on the move. For more information, visit Alcatel-Lucent on the Internet.

Notice The information contained in this document is subject to change without notice. At the time of publication, it reflects the latest information on Alcatel-Lucent’s offer, however, our policy of continuing development may result in improvement or change to the specifications described.

Trademarks Alcatel, Lucent Technologies, Alcatel-Lucent and the Alcatel-Lucent logo are trademarks of Alcatel-Lucent. All other trademarks are the property of their respective owners. AlcatelLucent assumes no responsibility for inaccuracies contained herein.


2 / 61

INTERNAL


History Changes

Date

Author

Ed 1.0 – 1st Release

Dec 2008

Keith Butterworth

Ed 2.0 - Quality review and edits, minor edits to section 4.1

Feb 2009

Keith Butterworth

Ed2.1 – Correction to interference margin definition

Mar 2009

Keith Butterworth

Ed2.2 – Updates to modem performances and active user & throughput computations. Revamp of parameter naming for air interface and modem computations. Addition of ACK/NACK link budget considerations.

Jun 2009

Keith Butterworth

Ed2.3 – Updates to the link budget aspects (modification of UL link budget + addition of revised DL link budget).

Nov 2009

Keith Butterworth

Ed2.3 – Minor updates and corrections

Dec 2009

Keith Butterworth

Ed2.5 – Alignment with Ed8.2 link budget (updated SINR figures, FSS Gain, revised IoT section, rework of DL section, spatial multiplexing gain)

Feb 2010

Keith Butterworth

Ed2.6 – Update inline with new dimensioning guidelines document structure + alignment with changes in Ed8.3.2 of link budget tool

Apr 2010

Keith Butterworth

Jul 2010

Keith Butterworth

Ed2.8 – Minor editorial updates (correction of interference margin equation). Updates to align with Ed 8.4 of the LKB tool. Addition of 8bit CQI report over PUCCH link budget.

Sept 2010

Keith Butterworth

Ed2.9 – Updates to align with Ed8.5 of the LKB tool. Correction of effective coding rates and other minor corrections.

Feb 2011

Laurent Demerville

Ed2.7 – Minor changes to sections 2.1.4.4, 3.1.3 and 3.1.5.4.

Reviewed by ARFCC (Advanced RF Competence Centre)


3 / 61

INTERNAL


CONTENTS 1

Introduction ....................................................................... 8

2

Uplink Link Budget..............................................................10 2.1

Uplink Link Budget Parameters ................................................. 11

2.1.1 2.1.2 2.1.3 2.1.4 2.1.5 2.1.6 2.1.7 2.1.8 2.1.9 2.1.10 2.1.11

2.2

Final MAPL and Cell Range....................................................... 32

2.2.1 2.2.2

2.3

Propagation Model .....................................................................33 Site Area .................................................................................34

Impact of RRH and TMA .......................................................... 35

2.3.1 2.3.2

RRH .......................................................................................35 TMA.......................................................................................35

2.4

Uplink Budget Example........................................................... 36

2.5

Uplink Common Control Channel Considerations ........................... 36

2.5.1 2.5.2 2.5.3

3

UE Characteristics......................................................................12 eNode-B Receiver Sensitivity.........................................................12 Noise Figure .............................................................................12 SINR Performances .....................................................................13 Handling of VoIP on the Uplink ......................................................21 Uplink Explicit Diversity Gains .......................................................23 Interference Margin....................................................................24 Shadowing Margin ......................................................................27 Handoff Gain / Best Server Selection Gain ........................................28 Frequency Selective Scheduling (FSS) Gain ........................................30 Penetration Losses .....................................................................32

Attach Procedure.......................................................................37 ACK/NACK Feedback...................................................................38 Periodic CQI Reports ...................................................................40

Downlink Link Budget ..........................................................42 3.1

Downlink Budget Parameters ................................................... 43

3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 3.1.6 3.1.7 3.1.8

3.2

SINR.......................................................................................43 RSRQ......................................................................................45 Interference Sources ..................................................................46 Geometry................................................................................47 Downlink SINR Performances .........................................................50 Resource Element Distribution.......................................................54 Energy Per Resource Element (EPRE) ...............................................55 Shadowing Margin & Handoff Gain ..................................................56

Downlink Budget Example ....................................................... 57

4

Downlink Output Power........................................................59

5

Radio Network Planning .......................................................60

6

Summary ..........................................................................61


4 / 61

INTERNAL


EXECUTIVE SUMMARY The purpose of this series of dimensioning guidelines is to describe details of AlcatelLucent’s dimensioning rules for the LTE Frequency Division Duplex (FDD) air interface and eNode-B modem hardware. A first step of the network design process consists of determining the number of sites required and deployment feasibility according to the following information:     

Site density of any legacy network deployments, Frequency band(s) used by the legacy system(s), if applicable Frequency band(s) used by the LTE system, Bandwidth available for LTE (1.4, 3, 5, 10, 15 or 20 MHz), Requirements in terms of LTE data rates at cell edge (e.g. uplink data edge to be guaranteed, best effort data, VoIP coverage requirements, etc.).

This initial number of sites is then typically refined by means of a Radio Network Planning (RNP) study, taking into account site locations, accurate terrain databases and calibrated propagation models. The figure below illustrates key inputs and outputs of the AlcatelLucent eNode-B dimensioning process: Coverage Outputs

Coverage Inputs • Area to be covered • Targeted service at cell edge

• Cell Range

Link Budget RF Planning

•Legacy Site Reuse •Number of Sites

• Indoor penetration level

+ Traffic Inputs

Network Information • Incumbent network info • LTE Frequency

eNodeB Configuration

• LTE Maximum bandwidth

• LTE Bandwidth

Air Interface Capacity Analysis

• MIMO Scheme, Output Power

Traffic Inputs

eNodeB configuration

• Number of subscribers • Traffic profile per subscriber Optional Requirements • Peak Throughput per Site

• Number of modems

Traffic Model Modem Dimensioning

• Modem configuration - No. connection tokens - UL & DL Throughput tokens

Figure 1: Alcatel-Lucent Dimensioning Process As implied in the figure, Alcatel-Lucent’s process relies on advanced dimensioning rules for Link Budget Analysis, Air Interface Capacity Analysis, eNode-B Modem Dimensioning, and Multi-service traffic modeling. The dimensioning process takes into account product release functionalities and will be updated regularly to follow product evolutions. As background to further discussion of this process, a qualitative overview of dimensioning challenges regarding the FDD radio interface and multi-service traffic mix is provided.


5 / 61

INTERNAL


Internal: These rules are implemented in the dedicated LTE tools used by Network Designers: “Alcatel-Lucent LTE Link Budget” for FDD and TDD link budget analysis, “9955 and ACCO” for radio network planning studies and “LTE eNode-B Dimensioning Tool” for air interface capacity and modem dimensioning.


6 / 61

INTERNAL


References [1]

Jakes W.C., “ Microwave Mobile Communications”, IEEE Press, 1994

[2]

K.M Rege, S. Nanda, C.F. Weaver, W.C. Peng, “ Analysis of Fade Margins for Soft and Hard Handoffs”, PIMRC, 1996

[3]

K.M Rege, S. Nanda, C.F. Weaver, W.C. Peng, “Fade margins for soft and hard handoffs”, Wireless Networks 2, 1996


7 / 61

INTERNAL


1

INTRODUCTION

This document forms one part of a series of network dimensioning guidelines, as detailed in Table 1.

Table 1: Design Topics Covered in the LTE Dimensioning Guidelines Package Design Topic

Document

Deployment Strategy

LTE Dimensioning Guidelines - Deployment Strategy

Radio Features

LTE Dimensioning Guidelines – Radio Features

Outdoor Link Budget

LTE Dimensioning Guidelines – Outdoor Link Budget

Indoor Link Budget

LTE Dimensioning Guidelines – Indoor Link Budget

Peak Throughput

LTE Dimensioning Guidelines – Peak Throughput

Radio Network Planning

LTE Dimensioning Guidelines – RNP

Air Interface Capacity

LTE Dimensioning Guidelines – Air Interface Capacity

eNode-B Dimensioning

LTE Dimensioning Guidelines – Modem

Token & Licensing Dimensioning

LTE Dimensioning Guidelines – Token & Licensing

S1/X2 Dimensioning

LTE Dimensioning Guidelines – S1 & X2

Frequency Reuse Considerations

LTE Dimensioning Guidelines – Frequency Reuse

Diversity & MIMO

LTE Dimensioning Guidelines – Diversity & MIMO

Traffic Power Control

LTE Dimensioning Guidelines – Power Control

Traffic Aggregation Modeling

LTE Dimensioning Guidelines – Traffic Aggregation Modeling

The purpose of this document is to detail the formulation of Alcatel-Lucent’s LTE link budget for outdoor macro cellular deployments. Link budgets are used by Alcatel-Lucent primarily to derive the expected LTE performances at cell edge on the uplink and compare them with legacy systems in the case of an overlay of an existing network. This enables the estimation of the proportion of sites that can be reused (additional constraints such as space for hardware deployment, etc, have to be considered on top of this) and/or the required number of sites for a Greenfield operator. Figure 2 illustrates the main inputs and outputs for an LTE link budget coverage analysis.


8 / 61

INTERNAL


Coverage Inputs • Area to be covered • Targeted service at cell edge • Indoor penetration level

Network Information

Coverage Outputs

Link Budget RF Planning

• Cell Range •Legacy Site Reuse

• Incumbent network info

•Number of Sites

• LTE Frequency • LTE Maximum bandwidth

Figure 2: Link Budget Coverage Analysis Inputs/Outputs Key factors influencing the link budget analysis include the frequency band for LTE operation, the cell edge performance requirements, and the depth of coverage expectations.


9 / 61

INTERNAL


2

UPLINK LINK BUDGET

On the uplink, a cell is generally dimensioned by its coverage, the maximum cell range at which a mobile station is received with enough quality by the base station.

MAPL

Max UE transmit Power

Required Received Signal cell radius

Figure 3: Uplink Link Budget Concept The signal threshold at which a signal is received with enough quality is called the eNode-B receive sensitivity. This sensitivity figure will depend upon the:    

Data rate targeted at cell edge, Target quality / HARQ operating point (such as Block Error Rate (BLER), maximum number of retransmissions), Radio environment conditions (multipath channel, mobile speed), eNode-B receiver characteristics (Noise Figure).

As for 2G and 3G systems, the uplink link budget involves the calculation of the Maximum Allowable Propagation Loss (or Pathloss), denoted as the MAPL, that can be sustained over the link between a mobile at cell edge and the eNode-B, while meeting the required sensitivity level at the eNode-B. As for 2G/3G systems, the uplink link budget calculations consider all the relevant gains and losses encountered on the link between the mobile and the eNode-B. The uplink link budget is formulated such that one service (UL_Guar_Serv) is targeted at the cell edge, while for more limiting service rates, link budgets are formulated under the assumption they are not guaranteed at cell edge but at a reduced coverage footprint, as is illustrated in Figure 4).


10 / 61

INTERNAL


UL Rates 128kbps 256kbps 512kbps

RangeUL_Guar_Serv

Figure 4: Rationale behind the Uplink LKB Formulation

2.1 Uplink Link Budget Parameters The power, Cj(UL), received at the eNode-B from a mobile (UE) located at cell edge transmitting with its maximal power, P MaxTX_PUSCH, is given by: C j(UL) dBm

=

PMaxTX_PUSCHdBm

−

+

Gain TxdB

M arg inPenetratio ndB

+

−

Loss Tx dB

(GainRx

dB

−

− LossRx

LossesPropagationdB R Service(UL) dB

− LossBody

dB

)

where 

PMaxTX _ PUSCHdBm is the maximum transmit power of the UE (see section 2.1.1)



GainTx and LossTx, the gains and losses at the transmitter side such as UE antenna gain GainRx and LossRx represent the gains and losses at the receiver side such as the eNode-B antenna gain and the feeder losses between the eNode-B and the antenna LossBody is the body losses induced by the user, typically 3dB body losses are considered for voice services and 0 dB for data services (handset position is far from the head when using data services) MarginPenetration is the losses (in dB) induced by buildings, windows or vehicles according to the penetration coverage objective (deep or light indoor, outdoor) (see section 2.1.11) Assuming a Hata-like propagation model, the propagation losses can be expressed according to the cell range, Losses Propagation (see section 2.2.1):

 





LossesPropagationdB R Service(UL)

=

K1(UL) + K 2(UL) ⋅ Log10 R Service(UL) .

To ensure reliable coverage, the received power at the eNode-B should be higher than the eNode-B receiver sensitivity (see section 2.1.2): C j(UL)dBm

≥

Sensitivity dBm

+

MarginIoTdB

+

MarginShadowingdB

−

GainHOdB

−

GainFSS dB

where   

MarginIoT is a margin accounting for inter-cell interference (see section 2.1.7) MarginShadowing is a margin that compensates for the slow variability in mean path loss about that predicted using the propagation model, e.g. Hata (see section 2.1.8) GainHO is a handoff gain or best server selection gain that models the benefits due to the ability to reselect to the best available serving site at any given location (see section 2.1.9)


11 / 61

INTERNAL

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET> 

GainFSS is a frequency selective scheduling gain that is due to the ability of the scheduler to select best frequency blocks per UE depending on their channel conditions

For each service to be offered by the operator, this relationship allows computation of the maximum propagation losses that can be afforded by a mobile located at the cell edge, that is to say the Maximum Allowable Path Loss (MAPL): MAPL j(UL) dB

=

PMaxTX_PUSCHdBm

+

Gain TxdB

−


−

MarginShadowing dB

−

+

−

Loss Tx dB

Sensitivity dBm

GainHOdB

+

+

GainRxdB

−

− LossRx

dB

− LossBody

dB

MarginIoTdB

GainFSSdB

2.1.1 UE Characteristics The maximum transmit power of an LTE UE, P MaxTX_PUSCH, depends on the power class of the UE. Currently, only one power class is defined in 3GPP TS 36.101: 

A 23dBm output power is considered with a 0 dBi antenna gain.

Internal: This is the case in the TS 36.101 version of January 2011. Only one class defined (Class 3) with 23dBm output power (with ±2dB tolerance, but we should not account for such a tolerance to define the UE output power).

2.1.2 eNode-B Receiver Sensitivity The sensitivity level can be derived from SINR figures calculated or measured for some given radio channel conditions (multipath channel, mobile speed) and quality target (e.g. 10-2 BLER):

Sensitivity dBm

=

SINRPUSCH_dB

+ 10 ⋅ Log10

FeNode_B.Nth .NRB(UL).WRB

where:     



SINRPUSCH_dB is the signal to interference ratio per Resource Block, required to reach a given PUSCH data rate and quality of service, FeNode-B.Nth.NRB(UL).WRB is the total thermal noise level seen at the eNode-B receiver within the required bandwidth to reach the given data rate, where: FeNode-B is the noise figure of the eNode-B receiver, Nth is the thermal noise density (-174dBm/Hz), NRB(UL) is the number of resource blocks (RB) required to reach a given data rate – it can be deduced from link level simulations selecting the best combination (e.g. the one that requires lowest SNR or lowest number of RB to maximize the capacity), WRB is the bandwidth used by one LTE Resource Block. One Resource Block is composed of 12 subcarriers, each of a 15kHz bandwidth – so WRB is equal to 180kHz.

2.1.3 Noise Figure The Noise Figure of the eNode-B is supplier dependent. Typically the Noise Figures of an eNode-Bs is 2.5dB.

Internal: Assumed Noise Figures for ALU RRH product variants (September 2010).


12 / 61

INTERNAL


Frequency Band

Typical Noise Figure

700 MHz

2.5 dB

800 MHz

2.1 dB

850 MHz

2.1 dB

900 MHz

2.1 dB

1800 MHz

2.0 dB

1900 MHz

2.0 dB

AWS

2.0 dB

2100 MHz

2.0 dB

2600 MHz

2.0 dB

Internal: These figures are dependent on the specific hardware realization and as such within a given frequency band there will be variation between different product variants. For precise figures it is recommended to verify the latest figures with LTE Portfolio Management.

2.1.4 SINR Performances The SINR figures are derived from link level simulations or better from equipment measurements (lab or on-field measurements). They depend on the eNode-B equipment performance, radio conditions (multipath fading profile, mobile speed), receive diversity configuration (2 branch by default and optionally 4 branch), targeted data rate and quality of service.

2.1.4.1 Multipath Channel For link budget analysis, the most typical UE speed and multipath profiles are considered according to the type of environment (e.g. dense urban, rural, etc). In terms of multipath channel, the dense urban, urban or suburban indoor Macrocell deployment environments are consider to be well characterized by the ITU Vehicular multipath profile, with mobiles moving at 3km/h and 50km/h for rural environments. Choosing one multipath channel for a given environment is a modeling assumption. In reality, in a cell, various multipath conditions exist. A better representation would be to consider a mix of multipath channel models (even though there is no one unique mix to represent a typical Macro cell environment that has been agreed across the radio community). However for a coverage assessment, the worst case model should be considered. The ITU VehA multipath channel model (2 equivalent main paths) is correspondingly a good compromise for a reasonable, worse case, link budget analysis. For LTE some evolved multipath channel models have been defined such as EVA5Hz or EPA5Hz. These are an extension of the VehA and PedA models used in UMTS to make them more suitable for the wider bandwidths encountered with LTE, e.g. >5MHz. Main difference lies in the definition of a doppler frequency instead of a speed, making the model useable for different frequency bands. Typical SINR performances used in Alcatel-Lucent link budgets are for EVehA3 and EVehA50 channel models. For the purposes of the link budget the underlying assumption is that the UE is at the cell edge and the main driver is to maximize the coverage.


13 / 61

INTERNAL


2.1.4.2 Number Resource Blocks & Modulation & Coding Scheme For a given target data rate the required target SINR depends upon (see Figure 5 for some definitions of the LTE channel structure):  

Number Resource Blocks, NRB Modulation & Coding Scheme Index (MCS) subframe Physical Resource Block (RB) = 14 OFDM Symbols x 12 Subcarrier

f

one Subcarrier

This is the minimum unit of allocation in LTE

one OFDM symbol

RB 15 kHz

Slot (0.5 ms)

t

Slot (0.5 ms)

Subframe (1 ms)

Figure 5: LTE Channel Structure - Some Definitions The Modulation & Coding Scheme Index (MCS) determines the Modulation Order which in turn determines the Transport Block Size (TBS) Index to be used (see Table 2).

Table 2: Extract from the Modulation and TBS index table for PUSCH (from 36.213) MCS Index, IMCS

Modulation Order, Q M

TBS Index, ITBS

0

QPSK

0

1

QPSK

1

2

QPSK

2

3

QPSK

3

…

…

…

For a given MCS Index the Transport Block Size (TBS) is given by Table 3 for different numbers of resource blocks


14 / 61

INTERNAL


Table 3: Extract from the Transport Block size table (from 36.213) ITBS

NRB = 1

NRB = 2

NRB = 3

NRB = 4

NRB = …

0

16

32

56

88

…

1

24

56

88

144

…

2

32

72

144

176

…

3

40

104

176

208

…

4

56

120

208

256

…

5

72

144

224

328

…

6

328

176

256

392

…

…

…

…

…

…

…

For example, for an MCS Index = 2 and NRB = 3 the corresponding TBS = 144 bits.

2.1.4.3 Hybrid Automatic Repeat request (HARQ) A key characteristic of the LTE air interface is the utilization of HARQ, a combination of ARQ and channel coding which provides greater robustness against fast fading; these schemes include incremental redundancy, whereby the code rate is progressively reduced by transmitting additional parity information with each retransmission. In LTE, asynchronous adaptive HARQ is used for the downlink, and synchronous HARQ for the uplink. In the uplink, the retransmissions may be either adaptive or non-adaptive, depending on whether new signaling of the transmission attributes is provided. In an adaptive HARQ scheme, transmission attributes such as the modulation and coding scheme, and transmission resource allocation in the frequency domain, can be changed at each retransmission in response to variations in the radio channel conditions. In a nonadaptive HARQ scheme, the retransmissions are performed without explicit signaling of new transmission attributes – either by using the same transmission attributes as those of the previous transmission, or by changing the attributes according to a predefined rule. Accordingly, adaptive schemes bring more scheduling gain at the expense of increased signaling overheads. There are multiple HARQ operating points that can be utilized for an LTE system: 



Either, a lower initial BLER with a correspondingly fewer overall number of HARQ transmissions, resulting in a higher SINR requirement with reduced latency and better spectral efficiency (e.g. 10% iBLER target for the 1st HARQ transmission) Or, a higher initial BLER with a correspondingly greater overall number of HARQ transmissions resulting in a lower SINR requirement with an increased latency and poorer spectral efficiency (e.g. 1% pBLER target after up to 4 HARQ transmissions – iBLER ~50-70%).

The former operating point is currently recommended by Alcatel-Lucent, this corresponds to a 10% iBLER target for the 1st HARQ transmission.

Internal: Ideally the later operating point is considered at cell edge locations (for which we perform the link budget) where the objective is to tradeoff spectral efficiency and latency for an improved SINR and receiver sensitivity. Whereas in locations that are not link budget constrained, e.g. closer to the eNode-B, the former HARQ operating point is more appropriate. The current Alcatel-Lucent implementation considers only a 10% iBLER, eventually a different operating point is likely to be supported, maybe even a dynamic operating point. Alcatel-Lucent – Proprietary Version 2.9

15 / 61

INTERNAL


2.1.4.4 Selection of Optimal MCS Index & N RB For each targeted uplink data rate there will be an optimal combination of NRB and MCS Index that will maximize the receiver sensitivity for the relevant HARQ operating point. Figure 6 provides an example of the selection of the optimal MCS and number of RB, N RB, for a given target effective data rate. This plot illustrates for the full range of possible MCS indices the corresponding required NRB and the resultant eNode-B receiver sensitivity.

y t i v i t i s n e S x R B e d o N e

-90.0 dBm

7 RB

-95.0 dBm

6 RB

-100.0 dBm

5 RB

MCS 2 provides the optimal tradeoff between Rx. Sens and NRB required

-105.0 dBm

4 RB

-110.0 dBm

3 RB

-115.0 dBm

2 RB

-120.0 dBm

1 RB

MCS 0

MCS 5

MCS 10

MCS 15

MCS 20

MCS 25

Figure 6: Selection of Optimal MCS and N RB for a target rate of 128kbps with 10% iBLER, EVehA3 From Figure 6 it can be seen that MCS 2 with 3 RB’s is optimal, as this provides the best receiver sensitivity while minimizing utilization of RB’s. Table 4 provides an example of comparison between the 10% iBLER operating point performance with that for a 1% pBLER operating point, for the same 128kbps target effective data rate:

Table 4: Example of Different HARQ Operating Points (128kbps) 1% pBLER (high initial BLER)

10% iBLER (low initial BLER)

MCS 9

MCS 2

2 RB

3 RB

296 bits

144 bits

0.606

0.212

Post HARQ Throughput

128 kbps

128 kbps

Required SINR

-0.5 dB

0.2 dB

-116.9 dBm

-114.4 dBm

MCS Index NRB TBS Size Effective Coding Rate

Receiver Sensitivity (NF=2dB)


16 / 61

INTERNAL

MCS 30

c i v r e S r o f B R # d e r i u q e R


Note: The 1% pBLER HARQ operating point (1% BLER after 4 HARQ Tx) corresponds to an iBLER (BLER for the 1st HARQ transmission) much greater than 10%. It can be seen from the example summarized in Table 4, that the same required data rate can be achieved with different combinations of N RB, MCS Index and number of HARQ transmissions. The receiver sensitivity comparison below highlights the different coverage for the same targeted data rate due to the different HARQ operating points: 

Sensitivity dBm



Sensitivity 1% BLER after 4 HARQ Tx = -0.5 + 10xlog10( 2.0dBxNthx2RBx180kHz ) = -116.9dBm Sensitivity 10% BLER after 1 HARQ Tx = 0.2 + 10xlog10( 2.0dBxNthx3RBx180kHz ) = -114.4dBm



=

SINR PUSCH_dB

+ 10log10


While the two solutions require a relatively similar SINR, they utilize a different number of resource blocks, NRB. The trade-off between the two is a combination of the required bandwidth (number of resource blocks) and the number of HARQ transmissions versus the receiver sensitivity. 



While the utilization of more HARQ transmissions enhances (reduces) the required SINR for an equivalent MCS, it also requires the same air interface resources for a longer period of time (more transmission time intervals). Utilizing more resource blocks degrades the receiver sensitivity due to an increased noise bandwidth (180 kHz x number of resource blocks).

Note that the difference between the receiver sensitivities in Table 4 is due to the difference in the required SINR and the difference in the number of resource blocks. Figure 7 shows an identical analysis to that presented in Figure 6 with the exception that here an effective data rate of 512kbps is targeted.

y t i v i t i s n e S x R B e d o N e

-90.0 dBm

26 RB

-95.0 dBm

21 RB

-100.0 dBm

MCS 3 provides the optimal tradeoff between Rx. Sens and NRB required

16 RB

-105.0 dBm

11 RB

-110.0 dBm

6 RB

-115.0 dBm MCS 0

1 RB MCS 5

MCS 10

MCS 15

MCS 20

MCS 25

Figure 7: Selection of Optimal MCS and NRB for a target rate of 512kbps with 10% iBLER, EVehA3 From Figure 7 it can be seen that now MCS 3 with 10 RB’s is optimal as this provides the best receiver sensitivity while minimizing utilization of RB’s. Table 5 provides a comparison between the 10% iBLER operating point performance with that for a 1% pBLER operating point, for the same 512kbps target effective data rate: Alcatel-Lucent – Proprietary Version 2.9

17 / 61

INTERNAL

MCS 30

c i v r e S r o f B R # d e r i u q e R


Table 5: Example of Different HARQ Operating Points (512kbps) 1% pBLER (high initial BLER)


MCS 8

MCS 3

8 RB

10 RB

1096 bits

568 bits

0.530

0.224

Post HARQ Throughput

512 kbps

512 kbps

Required SINR

-0.8 dB

0.2 dB

-111.2 dB

-109.2 dB

MCS Index NRB TBS Size Effective Coding Rate

Receiver Sensitivity (NF=2dB)

Making the same comparison of the receiver sensitivity: 

Sensitivity dBm



Sensitivity 1% BLER after 4 HARQ Tx = -0.8 + 10xlog10( 2.0dBxNthx8RBx180kHz ) = -111.2dBm Sensitivity 10% BLER after 1 HARQ Tx = 0.2 + 10xlog10( 2.0dBxNthx10RBx180kHz ) = -109.2dBm



=

SINR PUSCH_dB

+ 10log10


Here the difference between the receiver sensitivities is due to the combination of the differences in the required SINR and in the required bandwidth (dictated by the number of resource blocks, NRB). Thus it is important when comparing the required SINR for two services to consider also the required number of resource blocks.

2.1.4.5 Typical SINR Performances Based on link level simulations, for a HARQ operating point that targets 1% pBLER, the optimal combination of NRB, MCS Index and the corresponding SINR target for the typical data rates considered in Alcatel-Lucent uplink link budgets are summarized in Table 6 and Table 7 for EVehA3 and EVehA50 channel conditions respectively with 2-way Rx Diversity.

Table 6: Typical Rates Considered in Uplink Link Budget for EVehA3 channel conditions @ 700MHz with 2.5dB Noise Figure, 1% post HARQ BLER Post HARQ Peak T’put

9.3 kbps

64 kbps

128 kbps

256 kbps

512 kbps

1000 kbps

2000 kbps

MCS Index

MCS 0

MCS 9

MCS 9

MCS 8

MCS 8

MCS 6

MCS 4

Modulation

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

NRB(UL)

1 RB

1 RB

2 RB

4 RB

8 RB

20 RB

45 RB

HARQ Operating Point

1% pBLER

1% pBLER

1% pBLER

1% pBLER

1% pBLER

1% pBLER

1% pBLER

Initial BLER

52.3%

78.9%

79.5%

75.2%

78.8%

80.4%

51.3%

TBS Size

16 bits

136 bits

296 bits

536 bits

1096 bits

2088 bits

3240 bits

Effective Coding Rate

0.152

0.606

0.606

0.53

0.530

0.400

0.275

Average # HARQ Tx

1.71

2.13

2.31

2.09

2.14

2.09

1.62

SINR Target

-5.9 dB

-0.5 dB

-0.5 dB

-0.9 dB

-1.1 dB

-2.5 dB

-3.7 dB

Rx Sensitivity

-124.8 dBm

-119.4 dBm

-116.4 dBm

-113.9 dBm

-111.0 dBm

-108.5 dBm


18 / 61

INTERNAL

-106.1 dBm


Table 7: Typical Rates Considered in Uplink Link Budget for EVehA50 channel conditions @ 700MHz with 2.5dB Noise Figure, 1% post HARQ BLER Post HARQ Peak T’put

7.5 kbps

64 kbps

128 kbps

256 kbps

512 kbps

1000 kbps

2000 kbps

MCS Index

MCS 0

MCS 6

MCS 7

MCS 10

MCS 10

MCS 10

MCS 10

Modulation

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

NRB(UL)

1 RB

2 RB

3 RB

4 RB

8 RB

16 RB

32 RB


1% pBLER

1% pBLER

1% pBLER

1% pBLER

1% pBLER

1% pBLER

1% pBLER

Initial BLER

74.2%

86.2%

88.6%

95.6%

95.6%

95.6%

95.6%

TBS Size

16 bits

176 bits

328 bits

680 bits

1384 bits

2792 bits

5736 bits


0.152

0.379

0.444

0.667

0.667

0.667

0.682

Average # HARQ Tx

2.12

2.75

2.56

2.66

2.70

2.79

2.87

SINR Target

-6.4 dB

-2.5 dB

-2.2 dB

-0.6 dB

-0.9 dB

-1.4 dB

-1.7 dB

Rx Sensitivity

-125.3 dBm

-118.5 dBm

-116.3 dBm

-113.6 dBm

-110.8 dBm

-108.3 dBm

-105.5 dBm

Internal: If quoting SINR performances to customers the 10% iBLER figures (Table 8 and Table 9) should be presented (as they are more representative of current product characteristics) in preference to the 1% pBLER figures (Table 6 and Table 7). The above SINR figures have been derived from link level simulations which assume ideal scheduling and link adaptation, the reality in the field will not be as good. To compensate for such ideal assumptions, there are currently two key elements to the margins incorporated into in the SINR performances used in uplink budgets today: 



Implementation Margin: to account for the assumptions implicit in the link level simulations used to derive the SINR performances Currently considered to be ~1dB o o No variability is assumed for different environments or UE mobility conditions o Will be tuned based on SINR measurements (not yet performed) ACK/NACK Margin: to account for the puncturing of ACK/NACK onto the PUSCH A 1dB margin is applied for VoIP services and 0.5dB for higher data o throughputs

The SINR performances quoted in Table 6, Table 7 and subsequently in Table 8 and Table 9 account for the above mentioned implementation and ACK/NACK margins. Table 8 and Table 9 summarize the same for a 10% iBLER HARQ operating point.


19 / 61

INTERNAL


Table 8: Typical Rates Considered in Uplink Link Budget (for EVehA3 channel conditions @ 700MHz with 2.5dB Noise Figure, 10% iBLER) Post HARQ Peak T’put

14.5 kbps

64 kbps

128 kbps

256 kbps

512 kbps

1000 kbps

2000 kbps

MCS Index

MCS 0

MCS 5

MCS 2

MCS 5

MCS 3

MCS 4

MCS 5

Modulation

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

NRB(UL)

1 RB

1 RB

3 RB

4 RB

10 RB

16 RB

25 RB


10% iBLER

10% iBLER

10% iBLER

10% iBLER

10% iBLER

10% iBLER

10% iBLER

TBS Size

16 bits

72 bits

144 bits

328 bits

568 bits

1128 bits

2216 bits


0.152

0.364

0.212

0.333

0.224

0.273

0.339

Average # HARQ Tx

1.1

1.1

1.1

1.1

1.1

1.1

1.1

SINR Target (EVehA3)

-1.2 dB

2.8 dB

0.2 dB

1.9 dB

0.2 dB

0.4 dB

0.9 dB

Rx Sensitivity (EVehA3)

-120.2 dBm

-116.1 dBm

-113.9 dBm

-111.0 dBm

-108.7 dBm

-106.5 dBm

-104.1 dBm

Table 9: Typical Rates Considered in Uplink Link Budget (for EVehA50 channel conditions @ 700MHz with 2.5dB Noise Figure, 10% iBLER) Post HARQ Peak T’put

14.5 kbps

64 kbps

128 kbps

256 kbps

512 kbps

1000 kbps

2000 kbps

MCS Index

MCS 0

MCS 5

MCS 2

MCS 5

MCS 3

MCS 4

MCS 5

Modulation

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

QPSK

NRB(UL)

1 RB

1 RB

3 RB

4 RB

10 RB

16 RB

25 RB


10% iBLER

10% iBLER

10% iBLER

10% iBLER

10% iBLER

10% iBLER

10% iBLER

TBS Size

16 bits

72 bits

144 bits

328 bits

668 bits

1128 bits

2216 bits


0.152

0.364

0.212

0.333

0.224

0.273

0.339

Average # HARQ Tx

1.1

1.1

1.1

1.1

1.1

1.1

1.1

SINR Target (EVehA3)

-0.9 dB

3.2 dB

0.5 dB

2.4 dB

0.7 dB

1.1 dB

1.5 dB

Rx Sensitivity (EVehA3)

-119.9 dBm

-115.7 dBm

-113.7 dBm

-110.6 dBm

-108.2 dBm

-105.8 dBm

Figure 8 illustrates the receiver sensitivity figures quoted in Table 6, Table 7, Table 8 and Table 9 for 1%pBLER and 10% iBLER HARQ operating points and EVehA3 and EVehA50 km/h channel conditions.


20 / 61

INTERNAL

-103.5 dBm


-105 dBm

EVehA 3km/h - 10% iBLER EVehA 50km/h - 10% iBLER -110 dBm

y t i v i t i s n e S -115 dBm r e v i e c e R -120 dBm

EVehA 3km/h - 1% pBLER EVehA 50km/h - 1% pBLER

-125 dBm

10 kbps

100 kbps

1000 kbps

Uplink Average Effective Throughput

Figure 8: Receiver Sensitivity for Typical Rates Considered in Uplink Link Budget (for EVehA3 & EVehA50 channel conditions @ 700MHz with 2.5dB Noise Figure, 10% iBLER and 1% pBLER)

2.1.5 Handling of VoIP on the Uplink For VoIP, various approaches (L2 segmentation and TTI bundling) were discussed at 3GPP to offer good coverage performances of VoIP (see Figure 9). TTI bundling was adopted in 3GPP Rel8 (36.321). With TTI bundling, as opposed to RLC Segmentation, larger transport blocks are used. Relying on incremental redundancy, HARQ Transmissions are performed in consecutive TTIs without waiting for HARQ feedback. The HARQ receiver accumulates the received energy of all transmissions and responds with HARQ feedback only once after the entire bundle has been received and evaluated.

RLC Segmentation

4ms TTI Bundling

Figure 9: RLC Segmentation and 4ms TTI Bundling Operating Modes Alcatel-Lucent – Proprietary Version 2.9

21 / 61

INTERNAL


2.1.5.1 VoIP and TTI Bundling    

No segmentation of VoIP packets required Enhances link budget compared to transmission of a single packet by supporting more HARQ transmissions in short time period Not supported in initial UEs and product Otherwise known as VoIP with QoS

The VoIP packet size for an AMR 12.2 VoIP codec, after accounting for RLC and MAC overheads, is ~328 bits. The VoIP codec generates such packets with ~20ms periodicity. With 4ms TTI bundling each 328 bit VoIP packet is sent in 4 consecutive TTI’s with 4 different redundancy variants (think of this as doing 4 HARQ transmissions in successive TTI’s). These four transmissions can be sent up to a maximum of 4 times and on average 2 times. For each TTI, MCS Index 6 is utilized with a single RB. This yields a TBS (Transport Block Size) of 328 bits (MCS 6 & 1 RB is a special combination created especially for VoIP services). The average effective air interface rate for active transmission for an AMR 12.2 VoIP service over the air interface is 328 bits / 4 successive TTIs / 2 average transmissions = 41 kbps, with the maximum of 4 transmissions this drops to 20.5kbps. However, if we average the codec payload of 328 bits over the 20ms periodicity, the average throughput is 328 bits / 20ms = 16.4 kbps. Table 10 summarizes the VoIP with TTI bundling performance characteristics that are considered in UL budgets:

Table 10: VoIP with TTI Bundling (1% pBLER target, 2dB NF) AMR 12.2 Nominal Codec Rate

VoIP Packet Size (with overheads)

12.2 kbps

328 bits

MCS / N RB / SINR (EVehA3) Rx Sensitivity

MCS 6 / 1 RB / -3.4 dB -122.9 dBm

MCS / N RB / SINR (EVehA50) Rx Sensitivity

MCS 6 / 1 RB / -2.9 dB -122.4 dBm

2.1.5.2 VoIP and RLC Segmentation     

Segments VoIP packets into multiple smaller segments Enhances link budget compared to transmission of a single packet as the smaller segments result in a more favorable required MCS and NRB Substantially higher overheads in terms of required grants and signaling Otherwise known as “Over the Top” best effort VoIP Very poor link budget without substantial levels of segmentation

There are a wide range of possible VoIP codec’s that could be used for such solutions, e.g. G711 (64kbps) and G729 (8kbps), in fact it is possible to use RLC segmentation with an AMR 12.2 VoIP codec. Table 11 provides a summary of the required TBS size for, varying levels of segmentation for G.729 and G.711 VoIP codecs and IPv4 and IPv6.


22 / 61

INTERNAL


Table 11: “Over the Top” Best Effort VoIP Packet Sizes (with overheads) for Varying Levels of Segmentation VoIP Codec

G.729

G.729

G.711

G.711

IP Version

IPv4

IPv6

IPv4

IPv6

1 Way Segmentation

536 bit

696 bit

1664 bit

1824 bit

2 Way Segmentation

292 bit

372 bit

856 bit

936 bit

4 Way Segmentation

170 bit

210 bit

452 bit

492 bit

8 Way Segmentation

109 bit

129 bit

250 bit

270 bit

Note: The packet sizes summarized in Table 11 assume that Robust Header Compression (RoHC) is not utilized for these “over the top” VoIP services. For example, with 8 way segmentation, a G.729 codec and IPv4, a TBS = 109bits is required. This means that the UE must have 8 separate transmissions scheduled each of 109bits in size, during each 20mecs VoIP frame period. Without segmentation, the UE only requires a single transmission of 536 bits scheduled during each 20mecs VoIP frame period. Clearly less segmentation is less demanding on air interface resources. However, this comes at the expense of degraded receiver sensitivity, as is summarized in Table 12.

Table 12: “Over the Top” Best Effort VoIP Receiver Sensitivity for Varying Levels of Segmentation (for EVehA3 km/h, 2dB NF and 10% iBLER) VoIP Codec

G.729

G.729

G.711

G.711

IP Version

IPv4

IPv6

IPv4

IPv6

1 Way Segmentation

-108.7 dBm

-108.1 dBm

-104.6 dBm

-104.3 dBm

2 Way Segmentation

-110.8 dBm

-109.6 dBm

-107.4 dBm

-107.2 dBm

4 Way Segmentation

-113.5 dBm

-111.7 dBm

-109.5 dBm

-108.9 dBm

8 Way Segmentation

-114.5 dBm

-114.3 dBm

-111.4 dBm

-111.2 dBm

For the above mentioned example (G.729 & IPv4), the receive sensitivity ranges from 108.7dBm without segmentation to -114.5dBm with 8 way segmentation. Furthermore, comparing the receiver sensitivities in Table 10 and Table 12, the link budget benefits attributable to TTI bundling combined with more HARQ transmissions are immediately apparent, -122.9dBm for TTI bundled AMR 12.2 VoIP versus -114.5dBm for G.729, IPv4 and 8 way segmentation.

2.1.6 Uplink Explicit Diversity Gains The SINR performance figures considered by Alcatel-Lucent in uplink and downlink link budgets are based on link level simulations that already account for the corresponding transmit and receive diversity gains. For the uplink the default assumption is 1x2 receive diversity (2RxDiv), the gain associated with 2RxDiv is accounted for directly in the SINR figures. Table 13 summarizes the receive diversity gains observed from link level simulations performed for a range of different eNode-B receive antenna correlation assumptions.


23 / 61

INTERNAL


Table 13: Receive Diversity Gains From Link Level Simulations Low

Medium

High

4RxDiv Gain (QPSK)

4.2 dB

4.1 dB

3 dB

8RxDiv Gain‡ (QPSK)

7.5 dB

6.2 dB

5 dB

Correlation ‡

Power Combining Gain

3dB (4RxDiv) and 6dB (8RxDIv)

Spatial Diversity Gain

‡

Large

Medium

Small ‡‡

Relative to 2RxDiv performances

‡‡

MRC loss in highly correlated channels

It can be seen from Table 13 that 4RxDiv gains range from 3 to 4.2dB and 8RxDiv gains from 5 to 7.5dB. For high correlation conditions the 8RxDiv gains are less than that attributable to the power combining gain due to an MRC loss. Table 14 details the impact on the SINR figures considered by Alcatel-Lucent for link budget purposes for four different UL receive diversity schemes (these are aligned with the High correlation scenario from Table 13 with some additional margin):

Table 14: SINR and IoT Impact due to UL Receive Diversity Scheme UL Rx Diversity Scheme

SINR Impact

IoT Impact

1 RxDiv

-2.5 dB

+1 dB

2 RxDiv

0 dB

0 dB

4 RxDiv

+2.5 dB

-1 dB

8 RxDiv

+4.5 dB

-2 dB

For example, to account for 1x4 receive diversity (4RxDiv) on the uplink an additional 2.5dB gain is considered on the (2RxDiv) SINR figures from link level simulations. Also detailed in Table 14 is the assumed impact on the default average IoT (discussed more in section 2.1.7). The underlying assumption here is that the reduced SINR requirements associated with higher order receive diversity schemes leads to a reduced SINR for cell edge UEs which in turn corresponds to a reduction in the average IoT imposed on adjacent cells.

Internal: Currently we do not have simulations to strongly back these IoT reduction assumptions other than that which can be found at the following: https://sps.sg.alcatellucent.com/sites/Global Sales Organization/wreless_toolsanexptse/LTE Simulations WG/Shared Documents/03 System Level Simulations/2010_09 - UL - TDD - 8RxDiv vs 2RxDiv IoT Impact

2.1.7 Interference Margin Generally, sensitivity figures are derived considering only thermal noise. However, in a link budget analysis, the real interference, Ij(UL), should be considered and not only the thermal noise. This means that the received power, Cj(UL), should satisfy the following condition: C j(UL)dBm

≥

Sensitivity dBm

+

MarginInterference dB

where


24 / 61

INTERNAL


MarginInterferencedB

=

 Ij(UL)

10log 



+

N th W 

Nth W

 

The MarginInterference is the interference rise over that of thermal noise due to inter-cell interference. Nth is the thermal noise (-174 dBm/Hz) and W is the used PRB bandwidth (Hz). Note that the assessment of the interference margin is totally different from the classical relationship between uplink cell load and noise rise considered in CDMA and WCDMA systems. Ij(UL) is the interference due to adjacent cells utilizing the same PRB at the same time. Note that this interference could also be considered to comprise of external interference from other systems such as MediaFLO or DTC Channel 51. LTE resources are divided into resource blocks (set of OFDM symbols and frequencies). The interference per resource block will depend on the probability that resource blocks of same frequency are simultaneously used in the surrounding cells. However, LTE system is likely to be deployed with a frequency reuse of 1. The interference on a given resource block can therefore be high. Assessing the interference level enables the derivation of the interference margin to be accounted for in link budgets used for coverage (cell range) evaluation. In CDMA or WCDMA systems, the interference margin was derived from power control equations, these equations established a linkage between the number of users transmitting in the cell (or the cell load) to the interference margin (or noise rise). In LTE some specific power control schemes are defined with some flexibility in the definition of the parameters offering various power control strategies to be adopted and consequently impacting the interference margin, IoT, to be considered in link budget analyses. For overlay and Greenfield deployment scenarios different approaches can be adopted for selecting the system IoT target and tolerable adjacent cell RB loadings. 



For a pure 100% overlay, the inter-site distance of the incumbent system must be respected. The link budget enables the determination of the ideal IoT target so that the system can reach a given data rate at cell edge, o From this IoT target the tolerable RB loading of adjacent cells can be estimated. For a Greenfield network, there is more flexibility to set the IoT target versus the data rate expectations at cell edge. o This can be performed for a target RB loading for adjacent cells.

A typical IoT target considered in LTE link budgets is 3dB. Such an IoT target will have a corresponding loading for adjacent cells for the cell range computed using the link budget formulation presented in this document. The average IoT is dependent upon the cell edge data rate (SINR) that is targeted by UEs in adjacent cells.   

Higher cell edge SINR targeted by UEs in adjacent cells  Higher average IoT Larger cell sizes  Lower cell edge rates can be achieved by UEs in adjacent cells  Lower average IoT (e.g. NGMN Case 3) Smaller cell sizes  Higher cell edge rates can be achieved by UEs in adjacent cells  Higher average IoT (e.g. NGMN Case 1)

An example from some system level simulations performed under NGMN Case 3 conditions (a coverage/link budget limited scenario) is presented in Figure 10 (assuming 100% resource block loading, 10 UEs per sector, full buffer simulations, 10MHz bandwidth).


25 / 61

INTERNAL


7000 kbps 6000 kbps 5000 kbps

t u p h g 4000 u o r h T 3000 l l e C

kbps kbps

2000 kbps 1000 kbps 0 kbps 1.0 dB

1.5 dB

2.0 dB

2.5 dB

3.0 dB

3.5 dB

IoT Figure 10: NGMN Case 3 – Coverage limited scenario, 100% resource block loading, 10 UEs per sector, full buffer simulations Figure 10 illustrates the impact of allowing a different average IoT on the spectral efficiency of the uplink. It can be seen that for this particular scenario the optimal IoT is between 2.5 and 3dB. Such scenarios are more typical of deployments that are more coverage rather than interference limited which is typical of the cases commonly considered in link budget analyses. A further example performed under NGMN Case 1 conditions (an interference/capacity limited scenario) is presented in Figure 11 (assuming 100% resource block loading, 10 UEs per sector, full buffer simulations, 10MHz bandwidth).

10000 kbps 9000 kbps 8000 kbps t u p h g u o r h T l l e C

7000 kbps 6000 kbps 5000 kbps 4000 kbps 3000 kbps 2000 kbps 1000 kbps 0 kbps 0 dB

5 dB

10 dB

15 dB

20 dB

IoT Figure 11: NGMN Case 1 – Interference/capacity limited scenario, 100% resource block loading, 10 UEs per sector, full buffer simulations Alcatel-Lucent – Proprietary Version 2.9

26 / 61

INTERNAL


Figure 11 illustrates the impact of allowing a different average IoT on the spectral efficiency of the uplink. It can be seen that for this particular scenario the optimal IoT is greater than 5dB. However, in this case the link budget is not constraining and thus from a link budget perspective there is no issue with tolerating a higher IoT. Note that while the simulations indicate there are gains to be had at IoTs of up to 15dB or more, operating points greater ~5.5dB are not currently recommended by Alcatel-Lucent. As was mentioned in section 2.1.6, when considering different receive diversity schemes, the default IoT recommendations are offset according to the figures recommended in Table 14.

2.1.8 Shadowing Margin From the previous section, the link budget should satisfy the following equation: C j(UL)dBm

≥

Sensitivity dBm

+

Margin Interference dB

This equation should be satisfied from a statistical point of view with a given probability, Pcov, (coverage probability) within the cell. Typically, the received power should be better than the sensitivity over more than 95% of the cell area: Proba C j(UL)dBm

≥

Sensitivity dBm

+

MarginInterference dB

≥

Pcov

Generally, a target of 95% cell coverage is considered in dense urban, urban and suburban environments, while 90% is considered in rural environments, but this is dictated by the operator’s coverage quality objectives. The received power from a mobile within the cell will depend upon the shadowing conditions due to obstacles between the UE and the base station antennas. These slow shadowing variations (in dB) can be represented as a Gaussian random variable with a zeromean and a standard deviation that is dependent upon the environment (typically between 5 to 10 dB). Due to the Gaussian properties of the shadowing, a margin called the “shadowing margin” can be computed and incorporated in the link budget calculations to consider the coverage probability requirement, either probability at cell edge or over the cell. The following formulas are used to derive the shadowing margins according to the specified coverage probability: Pcov cell border

Pcov cell area

=

=

 Margin Shadowing 1 1 − erfc 2 σ 2 

dB

1 1 + erf (a ) + e 2 

1+2ab b2

  

  1 + ab    1 − erf   b  

Where 

a



b=



=

MarginShadowing σ 2 K2

ln(10 )σ 2 K2 is the propagation model coefficient.

More details on the way these equations are derived can be found in [1].


27 / 61

INTERNAL


In order to guarantee a given level of indoor coverage, a penetration margin is considered in the link budget (see sections 2.1 and 2.1.11). Either this penetration margin is defined as a worst-case (e.g. 95th percentile value) value for which indoor coverage must be ensured or as an average penetration loss value with an associated standard deviation. In the former case, both variations of penetration and shadowing can be considered together through a single Gaussian random variable with the following composite standard deviation: σ

=

σ 2shadowing

+

2 σ penetratio n

In order to simplify the link budget it is recommended to consider the former approach, i.e. the penetration margin defined in Section 2.1.11 is therefore considered as a worst case value, without the requirement to consider any additional standard deviation. Table 15 summarizes some typical shadowing margins for a typical path loss slope, K2 =35:

Table 15: Example of Shadowing Margins Shadowing Standard Deviation 10 dB

8 dB

7 dB

6 dB

Cell Area Coverage Probability

Cell Edge Coverage Probability

Shadowing Margin

95%

87.7%

11.7 dB

90%

77.7%

7.7 dB

95%

86.2%

8.7 dB

90%

75.1%

5.4 dB

95%

84.9%

7.2 dB

90%

73.3%

4.3 dB

95%

83.9%

5.9 dB

90%

70.9%

3.3 dB

2.1.9 Handoff Gain / Best Server Selection Gain Unlike UMTS/WCDMA or CDMA, there is no soft-handoff functionality for LTE. Therefore, no soft-handoff gain should be considered for LTE. However it would be too pessimistic to only consider the shadowing margin computed with one cell as in section 2.1.8: a mobile at the cell edge can still handover to or originate a call on a neighboring cell with more favorable shadowing, i.e. a lower path loss. Some models have been derived to compute such a hard handoff gain, taking into account handoff hysteresis thresholds and connection delays [2] [3]. Such a model collapses to that of soft-handoff computations when the handoff threshold and the connection delays are equal to zero. It is also important to note that while this is referred to in the link budget as a “handoff gain” it could equally well be referenced as a “best server selection gain”. Note that this hard handoff gain can be considered for any system without soft handoff. So this is the case for GSM. Note that the handoff gain for LTE should be somewhere in between that which may be considered for GSM and that for a soft handoff scenario for WCDMA or CDMA. A shadowing margin, which is partially mitigated by the handoff gain, is only considered in the link budget due to uncertainties in the estimation of the path loss and cell range. As the uncertainty in the prediction of the path loss is reduced (a reduction in the standard deviation of shadowing) the shadowing margin and handoff gain will also be reduced. If Alcatel-Lucent – Proprietary Version 2.9

28 / 61

INTERNAL


there are no uncertainties in the estimation of the path loss and the corresponding cell range, there will be no need to consider any shadowing margin or handoff gain.

Internal: However we are not used to considering such a gain in GSM. It is highly recommended to consider such a hard handoff gain, above all to have favorable link budget comparison with CDMA or WCDMA, both of which consider a soft handoff gain in their link budgets. Table 17 provides some examples of the shadowing margin and handoff gain for different coverage probability targets and shadowing standard deviations. This example is based on the assumptions listed in Table 16:

Table 16: Assumptions for Hard Handoff Gain Computations Antenna Height

30 m

K2 Propagation Model

35.2

Shadowing Correlation

0.5

Hysteresis

3 dB

HO sampling time

20 msec

# of samples to decide HO

4 samples

Correlation distance

50 m

Note that the assumptions in Table 16 for the Hysteresis and HO sampling time are relatively conservative so as to ensure that the handoff gains considered in the LKB are evaluated with a reasonable degree of confidence.

Table 17: Example of Hard Handoff Gain Handoff Gain

Shadowing Standard Deviation

Cell Area Coverage Probability

Cell Edge Coverage Probability

Shadowing Margin

Soft Handoff Gain

3 km/h

50 km/h

100 km/h

6 dB

90%

71%

3.3 dB

2.7 dB

2.3 dB

2.1 dB

2.0 dB

6 dB

95%

84%

5.9 dB

2.8 dB

2.5 dB

2.2 dB

2.0 dB

7 dB

90%

73%

4.3 dB

3.1 dB

2.8 dB

2.6 dB

2.4 dB

7 dB

95%

85%

7.2 dB

3.4 dB

3.1 dB

2.8 dB

2.6 dB

8 dB

90%

75%

5.4 dB

3.6 dB

3.4 dB

3.1 dB

2.8 dB

8 dB

95%

86%

8.7 dB

3.9 dB

3.6 dB

3.3 dB

3.0 dB

10 dB

90%

78%

7.7 dB

4.7 dB

4.4 dB

4.1 dB

3.7 dB

10 dB

95%

88%

11.7 dB

5.0 dB

4.8 dB

4.4 dB

4.0 dB

Based on these results, a 3.6dB handoff gain can be assumed for typical DU, U and SU deployment conditions (95% area reliability, 8dB shadowing standard deviation and 3km/h) and 2.6dB in typical Rural conditions (90% area reliability, 7dB shadowing standard deviation and 50km/h). Note that the full handoff gain is only applicable for UE’s located at the cell edge. In the uplink link budget we consider one service (data rate) that is guaranteed at the cell edge, the more demanding services are supported in a subset of the coverage area. Consequently, the other services will not take benefit of the full handoff gain. Figure 12 illustrates the Alcatel-Lucent – Proprietary Version 2.9

29 / 61

INTERNAL


handoff gains computed for UE locations between the eNode-B and the cell edge. Note that this is an example for the same assumption as shown in Table 16 for a shadowing standard deviation of 8dB and 95% coverage reliability. 4.0 dB 3.5 dB 3.0 dB

n i a 2.5 dB G f f 2.0 dB o d n 1.5 dB a H

1.0 dB 0.5 dB 0.0 dB 0%

20%

40%

60%

80%

100%

% of Cell Range Figure 12: Handoff Gains for UE Locations between the eNode-B and the Cell Edge

2.1.10

Frequency Selective Scheduling (FSS) Gain

There are a number of ways the LTE system can manage the potentially considerably frequency selective channel:  

Schedule the best groups of RBs (Resource Blocks) to individual UEs according to the channel conditions for specific UEs (frequency selective scheduling) Make no specific consideration to the frequency selectivity o Frequency non-selective scheduling A variant upon this is to randomly hop frequencies (RBs) for retransmissions o and/or successive TTIs

For frequency selective scheduling, consider as an example, an uplink where an eNode-B is serving 3 contending UEs. For each UE, the eNode-B has knowledge of the quality of the radio channel (by means of the uplink SRS) and as such can form quality metrics for each individual RB for each UE on the UL. Based on these quality metrics the scheduler can formulate which resource block or group of resource blocks is most advantageous to allocate to each of the contending UEs on the uplink. This process is highlighted on Figure 13.


30 / 61

INTERNAL


12

8

7

10 6

8

5

4

c i r t e M y 6 t i r o i r P 4

Priority Metric

3

2

UE 1 1

UE 2 2

0

UE3 UE 2 UE 1

3 1

2

4

5

7

6

8

UE 3

9

0 1

PRB Index

2

3

4

5

6

7

8

9

PRB Index

Figure 13: Per UE quality metrics for each RB and the consolidated priority metric for each RB By allocation of the RB groupings according to the right hand diagram in Figure 13 it is possible to ensure that each UE is more likely to get allocated individual resource blocks that have more favorable channel conditions, thus resulting in enhanced link budget performances. This can be thought of a type of interference co-ordination scheme, whereby it is possible for the system to avoid interference by appropriate resource block allocation. A similar principle also applies on the downlink. One alternative to such a frequency selective scheduling approach is to consider only an average of the channel qualities across the entire band for each UE, see Figure 14.

6

5

4

3

Priority Metric

2

1

UE 1 UE 2 UE 3

0

UE 3 8

UE 2

6

9

7

5

UE 1

4 3 2

Resource Unit Index

1

Figure 14: Frequency Non-Selective Scheduling With such an approach the scheduler losses the ability to differentiate the best RB or group of RBs depending on the channel quality of individual resource blocks. Thus as a consequence the system can not take benefit of the corresponding link budget benefits. The gains attributable to frequency selective scheduling are dependent upon the channel model and the HARQ operating point. The gains can be estimated by means of system level simulations performed both with and without consideration of frequency selective scheduling. The difference in cell edge performances dictates the link budget gain that can be attributed to frequency selective scheduling. Table 18 summarizes the frequency selective scheduling gains, derived from simulations, for two HARQ operating points and three different channel models.


31 / 61

INTERNAL


Table 18: Frequency Selective Scheduling Gains Channel Model

1% pBLER (high initial BLER)


VehA3

0.5 dB

1.8 dB

VehA50

0.0 dB

0.0 dB

VehA120

0.0 dB

0.0 dB

Consider as an example from Table 18:   

10% iBLER HARQ operating point, VehA3 channel conditions FSS Gain = 1.8dB This means the throughput with FSS is 50% greater than the case without FSS

Note: the FSS gain is only applied for services in the UL link budget where the RB’s for the service can be allocated from all the available RB’s. For example the PUCCH and Attach link budgets do not benefit from this gain as the RB allocation is restricted.

2.1.11

Penetration Losses

The penetration losses characterize the level of indoor coverage targeted by the operator (deep indoor, indoor daylight, window, in-car, outdoor, etc). They are highly dependent on the wall materials and number of walls/windows to be penetrated. As mentioned earlier, Section 2.1.8, the penetration losses can be specified either as an average penetration loss with an associated standard deviation or as a single “worst case” penetration margin (the later case is recommended). Table 19 summarizes some typical “worst case” penetration losses for the most common frequency bands.

Table 19: Typical Penetration Losses for Common Frequency Bands

Frequency band

Penetration Margin Dense Urban

Urban

Suburban Indoor

Suburban Incar

Rural Incar

700 MHz

17 dB

14 dB

11 dB

5 dB

5 dB

800 MHz

17 dB

14 dB

11 dB

5 dB

5 dB

850 MHz

18 dB

15 dB

12 dB

6 dB

6 dB

900 MHz

18 dB

15 dB

12 dB

6 dB

6 dB

1800 MHz

20 dB

17 dB

14 dB

8 dB

8 dB

1900 MHz

20 dB

17 dB

14 dB

8 dB

8 dB

AWS

20 dB

17 dB

14 dB

8 dB

8 dB

2100 MHz

20 dB

17 dB

14 dB

8 dB

8 dB

2600 MHz

21 dB

18 dB

15 dB

9 dB

9 dB

2.2 Final MAPL and Cell Range The final uplink link budget equations become:


32 / 61

INTERNAL

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET> C j(UL) dBm

=

PMaxTX_PUSCHdBm

−

Gain Tx dB

+


+

−

Loss TxdB

(GainRx

−

− LossRx

dB

LossesPropagationdB dB

−

)

Loss Body dB

And C j(UL)dBm

≥

Sensitivity dBm

+

MarginIoTdB

+

MarginShadowingdB

−

GainHOdB

−

GainFSS dB

For each service to be offered by the operator, this relationship allows computation of the maximum propagation losses that can be afforded by a mobile located at the cell edge, that is to say the Maximum Allowable Path Loss (MAPL): MAPL j(UL) dB

=

PMaxTX_PUSCHdBm

+

Gain TxdB

−


−

MarginShadowing dB

−

+

−

Loss Tx dB

+

Sensitivity dBm

GainHOdB

+

GainRxdB

−

− LossRx

dB

− LossBody

dB

MarginIoTdB

GainFSSdB

Transmit Power

Max UE transmit Power Maximum Allowable Pathloss

Losses and Margins Gains

Interference margin extra cell interference

Reference Sensitivity

Gains - Losses- Margins Reference Sensitivity

Interference

cell radius

•

= MAPL

Figure 15: Uplink Link Budget Elements Considering the most demanding service for which contiguous coverage is to be offered, the following can be used to determine the maximum allowable cell range for deployment of the system: MAPL (UL)dB

=

Min

MAPL j(UL) dB

=

K 1(UL)

+ K 2(UL) ⋅ Log10

R Service(UL)

2.2.1 Propagation Model K1 and K2 characterize the propagation model. For Macro-cell coverage, the following propagation models are used: 

LossesPr opagation (R km ) = K 1 + K 2 ⋅ Log10 (R km )



For 700, 850 or 900 MHz, the Okumura-Hata model is used: o K 1 = 69.55 + 26.16 ⋅ Log10 (FMhz ) − 13.82 ⋅ Log10 (Hb ) − a(Hm ) + K c



For AWS, 1.9GHz or 2.1GHz band, the COST-231 Hata model is used: o K 1 = 46.3 + 33.9 ⋅ Log10 (FMhz ) − 13.82 ⋅ Log10 (Hb ) − a(Hm ) + K c



For 2.6GHz, a Modified COST-231 Hata model is used: The COST-231 Hata is limited to frequency between 1.5GHz and 2GHz. o Based on measurements at higher frequency (3.5GHz, 2.5GHz), AlcatelLucent proposed the following modified formula:  F  o K 1 = 46.3 + 33.9 ⋅ Log10 (2000 ) + 20 ⋅ Log10  MHz  − 13.82 ⋅ Log10 (Hb ) − a(Hm ) + K c  2000 


33 / 61

INTERNAL

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET> o



The Modified Cost-231 Hata model is only considered applicable for Suburban and Rural morphologies. For Dense Urban and Urban morphologies the Cost-231 Hata model is considered to be a better representation.

Where 44.9 − 6.55 ⋅ Log10 (Hb )

o

K2

o

a( H m ) = 3.2 ⋅ [(Log10 (11.75 xH m )]

o

a( H m ) = [1.1 ⋅ Log 10 (F MHz ) − 0.7] ⋅ H m

=

2

−

4.97 −

for K c

> −5

[1.56 ⋅ Log10 (FMHz ) - 0.8]

for K c

FMHz represents the operating frequency in MHz. Hb is the height of the eNode-B antenna in meters and Hm is the height of the UE antenna in meters (typically 1.5m). A morphology correction factor, Kc, is used depending on the type of environment, e.g. dense urban, urban, suburban, rural (typical values from calibration measurement campaign).

Internal: For the propagation model, it is always better to use a calibrated propagation model for the country or city you are studying – if a calibration measurement campaign is available. Otherwise, use the default morpho correction factors defined in the document “Clutter Classes For Radio Network Planning”.

2.2.2 Site Area Tri-sector sites are commonly used to offer 3G coverage and this is also the case for LTE.

Figure 16: Intersite Distance and Site Area The relationship between the cell range and the site area (3 sector sites) is defined by the following: SiteArea

=

9 3 2 R Service(UL) 8

= 1.95

R Service(UL)

2

The number of sites to cover a given area (due to coverage limitation) can then be derived.

Note: In the case of tri-sector configurations it is assumed that the antenna is tilted such that the antenna boresight is directed at the cell edge.


34 / 61

INTERNAL

≤ −5


2.3 Impact of RRH and TMA 2.3.1 RRH Remote Radio Heads (RRH) are a popular solution that enables to separate the RF part of the eNode-B and locate it physically close to the antenna, resulting lower feeder losses between the eNode-B and the antenna (lower losses on UL, more effective radiated power on the DL). Depending on where the RRH is located relative to the antenna, more or less losses have to be considered in the uplink link budget: 



At least 0.5dB losses should be considered due to the jumper required between the RRH and the antenna, applicable where the RRH is deployed very close to the antenna, Higher losses should be considered if the RRH is installed farther from the antenna (e.g. RRH at rooftop but still some non-zero length of feeder between the RRH and the antenna).

The other parameters of the link budget are not modified.

2.3.2 TMA Tower Mounted Amplifiers (TMA) (also called Mast Head Amplifiers (MHA) or Tower Top Low Noise Amplifiers (TTLNA)) can be used to enhance the uplink coverage of eNode-Bs with high feeder losses between the eNode-B and the antenna, allowing the required number of sites to be minimized (in the case of coverage-limited scenarios but not for capacitylimited scenarios) or allowing the reuse of incumbent 2G/3G sites to be maximized while offering higher data rates than in 2G/3G. For example, TMAs can be particularly beneficial if LTE is deployed in the 2.6GHz band, while incumbent 2G/3G sites were deployed in a lower band (e.g. 2GHz or even 850 or 900MHz), this allows the uplink LTE cell range, affected by higher propagation losses at the higher frequency, to be enhanced. As for any active element inserted in the reception chain of an eNode-B, the impact of a TMA on the link budget can be assessed by means of the Friis formula. n overall

=

n TMA

+

n feeder − 1 n eNode−B − 1 with n element + g TMA g TMA ⋅ g feeder

NFelement

=

10

10

Gelement

and g element

= 10

10

,

where NFfeeder = -Gfeeder = Feeder Losses. The typical TMA characteristics are NFTMA = 2dB, GTMA = 12dB and Insertion Losses = 0.4dB This has 2 key impacts on the link budget parameters:  

Compensation of the feeder losses, Reduction in the overall Noise Figure of the eNode-B.

However, TMA insertion losses of 0.4dB must be considered in the DL link budget. The typical gain on the MAPL for a 3dB feeder loss is approximately 2.7dB, which corresponds to 36% less sites, thanks to TMA usage. Note that such gains are only applicable for scenarios where uplink coverage remains as the limitation (i.e. low traffic scenarios).


35 / 61

INTERNAL


2.4 Uplink Budget Example Table 20 presents some example of the entire uplink budget analysis for a dense urban environment with deep indoor penetration for a range of different services.

The key objective of the air interface coverage analysis is to formulate a link budget from which the per-service MAPLs and the corresponding cell ranges can be computed (see the rows in red in Table 20). Table 20: Typical PUSCH link budgets for a tower mounted RRH deployment in Dense Urban VehA3 channel conditions at 700MHz (128kbps guaranteed at cell edge)

The cell ranges computed above are for a Hata propagation model for a 25m eNode-B antenna height, a 1.5m UE antenna height a clutter correction factor of 0dB. Where PL=K1+K2xlog10(dkm), K1=124.8 and K2=35.7.

Internal: The default ALU link budget can be found on the intranet: Alcatel-Lucent LTEFDD Link Budget. Based on the services to be guaranteed at cell edge the limiting Maximum Acceptable Path Loss (MAPL) can be derived.

2.5 Uplink Common Control Channel Considerations There are two main common and control channel considerations that should be assessed for an LTE network design to ensure that they will not limit the coverage. These include: Alcatel-Lucent – Proprietary Version 2.9

36 / 61

INTERNAL

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET>  



Attach Procedure (limiting message  RRC Connection Request) ACK/NACK Feedback Either punctured onto the Physical Uplink Shared Channel (PUSCH) o o Or over the Physical Uplink Control Channel (PUCCH) Periodic CQI Reports Over the Physical Uplink Control Channel (PUCCH) o

2.5.1 Attach Procedure Figure 17 illustrates the procedure that the UE must go through to Attach to an LTE network. From a link budget perspective the limiting message from messages 1, 2, 3, 4, 5, 15 and 16 (that involve the air interface) must be considered to assess any link budget constraints. UE

eNB

PGW

SGW

MME

RACH Preamble (1) Grant and TA (2) RRC Connection Request (3) RRC Connection Setup (4) RRC Connection Setup Complete (5) Attach request (6)

Authentication (optional)/ security (7-8) Create Default Bearer Request (9)

RRC Connection reconfiguration (14)

Attach accepted (13)

Create Default Bearer Response (12)

CDB Request (10) CDB Response (11)

RRC Connection reconfiguration complete (15)

Attach complete (16) 1st UL bearer packet

Update Bearer Request (20) Update Bearer Response (21)

1st DL bearer packet

No MME Relocation

Figure 17: LTE Attach Procedure The limiting message of the attach procedure over the air interface is message 3 (RRC Connection Request). This message utilizes 2 resource blocks with MCS 3, delivering an average effective data rate of 41.6 kbps after an average of 2.5 HARQ transmissions (maximum of 5). The SINR requirements for this message is -4.4 dB (including margins), based on link level simulation studies.


37 / 61

INTERNAL


Figure 18 summarizes an uplink budget formulated for a dense urban morphology in the 700MHz band. This link budget compares the Attach link budget with VoIP, 32, 64 and 128kbps services.

Figure 18: LTE Link Budget for Message 3 of the LTE Attach Procedure (compared with VoIP, 32, 64, 128kbps services) Note: For that the RRC Connection Request message can not be assigned to any available resource blocks on the uplink. As a consequence no frequency selective scheduling gain is considered for this link budget. It can be seen from Figure 18 that the Attach link budget is not limiting since equivalent to a 32kbps cell edge service.

2.5.2 ACK/NACK Feedback When users are receiving packets on the DL over the Physical Downlink Shared Channel (PDSCH) they must send steady streams of ACK/NACK transmissions over the UL to either acknowledge or not acknowledge the reception of the downlink packets. Correct reception of such ACK/NACK transmissions is critical for optimizing the efficiency of the DL transmissions. Alcatel-Lucent – Proprietary Version 2.9

38 / 61

INTERNAL


The first Alcatel-Lucent implementation implementation for such transmissions is to puncture the ACK/NACK transmissions onto the Physical Uplink Shared Channel (PUSCH). In the longer term it is expected to carry such transmissions over the Physical Uplink Control Channel (PUCCH). For either solution the ACK/NACK transmission utilizes 1 resource block with QPSK. The SINR requirements for this message are -1.7dB and -5.8dB for puncturing on the PUSCH and PUCCH, respectively (including margins), based on link level simulation studies. Figure 19 summarizes an UL link budget formulated for a dense urban morphology in the 700MHz band. This link budget compares the ACK/NACK link budgets for puncturing over the PUSCH and PUCCH options with VoIP, 32, 64, and 128kbps services.

Figure 19: LTE Link Budget ACK punctured o nto PUSCH and over PUCCH (compared with VoIP, 32, 64 and 128kbps services) Note: As the PUCCH only uses a subset of the uplink resource blocks no frequency selective scheduling gain is considered for the ACK/NACK over PUCCH link budget. However, this is not the case for ACK/NACK over PUSCH. From Figure 19 it can be seen that for either option (PUSCH or PUCCH) the ACK/NACK link budget does not limit the LTE coverage but rather coverage will be first limited by the UL service link budgets, e.g. VoIP AMR 12.2 or 32kbps.


/ 61 39 /

INTERNAL


2.5.3 Periodic CQI Reports The periodicity and frequency resolution to be used by a UE to report CQI are both controlled by the eNode-B. In the time domain, both periodic and aperiodic CQI reporting are supported. The Physical Uplink Control Channel (PUCCH) is used for periodic CQI reporting only; the Physical Uplink Shared Channel (PUSCH) is used for aperiodic reporting of the CQI, whereby the eNode-B specifically instructs the UE to send an individual CQI report embedded into a resource which is scheduled for uplink data transmission.

Internal: Alcatel-Lucent does not yet support CQI reporting over PUCCH (as at LA2.0) but this is planned for LA3.0 Focusing on the periodic CQI reports over the PUCCH, the most limiting 8bit CQI report utilizes 1 resource block with QPSK. The SINR requirements for this message ia -2.9dB (including margins), based on link level simulation studies. Figure 20 summarizes an UL link budget formulated for a dense urban morphology in the 700MHz band. This link l ink budget compares compare s the 8bit CQI link lin k budget over the PUCCH with VoIP, 32, 64, and 128kbps services.

Figure 20: LTE Link Budget for an 8bit CQI Report over the PUCCH (compared with VoIP, 32, 64 and 128kbps services) Alcatel-Lucent – Proprietary Version 2.9

/ 61 40 /

INTERNAL


Note: As the PUCCH only uses a subset of the uplink resource blocks no frequency selective scheduling gain is considered for the 8bit CQI report over PUCCH link budget. From Figure 20 it can be seen that the 8bit CQI link budget over the PUCCH does not limit the LTE coverage but rather coverage will be first limited by the UL service link budgets, e.g. VoIP AMR 12.2 or 32kbps.


41 / 61

INTERNAL


3

DOWNLINK LINK BUDGET

For typical eNode-B output powers and deployment scenarios with the classical UE output power class of 23dBm, link budgets should remain uplink limited. The downlink cell edge performances depend primarily upon the scheduler parameters (e.g. tuning of the fairness of the proportional fair scheduler algorithm) or the available bandwidth (e.g. 10MHz vs 5MHz). For the downlink, link budgets need to be carefully tuned with system level simulations to well assess the interference margin that is location dependent. The preferred approach by Alcatel-Lucent is to perform system level simulations to well assess the downlink performances with or without MIMO. Alcatel-Lucent extensively contributed to such systemlevel performances assessment at 3GPP and in the NGMN initiative. In addition to system level simulations it is the preferred practice of Alcatel-Lucent to rely upon Radio Network Planning (RNP) analysis. However, it is possible to formulate a reasonably meaningful downlink budget. The approach preferred by Alcatel-Lucent is as follows: 

 

Downlink cell range is defined by the uplink cell edge service link budget, i.e. the same cell ranges as those considered for the uplink are also considered for the downlink. On the uplink the objective was to compute the cell range for a target data rate, on the downlink the objective is to compute the data rate for a known cell range. Downlink throughputs computed for coverage reliabilities associated with each corresponding uplink service Geometry distributions (see section 3.1.3) are used to determine the cell edge SINR for the PDSCH, from which an estimate of the downlink cell edge throughput can be made

Figure 21 illustrates the downlink link budget approach utilized by Alcatel-Lucent. Section 2 described the methodology used to compute the cell range for different uplink services. Some examples of such services and their relative cell ranges are illustrated in blue in Figure 21. Also shown in Figure 21 are the downlink data rate estimates, illustrated in purple, corresponding to the various uplink data rates. UL Rates 128kbps (3RB) - guaranteed at cell edge 256kbps (4RB) 512kbps (9RB)

RangeUL UL_Guar_Serv _Guar_Serv

13152kbps (50RB) 7875kbps (50RB) 3786kbps (50RB) DL Rates

Figure 21: Rationale behind the Downlink LKB Formulation Alcatel-Lucent – Proprietary Version 2.9

42 / 61

INTERNAL


The example shown in Figure 21 is for an urban morphology, indoor 0dBi omni UE configuration, cell range fixed for uplink 128kbps, 50% adjacent cell downlink resource block loading, no TMA, 700MHz and 10MHz bandwidth. This example illustrates the concept behind downlink link budget approach that is described in this section.

Note: The diagram shown in Figure 21 is not to scale and does not include all rates.

3.1 Downlink Budget Parameters 3.1.1 SINR One of the measures of quality used on the downlink is the SINR. There are a number of important channels for which the SINR is of interest, these include:   

SINRPDSCH SINRPDCCH SINRDLRS

Of the above, the SINRPDSCH is the most commonly referenced downlink SINR metric.

Note: There is no consistent standard or industry defined measure of SINRPDSCH exists that is a completely unambiguous and can be used as a concise reference measure of downlink signal quality in the field. For example, the SINRPDSCH can be quantified both with and without inclusion of a combining gain at the UE (the default for Alcatel-Lucent is to incorporate such a combining gain). While Alcatel-Lucent link level performances are quantified in terms of the SINRPDSCH, the reference to be used, measured and validated in the field is the RSRQ (see section 3.1.2), which is more unambiguously defined. Unlike the uplink, the observed downlink SINR performances are dependent upon the UE location, i.e. the signal to interference plus noise ratio for the PDSCH, PDCCH or DLRS channels, SINRChannel_Des, is dependent on the user location. Thus, for a given UE location, SINRChannel_Des, for a number of transmit paths, PathDL, is given by: SINR Channel_Des

=

C j_Channel_ Des (DL) ⋅ PathsDL Ij(DL)

+

N(DL)

,

where Channel_Des is the desired channel for the SINR computation. This can be either PDSCH, PDCCH or DLRS.

Note: PathsDL = 1 when computing SINRDLRS as the DLRS is not transmitted on the same RE from all transmit paths. The worst performances will be experienced when the UE is at cell edge far from the eNode-B. The relationship between the SINR PDSCH, and downlink throughput is discussed in more detail in section 3.1.5. Cj_Channel_Des (DL) is the power of the considered channel, PDSCH, PDCCH or DLRS, received at the UE located at the uplink service cell range, R Service(UL), per Resource Element (RE) from the UE’s serving eNode-B, that is transmitting with its maximal power and is given by: C j_Channel_ Des (DL) dBm C j_DLRS(DL) dBm

= −

EPRE DLRS

+

Gain TxdB



+

−

=

C j_DLRS(DL) dBm

Loss TxdB

(GainRx

dB

−

+

Offset Channel

− LossesPropagatio n

dB

R Service(UL)

LossRx dB ) − MarginShadowingdB

43 / 61

INTERNAL

+

− LossBody

GainHOdB

dB


where  

  





 

EPREDLRS is the Energy Per Resource Element for the reference symbol (see Section 3.1.7) OffsetPDSCH / PDCCH / DLRS is the margin by which PDSCH, PDCCH and DLRS RE’s are offset in power from the EPREDLRS. OffsetDLRS is by definition always equal to 0dB. OffsetPDCCH is driven by dynamic PDCCH power control. GainTx and LossTx, represent the gains and losses at the transmitter side such as the eNode-B antenna gain and the feeder losses between the eNode-B and the antenna, GainRx and LossRx, the gains and losses at the receiver side such as UE antenna gain, LossBody is the body loss induced by the users proximity to the UE, typically 3dB body losses are considered for voice services and 0 dB for data services (handset position is far from the head when using data services), MarginPenetration is the loss (in dB) induced by buildings, windows or vehicles according to the penetration coverage objective (deep or light indoor, outdoor) (see Section 2.1.11), The downlink MAPL, MAPL(DL)dB, that corresponds to an uplink service cell range, RService(UL) (as computed in section 2.2), is dependent upon the propagation model differences (K1(DL) & K2(DL)) due to the frequency duplex difference between uplink and downlink, MarginShadowing is a margin that compensates for the slow variability in mean path loss about that predicted using the propagation model, e.g. Hata (see section 0) GainHO is a handoff gain or best server selection gain that models the benefits due to the ability to reselect to the best available serving site at any given location (see section 0) LossesPropagatio n

dB

=

K 1(DL)

+

K 2(DL)log(R Service(UL) ) .

Ij(DL), is the average received interfering power at the UE from all adjacent cells per RE. Averaging is based upon the average number of RE allocated to the various interfering channels. The channels considered as sources of interference, Channel_Int, are detailed in section 3.1.3, these can include a combination of the PDSCH, RS, PDCCH, SCH, BCH, PCFICH and PHICH. The average number of RE elements per TTI assigned to each channel type is detailed in Section 3.1.6. Ij(DL) dBm

 Ij_PDSCH(DL ) Avg + Ij_PDCCH(DL ) Avg ⋅ LoadingDL + =   Ij_DLRS(DL) + Ij_SCH(DL) + Ij_BCH(DL) + Ij_PCFICH&PHICH(DL)  Avg

Avg

Avg

Avg

  ⋅ Paths - Geometry DL Percentile  

+

M arg inShad _ SINR

where 

Ij_Channel_Int(DL) is the average interference contribution due to RE’s allocated to the various interfering channels (where Channel_Int can be PDSCH, PDCCH, SCH, BCH and PCFICH & PHICH) and is given by (Note: power offset for PDCCH interferences is assumed to be null to account for dynamic PDCCH power control): Ij_Channel_ Int(DL) Avg



=

(C

j_RS(DL) dBm +

Offset Channel ) ⋅

ConsideredChannel _ Int ⋅ REChannel _ TTI RE Total _ TTI

,

Ij_RS(DL) is the average interference contribution due to RE’s allocated to RS’s and is given by (Note: The 3rd and 4th antennas, if present, only transmit the RS on half the number of RE’s as the 1st two antennas): Ij_RS(DL) Avg

=

C j_RS(DL) ⋅ If (PathsDL


=

4, Then 0.75 Else 1) ⋅

44 / 61

INTERNAL

ConsideredDLRS _ Int ⋅ REDLRS _ TTI RE Total _ TTI

,

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET> 

RETotal_TTI, is the sum of the RE for the considered interfering channels and is given by (where ConsideredChannels_Int for PDSCH, DLRS, PDCCH, SCH, BCH and PCFICH & PHICH is defined in section 3.1.3 and depends upon the considered Channel_Des): RE Total _ TTI

∑ Considered

=

Channel _ Int ⋅

RE Channel _ TTI

All Channels



 

LoadingDL is the assumed average resource blocking loading of adjacent cells on the downlink. Note: It is assumed that the interference due to both PDSCH and PDCCH from adjacent cells will be reduced with reduced average RB loading. GeometryReliability, represents the downlink geometry that corresponds to the UL cell range, RService(UL) (discussed in more detailed in section 3.1.3), MarginShad_SINR is the shadowing margin applied to the SINR distribution to account for the fact that the desired and interfering signals are not perfectly correlated with each other (see section 0)

The thermal noise for a single subcarrier is given by: N(DL)dBm

=

Nth

+

FUE

+ 10 ⋅ Log 10

(WSC )

where   

Nth is the thermal noise density (-174dBm/Hz), FUE is the noise figure of the UE receiver (8dB by default), WSC is the bandwidth used by one subcarrier, each of a 15kHz bandwidth.

Internal: Note that currently there is no consideration of any frequency selective scheduling benefits in the downlink budget as there is in the uplink (see section 2.1.10). This does not mean such gains are not realized in the field on the downlink, but rather that such gains are not yet accounted for in the current downlink budget. This may be addressed in future link budget updates.

3.1.2 RSRQ While, SINRPDSCH, is a meaningful measure of the cell edge quality (see section 3.1.1), this is not a measure that is standardized by 3GPP and as such is somewhat open to interpretation when it comes to measurement in the field, i.e. whether a power combining gain is accounted for in the computation of SINR. The standardized measure of the downlink quality is RSRQ (Reference Symbol Receive Quality) and is given by: RSRQ Service(UL)

= RSRP + 10 ⋅ log10 (NRB ) − RSSI Total

where   

RSRP is the Reference Signal Received Power at the UE from its serving cell and is given by Cj_DLRS(DL) (see above) NRB is the maximum number of RB’s for the consider carrier bandwidth RSSITotal is the total received power at the UE from its serving cell and all adjacent cells across the entire bandwidth and is given by: RSSI Total

 

=

RSSIOwn _ Cell

+ Ij(DL) +

N(DL)

⋅ SubCarriers RB ⋅ NRB

SubCarriersRB is the number of sub carries per RB, this is defined by the standards to be 12 sub-carriers per RB RSSIOwn_Cell is the average power received at the UE from its serving cell per RE. The averaging is based upon the average number of RE allocated to the various interfering channels (see Section 3.1.6 for details of the RE distribution) and is given by:


45 / 61

INTERNAL

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET> RSSIOwn_Cell

= Ij(DL) +

Geometry Percentile

−

M arg in Shad _ SINR

Note: RSRQ is dependent upon the number of downlink transmit paths, PathsDL. Table 21 provides examples of RSRQ values in typical scenarios. Table 21: Typical RSRQ Values for # Tx Paths and Average Adjacent Cells RB Loading RSRQ

Load

1 Tx Path

2 Tx Paths

100%

-17 dB

-20 dB

50%

-14 dB

-17 dB

3.1.3 Interference Sources As mentioned in sections 3.1.1 and 3.1.2 the channels considered as sources of interference can include a combination of the PDSCH, RS, PDCCH, SCH, BCH, PCFICH and PHICH. The channels to be considered depend upon a number of factors including:  

Whether phase synchronization is available in the network The type of SINR or RSRQ computation

Table 22 summarizes the sources of interference considered for various types of SINR and RSRQ computations. When the network is not phase synchronized there is no guarantee that the same OFDM symbols of a given subframe from two different eNode-Bs will be aligned in time. As such it is possible that different RE (Resource Elements) of different channels will be consider as sources of interference.

Table 22: Sources of Interference for SINR and RSRQ Considered for Computations With and Without Phase Synchronization Interference Source, Channel_Int

ConsiderChannel_Int (Freq Sync.) SINR PDSCH SINR PDCCH

SINR RS

RSRQ

ConsiderChannel_Int (Freq + Phase Sync) SINR PDSCH SINR PDCCH

SINR RS

RSRQ

PDSCH

Yes

n/a

Yes

Yes

Yes

n/a

Yes

Yes

DLRS

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

PDCCH

Yes

Yes

Yes

Yes

n/a

Yes

Yes

Yes

SCH

Yes

Yes

Yes

Yes

n/a

n/a

n/a

Yes

BCH

n/a

Yes

Yes

Yes

n/a

n/a

Yes

Yes

PCFICH

Yes

Yes

Yes

Yes

n/a

Yes

Yes

Yes

PHICH

Yes

Yes

Yes

Yes

n/a

Yes

Yes

Yes

For example, from Table 22 consider the computation of the PDSCH SINR for a frequency and phase synchronized configuration. For such calculations only RE used by the PDSCH and DLRS will be considered as sources of interference. Whereas without phase synchronization all possible sources of interfering RE will be considered with the exception of those used for the BCH.

Internal: The default Alcatel-Lucent configuration is frequency synchronization only, i.e. no phase synchronization. Phase synchronization is required for services such as OTDOA Location Based Services (e.g. E911 services in North America), Handover to CDMA (HRPD) and eMBMS support. It is also important to understand that this has minimal impact on the SINR and RSRQ computations. Alcatel-Lucent – Proprietary Version 2.9

46 / 61

INTERNAL


3.1.4 Geometry The geometry at a specific UE location is defined as the ratio between the total power received from the eNode-B serving that location and the total power received from all adjacent eNode-Bs, under the assumption that all eNode-Bs are transmitting at the same power.

Figure 22: Signals Contributing to the Downlink Geometry (serving site is solid green and adjacent sites are dashed maroon color) The geometry at a given UE location is given by: Geometry

=

Rx PowerServing Site

∑ Rx Power

Adjacent Site

All

Consequently the geometry is influenced by the parameters such as the relative positioning of adjacent sites, degree of overlapping coverage, variability of the propagation environment and directivity of eNode-B and UE antennas. The geometry distributions considered in the link budget are based upon the geometry distributions computed with the 9955 Radio Network Planning (RNP) tool, for a range of LTE trial network deployments, across a number of markets. These geometry distributions are considered to be representative of the typical geometries that are expected in a well optimized LTE deployment.

Note: The downlink geometries do not account for lognormal shadowing as such an additional shadowing margin must be applied to SINR and RSRQ computations (see section 3.1.8) A significant factor influencing the geometry distribution is the directivity and placement of the UE antenna. While the majority of LTE deployments are focused on a typical cellular mobility deployment model there is also interest in considering fixed wireless deployment scenarios where it is not uncommon to consider a directional non-zero gain UE antenna that can be roof mounted and directed at the best serving site. Figure 23 provides some examples of the geometry distributions used in the downlink budget for omni directional as well as direction UE antenna configurations.


47 / 61

INTERNAL


100% 90% 80% 70% 60% 50% 40% Indoor - 0 dBi - Omni

30%

Outdoor - 4 dBi - Direc.

20%


10%


0% -5 dB

0 dB

5 dB

10 dB 15 dB 20 dB 25 dB 30 dB 35 dB 40 dB

Geometry

Figure 23: Geometry Distributions Considered in Link Budget (for different UE Antenna Configurations) In the computation of the Cell Edge SINR (as mentioned in section 3.1.1), the average received interfering power, Ij(DL), at the UE from all adjacent cells per RE is given by: Ij(DL) dBm

=

 I j_PDSCH(DL) Avg + Ij_PDCCH(DL) Avg ⋅ Loading DL +   I j_DLRS(DL) + I j_SCH(DL) + I j_BCH(DL) + I j_PCFICH&PHICH(DL)  Avg

Avg

Avg

Avg

  ⋅ Paths - Geometry DL Percentile  

+

M arg in Shad _ SINR

The percentile of the geometry distribution shown in Figure 23 is approximated to be dependent upon the targeted coverage reliability, PCov, and the percentage of the overall coverage area for which the downlink service is to be guaranteed. Percentile

=

PCov

⋅

R UL_Service(UL)

2

R UL_Guaranteed_Service(UL)

2

Where RUL_Service(UL) is the cell range for the uplink service for which the equivalent downlink data rate is being computed and RUL_Guaranteed_Service(UL) is the uplink service that is guaranteed at the cell edge on the uplink. See the example in Figure 24 (based on the uplink budget summarized in Table 20) where RUL_Guaranteed_Service(UL) = 1.24km is for a 128kbps cell edge service and the Percentile is computed for the UL cell range, R UL_Service(UL) = 0.98km, that corresponds to an uplink 256kbps service.


48 / 61

INTERNAL


UL Rates 128kbps (3RB) - guaranteed at cell edge 256kbps (4RB)

RangeUL_Guaranteed_Service =1.24km RangeUL_Service =0.98km

7860kbps (50RB) 3790kbps (50RB) DL Rates

Figure 24: Example of Geometry Percentile Computation for 256kbps UL Cell Range within a 128kbps Coverage Footprint In this example the cell area reliability is 95%. Thus the percentiles can be calculated as follows:  

For 128kbps uplink cell range, 95% x 1.242 / 1.242 = 95% For 256kbps uplink cell range, 95% x 0.982 / 1.242 = 59%

Referring to Figure 25, estimates of the corresponding geometries can be read off the chart for these two uplink cell ranges, i.e. percentiles of 95% and 59% yield GeometryPercentile values of -2.2 & 4.7dB, respectively, for a 0dBi omni UE configuration. 100% 90% 80% 70% 60% 50%

(100% - 59%) = 41%

40% Indoor - 0 dBi - Omni

30%


20% 10%

Outdoor - 8 dBi - Direc. Outdoor - 10 dBi - Direc.

(100% - 95%) = 5%

0% -5 dB

0 dB

-2.2dB

5 dB

4.7dB

10 dB 15 dB 20 dB 25 dB 30 dB 35 dB 40 dB

Geometry

Cell Edge

Cell Centre

Figure 25: Example of Geometry distribution


49 / 61

INTERNAL


3.1.5 Downlink SINR Performances The downlink SINR figures, like those for the uplink (see section 2.1.4), are derived from link level simulations or from equipment measurements (lab or field measurements). They depend on the UE performance, radio conditions (multipath fading profile, mobile speed), antenna scheme (TxDiv/SFBC, closed loop rank 1, spatial multiplexing, etc), targeted data rate and the quality of service. Figure 26 illustrates a sample set of link level simulation results for the full set of MCS Indices for a wide range of SINR conditions. 35000 kbps

30000 kbps

25000 kbps

t u 20000 kbps p h g u o r h 15000 kbps T

MCS 0

MCS 1

MCS 2

MCS 3

MCS 4

MCS 5

MCS 6

MCS 7

MCS 8

MCS 9

MCS 10

MCS 11

MCS 12

MCS 13

MCS 14

MCS 15

MCS 16

MCS 17

MCS 18

MCS 19

MCS 20

MCS 21

MCS 22

MCS 23

MCS 24

MCS 25

MCS 26

MCS 27

MCS 28

10000 kbps

5000 kbps

0 kbps -15 dB

-10 dB

-5 dB

0 dB

5 dB

10 dB

15 dB

20 dB

25 dB

SINR

Figure 26: Example of link level simulations results for downlink, N RB=50, 10MHz Bandwidth (Closed Loop Rank 1)

3.1.5.1 Multipath Channel The equivalent channel models to those consider on the uplink (see section 2.1.4.1) are also assumed on the downlink, i.e. EVehA 3km/h for dense urban, urban or suburban indoor Macrocell deployment environments and EVehA 50km/h for suburban incar and rural environments.

3.1.5.2 Number Resource Blocks & Modulation & Coding Scheme For the uplink, the focus was to determine the required SINR for a given target data rate (see section 2.1.4.4). For the downlink, the reverse is performed, the data rate that is achievable for a given SINR value is determined. However, the same principles apply. For a given SINR and Number of Resource Blocks, NRB, there will be an optimal Modulation & Coding Scheme Index (MCS) that maximizes the data rate while also satisfying the targeted HARQ operating point. The same process is used for determining the Transport Block Size (TBS) that corresponds to a given combination of NRB and MCS Index (as described in section 2.1.4.2, with the exception that the PDSCH version of the MCS to TBS index mapping is used instead of the PUSCH version shown in Table 2).

3.1.5.3 Hybrid Automatic Repeat request (HARQ) As mentioned in section 2.1.4.3, asynchronous adaptive HARQ is used for the downlink where transmission attributes such as the modulation and coding scheme, and transmission Alcatel-Lucent – Proprietary Version 2.9

50 / 61

INTERNAL


resource allocation in the frequency domain, can be changed at each retransmission in response to variations in the radio channel conditions. Like the uplink there are multiple HARQ operating points that can be utilized (with the corresponding tradeoffs), the current recommended operating point for the downlink is a 10% iBLER. The Figure 27 illustrates the average effective L2 post HARQ frame averaged throughput versus the BLER for the 1st HARQ transmission for MCS index 27. 100%

35000 kbps

90% 30000 kbps 80%

27.5 Mbps Throughput 70% 60%

t

R N I S B d 2 . 0 2

R E 50% L B i

40% 30%

25000 kbps 20000 kbps u p

h g u o 15000 kbps r h T

10000 kbps

20%

10 % iBLER

10%

5000 kbps

0% 12 dB

0 kbps 14 dB

16 dB

18 dB

20 dB

22 dB

24 dB

SINR

Figure 27: Throughput mapping for 20.2dB SINR, respecting 10% iBLER HARQ operating point for 50 RB, MCS index 27, Closed Loop Rank 1 and 10MHz Bandwidth For the recommended 10% iBLER HARQ operating point it can be seen that an SINRPDSCH = 20.2dB is required which corresponds to a throughput of 27.5Mbps. This is an example for MCS index 27, the same can be done for the full range of MCS indices resulting in the plot shown in Figure 28 in Section 3.1.5.4.

3.1.5.4 Selection of the Optimal MCS Index In order to select the optimal MCS index for the SINRPDSCH conditions at a specific UE location. First the same process to that identified in Figure 27 must be performed for the full range of MCS indices. Figure 28 illustrates for a range of SINRPDSCH values the corresponding optimal MCS indices and post HARQ average effective frame throughputs for a 10% iBLER HARQ operating point.


51 / 61

INTERNAL

< LTE DIMENSIONING GUIDELINES – OUTDOOR LINK BUDGET> MCS 30

35000 k bps

30000 k bps

MCS 25

25000 k bps MCS 20 t u

20000 k bps p

h g u o r 15000 k bps h T

S CMCS 15 M

MCS 10 10000 k bps MCS 5

MCS 4

5000 kbps

3.2 Mbps

˜

MCS 0 -7 dB

0 kbps -2 dB

3 dB

8 dB

13 dB

18 dB

SINR

Figure 28: Optimal MCS Index Selection for a -2dB cell edge SINR, 50 RB, 10MHz Bandwidth, Closed Loop Rank 1 Assuming a specific UE location the cell edge SINR can be computed (see section 3.1.1), SINRPDSCH, the next step is to select the optimal MCS index for such conditions. As an example here it is assumed that for the considered UE location the SINRPDSCH is computed to be -2dB. Referring to Figure 28 it can be seen that the optimal MCS Index = 4 and the corresponding post HARQ throughput is around 3.2Mbps for SINRPDSCH = -2dB.

Note: This relationship has been derived based on Closed Loop Rank 1 link level performances. On top of these performances there will be additional gains in very good channel conditions, due to spatial multiplexing / Rank 2 MIMO (this is discussed in more detail in section 3.1.5.6).

3.1.5.5 Downlink Explicit Diversity Gains The default SINR performances considered in the Alcatel-Lucent downlink budgets are for a 2x2 Rank 1 configuration, these performances account for SFBC pre-coding or Closed Loop Rank 1 gains and for a 2RxDiv gain at the UE. The choice to base the link budget on Rank 1 link level performances was made as the channel conditions typical of the cell edge are not generally conducive to effective utilization of Spatial Multiplexing. However, when in very good SINR conditions, a spatial multiplexing is applied in the downlink link budget. This gain is applied on top of the 2x2 Rank 1 (see section 3.1.5.6).

Note: An additional over the air power combining gain is also considered on the downlink, e.g. a 3dB gain is applied in the downlink budget for PathsDL≥2 to account for the fact that the same RE’s are transmitted on each transmit path (with the exception of the RS). Where the number of transmit paths, PathsDL, considered does not match the 2x2 configuration assumed in the underlying link level simulation data, an additional gain or loss is applied to the computed SINR depending on the number of transmit paths, as detailed in Table 23: Alcatel-Lucent – Proprietary Version 2.9

52 / 61

INTERNAL


Table 23: Approximation for Impact of <> 2 Transmit Paths Downlink Transmit Paths, Path DL

SINR Impact

1 path

-1.0 dB

2 paths

0.0 dB

4 paths

1.0 dB

Warning : the SINR impact detailed in Table 23 is a very rough approximation to the expected performances with 1 and 4 transmit paths. Ensuring higher confidence in the link budget results would require consideration of dedicated link level results for such configurations.

3.1.5.6 Spatial Multiplexing / MIMO Gain As outlined in section 3.1.5.5, the underlying link level performances used to select the optimal MCS Index and the corresponding throughput for a given number of resource blocks are for a Rank 1 configuration. In very good channel conditions (channel rank >1 and high SINR) an additional spatial multiplexing gain on top of the underlying link level simulation performances is applied. Such a gain is based upon a comparison of link level performances between a Rank 1 configuration and a Rank2 spatial multiplexing configuration. Figure 29 illustrates a summary of the gain computed from such a comparison for three different channel conditions:

2.00 VehA 3km/h - Med

1.90

n i a G1.80 g n1.70 i x e l p1.60 i t l u M1.50 l a i t1.40 a p S 1.30 2 k n1.20 a R

VehA 50km/h - Med VehA 120km/h - Med

1.10 1.00 10.0 dB

15.0 dB

20.0 dB

25.0 dB

30.0 dB

35.0 dB

40.0 dB

45.0 dB

SINR

Figure 29: Gains Associated with Spatial Multiplexing (MIMO Rank 2) Note: The antenna correlation that best represents what has been seen in the field to date is best represented by the medium correlation assumptions. From Figure 29 it can be seen that spatial multiplexing gains become significant beyond an SINRPDSCH = 16dB and progressively increase with increasing SINR PDSCH. The precise gain attributable to spatial multiplexing is dependent upon the SINR, the antenna correlation and the channel model. Medium antenna correlation is assumed to address average performances.


53 / 61

INTERNAL


As an example, consider an SINRPDSCH = 35dB and VehA3 channel conditions. Referring to Figure 29 the estimated spatial multiplexing gain is around 1.86. If we assume the Rank 1 link level performances indicated a 30.4Mbps throughput for 50RB and a 10MHz bandwidth then the final throughput after accounting for the spatial multiplexing gain would be 30.4Mbps x 1.86 = 56.5Mbps.

Note: One thing that is not possible to compute from a link budget analysis is whether the channel can support rank 2 transmissions. The best that can be done is to assume/approximate that with a high SINR there is a reasonable probability that the channel rank will also be sufficiently good. For example, high SINR is most commonly observed close to the serving eNode-B and so is higher channel rank.

3.1.6 Resource Element Distribution Computation of the downlink SINR and RSRQ detailed in sections 3.1.1 and 3.1.2 is dependent upon the average resource element allocation to the various downlink channels. An example of the RE distribution for 2 transmit paths, Control Format Indicator (CFI) = 3, and 10MHz bandwidth is summarized in Table 24 for the first transmit path.

Table 24: Example Average RE Distribution Across the 14 OFDM Symbols of a Single TTI (2 Transmit Paths, CFI=3, 10MHz Bandwidth) Type of RE

0 t o l S

1 t o l S

DLRS1

P-SCH

S-SCH

PBCH4

PDCCH

PCFICH

PHICH

PDSCHA

PDSCHB

Unused6

sym. 0

100 RE

0 RE

0 RE

0 RE

300 RE5

16 RE

84 RE

0 RE

0 RE

100 RE

sym. 1

0 RE

0 RE

0 RE

0 RE

600 RE

0 RE

0 RE

0 RE

0 RE

0 RE

sym. 2

0 RE

0 RE

0 RE

0 RE

600 RE

0 RE

0 RE

0 RE

0 RE

0 RE

sym. 3

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

600 RE

0 RE

sym. 4

100 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

400 RE

0 RE

100 RE

sym. 5

0 RE

0 RE

12.4 RE3

0 RE

0 RE

0 RE

0 RE

0 RE

585.6 RE

2 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

585.6 RE

2 RE

2

sym. 6

0 RE

12.4 RE

sym. 0

100 RE

0 RE

0 RE

1.2 RE

0 RE

0 RE

0 RE

398.5 RE

0 RE

100.3 RE

sym. 1

0 RE

0 RE

0 RE

1.2 RE

0 RE

0 RE

0 RE

0 RE

598.2 RE

0.6 RE

sym. 2

0 RE

0 RE

0 RE

1.8 RE

0 RE

0 RE

0 RE

0 RE

598.2 RE

0 RE

sym. 3

0 RE

0 RE

0 RE

1.8 RE

0 RE

0 RE

0 RE

0 RE

598.2 RE

0 RE

sym. 4

100 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

400 RE

0 RE

100 RE

sym. 5

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

600 RE

0 RE

sym. 6

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

0 RE

600 RE

0 RE

Notes: 1

Two RE allocated per Resource Block (RB) for OFDM symbols 0, 4, 7 and 11

2

P-SCH is always located in the last OFDM symbol of the 1st and 11th slots of each radio frame for the center 6 RB's (figures averaged across 1 radio frame) 3

S-SCH is always located on the symbol before the P-SCH in the 1st and 11th slots of each radio frame for the center 6 RB's (figures averaged across 1 radio frame)


54 / 61

INTERNAL


4

The centre 6 RB's (72 subcarriers) the non-RS RE are used for the PBCH for the 1st 4 symbols of the 2nd slot with a 40msec periodicity (figures are averaged across 4 radio frames = 40msec). Note for symbols 0 and 1 only 8 RE are considered per RB as the remainder are reserved for the DLRS, for symbols 2 and 3 all 12 RE are considered per RB. 5

PDCCH RE after accounting for DLRS, PCFICH and PHICH REs

6

There remains some unused RE due primarily to RE reserved for RS transmission on the 2nd transmit path and also some RE reserved around the SCH RE A summary of the average number of Resource Elements (REs) that are transmitted per TTI for 1, 2 and 4 transmit paths is presented in Table 25. This is based on equivalent analyses to that presented in Table 24. The averaging is performed over 4 radio frame (40msec).

Table 25: Average Number of RE Transmitted per TTI per Transmit Path 1 Tx Path/s

2 Tx Path/s

4 Tx Path/s

Nsym-PDSCH

6265 RE

5965 RE

5765 RE

Nsym-DLRS

400 RE

400 RE

400 RE

Nsym-PDCCH

1600 RE

1500 RE

1300 RE

Nsym-SCH

25 RE

25 RE

25 RE

Nsym-BCH

6 RE

6 RE

6 RE

100 RE

100 RE

100 RE

Nsym-PCFICH_PHICH

3.1.7 Energy Per Resource Element (EPRE) The Energy Per Resource Element (EPRE) is the transmitted energy associated with a single resource element. This parameter is dictated by the overall output power setting for the eNode-B, the carrier bandwidth and the product variant. For each product variant the following set of information is defined (as summarized in Table 26):   

PowerRef – the reference downlink eNode-B transmit power per transmit path BWRef – Reference bandwidth EPREDLRS(Ref) – the EPREDLRS for the corresponding reference power and bandwidth

Table 26: Product EPRE Reference Hardware

Power Ref

BWRef

EPREDLRS(Ref)

RRH 2x40

30 W

10 MHz

17.0 dBm

TRDU 2x40

40 W

10 MHz

18.0 dBm

RRH 2x30

30 W

10 MHz

17.0 dBm

RRH 1x60

60 W

10 MHz

20.0 dBm

RRH 2x60

60 W

10 MHz

20.0 dBm

MC-RRH

40 W

10 MHz

18.0 dBm

MC-TRX

60 W

10 MHz

20.0 dBm

As it is possible to use the same power amplifier with a different power setting, PowerCurrent and different bandwidth, BWCurrent, in such cases the EPREDLRS is given by:


55 / 61

INTERNAL


EPREDLRS

=

EPREDLRS(Re f )

 PowerRe f   BWRe f    + 10 ⋅ Log10   Power   BW Current    Current 

− 10 ⋅ Log 10 

Note: The recommended overall power settings, PowerCurrent, for different carrier bandwidths are detailed in section 4). For example, consider for PowerCurrent = 40W and BWCurrent = 20MHz. The ERPERS for RRH40 hardware is given by: EPRERS

=

 30 W   10 MHz   + 10 ⋅ Log10    40 W   20 MHz 

17 dBm − 10 ⋅ Log10 

=

15.2dBm

Table 27 summarizes the power offsets from the EPREDLRS power setting for channels other than the RS:

Table 27: Power Offsets from EPRE DLRS RE Type

Power Offset from EPRE DLRS

OffsetPDSCH

0.0 dB

OffsetPDCCH

3.0 dB

OffsetSCH

0.7 dB

OffsetBCH

1.6 dB

OffsetPCFICH & PHICH

1.0 dB

Internal: The power offset detailed in Table 27 are dependent upon the recommendations for the specific software release, the RF hardware variant and the PA output power. Also, the Offset for PDCCH has been set to account for impacts of PDCCH power control. However, to simplify the link budget these values have been defined as fixed values.

3.1.8 Shadowing Margin & Handoff Gain For the downlink the same assumptions are considered to hold true, for reasons of reciprocity, when computing the received signal level at the UE, as they do on the uplink when computing the received signal level at the eNode-B. Thus the same relationships and rationale to those presented in section 2.1.8 (shadowing margin) and section 2.1.9 (handoff gain) are assumed to be equally applicable on the downlink. The only exception arises when considering the SINR and RSRQ on the downlink. For such computations the above mentioned shadowing margin and handoff gains are applied equally to both the desired and interfering signals and thus the net effect is only to bring the signal closer to the noise floor. In reality the desired and interfering signals are not perfectly correlated with each another. To account for such non-ideal correlation an approximation is applied in the downlink link budget to account for an additional shadowing margin on the SINR and the RSRQ. The shadowing standard deviation considered for the SINR shadowing margin is determined by the standard deviation considered for the given environment. Table 28 summarizes the mapping from the environment shadowing standard deviation to that considered for the computing the SINR shadowing margin.


56 / 61

INTERNAL


Table 28: Mapping from Environment Shadowing Standard Deviation to the Shadowing Standard Deviation Used for Computing SINR Shadowing Margin

Environment Shadowing Std. Dev.

DL SINR Shadowing Std. Dev.

6 dB

0 dB

7 dB

1 dB

8 dB

2 dB

The same method as detailed in section 2.1.8 is used to compute the SINR shadowing margin based on the SINR shadowing standard deviations presented in Table 28.

Note: Close agreement has been observed when comparing field measured SINR and RSRQ distributions with predicted SINR and RSRQ distributions that account for a shadowing margin based on the standard deviations presented in Table 28.

3.2 Downlink Budget Example Table 29 presents some example of the entire downlink budget analysis for a dense urban environment with deep indoor penetration for a range of different services. Note that this is the downlink link budget that corresponds to the uplink budget presented in Table 20.

The key objective of the downlink link budget analysis is to formulate estimates of the data rate expectations for the cell ranges of some nominal uplink data rates (see the rows in red in Table 29).


57 / 61

INTERNAL


Table 29: Typical PDSCH link budgets for a RRH deployment in Dense Urban VehA3 channel conditions at 700MHz (uplink 128kbps guaranteed at cell edge)

It is important to note that the downlink data rate estimates presented in the last row of Table 29 are achievable with 95% coverage reliability over the downlink cell ranges indicated in the row titled “UL Service Cell Range”. Note also that the same data rates are achieved over the entire coverage area (1.24km cell range) with reduced reliabilities for the higher data rates.

Internal: The default ALU link budget can be found on the intranet: Alcatel-Lucent LTEFDD & TDD Link Budget .


58 / 61

INTERNAL


4

DOWNLINK OUTPUT POWER

Assuming that the carrier bandwidth is known it is important to size the downlink power amplifier to ensure sufficient DL power resources to match the targeted uplink coverage. A series of system simulation studies were performed to assess the required Power Amplifier (PA) sizing for 4 different cases    

700 MHz (10 MHz) 2.1 GHz (10 MHz) 2.1 GHz/AWS (5 MHz) 2.6 GHz (20 MHz)

All scenarios considered 2x2 MIMO on the DL and 2RxDiv on the UL. In principle, all of the studies concluded that spectrum efficiency for “reasonable” cell sizes is relatively invariant to reasonable choices of PA sizes and that edge rates become much more sensitive to the choice of power at large cell radiuses. The process for assessing the downlink power requirements is summarized below:   





Step 1: Confirm uplink edge data rate MAPL, i.e. 64 kb/s or higher through the use of link budget Step 2: Determine the cell range based on the link budget for 64kb/s or other rate Step 3a: Run downlink system simulations using same cell size/range & other link budget requirements (use CDMA power levels for guidance) and variable PA Size to confirm that edge data rate requirements for the downlink are met Step 3b: Alternatively construct a downlink link budget (variations in loading & interference, benefit of multi-user scheduling, frequency selective scheduling, HARQ, can not be well modeled in downlink link budgets) and confirm that edge data rate requirement can be met Step 4: Perform uplink system simulations and observe cell edge rate meets link budget requirements

Table 30 summarizes the recommended PA sizing based on the observations from the above mentioned study (independent of the frequency band).

Table 30: Recommended Power Amplifier Sizing


Carrier Bandwidths

PA Power

1.4 MHz

2 x 20 W

3.0 MHz

2 x 20 W

5.0 MHz

2 x 20 W

10.0 MHz

2 x 30 W

15.0 MHz

2x40 W

20.0 MHz

2 x 40 W

59 / 61

INTERNAL


5

RADIO NETWORK PLANNING

The key motivations for utilizing an RNP tool for LTE design purposes include: 



Enhancement of the accuracy of the initial link budget design by accounting for field constraints such as the topology, morphology and traffic distribution. Site positions, antenna heights, antenna tilts can also be optimized. Accounting more accurately for the interference (geometry) scenario encountered for the specific deployment

It is the later point that is particularly important for LTE downlink coverage considerations. The approach recommended by Alcatel-Lucent is to formulate an UL link budget to define the cell range and then within that cell range perform an RNP study on the DL to assess the SINR performances achievable at the cell edge. This allows the following points to be assessed:  

Whether the DL is interference or noise limited If noise limited, a higher output power configuration should be considered o Whether the corresponding DL cell edge performances satisfy the DL cell edge performance expectations o This will drive the required bandwidth, output power and MIMO configuration on the DL

Alcatel-Lucent relies on the 9955 RNP tool, based on the ATOLL platform developed by FORSK. 9955 is fully aligned with Alcatel-Lucent’s products and engineering tool chain. Alcatel-Lucent customers can fully benefit from this tool since it is included in AlcatelLucent’s product portfolio.


60 / 61

INTERNAL

03 - LTE Dimensioning Guidelines - Outdoor Link Budget - FDD - Ed2.9 - Internal (1)

Recommend Documents