Ppt0000000.ppt [Lecture seule] - ITU: Committed to ... · Transversal Supporting Activities WP4: Impact of innovations on the RAN ... interference management (with 3D beam-forming)

Advanced Radio Interface TechnologIes for 4G SysTems

Overview

Date: Monday, July 16th 2012 Raphael Visoz - OrangeITU-R WP5D Workshop “Research Views on IMT Technology Evolution”

Advanced Radio Interface TechnologIes for 4G SysTems 1


Outline

ARTIST4G drivers Objectives and concepts Management structure Some Results Conclusion

2


ARTIST4G Drivers

Customers• Have new devices and greedy usages (iPhone, video, web 2.0)• Won’t pay for the traffic increase they generate• Expect an even more uniform Quality of Service

The market• Has limited investments capabilities (operators and vendors)• Must capitalize on existing infrastructures: LTE (backward

compatibility)

3


Objectives and Concepts

Increase the average performance for the benefit of cell-edge users

• In limiting the interferences, especially at the cell-edge

• Or via densification at low cost

4


Chalmers University of Technology

Consortium

5


Main Research Directions WP1: Interference avoidance

• avoiding interference e.g. through coordination among different non co-located transmission points and designing the generated radio signals accordingly

WP2: Interference exploitation• resources allocation algorithms allowing for efficient

interference cancellation, with a two-step approach– Investigation of interference cancelling receiver algorithms

– Design of higher-layer allocation schemes and control signaling

WP3: Advanced relays• Providing ubiquitous user experience: advanced relays need

to provide capacity on top of coverage (current relays in LTE-A are designed for coverage extension)

Limiting Interference

Densification

6


Transversal Supporting Activities WP4: Impact of innovations on the RAN

architecture• Proposed RAN innovations may enforce additional requirements

on backhaul networks: innovations need to be classified– No cooperation through the RAN– Cooperation with control-plane exchange– Cooperation with user-plane exchange

WP5: Evaluating innovations• Agreed procedures have to be settled to allow fair comparison of

innovations– Scenarios, metrics and methodologies– A main KPI is needed to measure uniformity of performances

WP6: Live testing of innovations• Several test beds available to experiment the most promising

innovations

7


Field Trials in Dresden- 16 Base Stations interconnected via microwaves links- Real time transmissions over the air

8


Advanced receiver proof of concept

Performance and Complexity trade-off are first evaluated before implementation into real production chain

Complexity and energy consumption are evaluated on CEA’s Network on Chip platform

Many field trial campaigns have been performed to evaluate advanced receivers performance gain with recorded radio channel samples.

9


Some Results 3D beamforming evaluations A Working CoMP framework Advanced relaying techniques

• Moving relays, multi-hop relays, Improved backhaul performances

Interference cancellation receivers• Innovative receivers algorithms • Performance prediction methodologies for iterative receivers• Flexible interference concepts: adapt resource allocation to the

receivers interference cancellation capabilities

10


3D-Beamforming

This concept is not part of LTE R10

11

Conventional fixed down-tilt causes interference

Objective: exploit also elevation dimension dynamically

Ground breaking innovation

LTE-A ARTIST4G

Simultaneous gains for both spectral efficiency (about 15 – 20%) andcell edge throughput (up to 50%)

Coordination gain and vertical BF gain (3D BF) are almost additive Exact vertical downtilt adaptation is approximated reasonably with 2 or 3 fixed

downtilts 3D beamforming already became part of ALUD product strategy


Coordinated Multi-Point (CoMP)

Study item stopped in R10

Not included in LTE R10• Technology not mature

enough

Study Item R11• No convincing results

12

Enlarged cluster of coordinated cells (e.g., 9 cells)

Better inter-cluster interference management (with 3D beam-forming)

Overlapping clusters (cover shift) in different frequency subbands

User centric scheduling (choose best cover shift per user)

Improved feedback in terms of efficiency and robustness by use of channel predictors

LTE-A ARTIST4G

Spectral efficiency up to 6..7 bit/s/Hz/cell about 100% CoMP gains


Interference Cancellation Full orthogonality provides interference avoidance

• But known to be suboptimal in terms of capacity

Cancelling interference at Rx increases the capacity through resource re-use

• But interference is not always easy to estimate

Controlling and structuring the interference conditional on advanced receiver processing opens ways to optimize the spectral efficiency

WP2 focused on non linear receivers using iterative processing (turbo principle)

• various functions in the receiver collaborate (e.g. channel estimator, detector and decoder) to better estimate the transmitted information

13

New physical layer abstraction for turbo-sic receivers enabling CLO (proven with measurements)9dB gains on colliding reference signals in Almost Blank Sub-frames used by pico cells.


Advanced Relays

The relay standardized in LTE-A has the full eNB functionalities

• can be seen as an eNodeB with a wireless backhaul operating in LTE spectrum

Expected benefits• Reduced cost: no backhaul, low-cost

device• Easy and quick to deploy (no

backhaul, small, low-power)

Foreseen main scenario: coverage enhancement

14

Advanced relays need to provide capacity in addition to coverage

CA on the relay backhaul

Adaptive HARQ

Distributed coding

Multi-hop relays

Moving relays

LTE-A features ARTIST4G innovations

CA on backhaul link gains:+20% in 5%-ile throughputs+11% in throughput Jain’s index


Conclusion

15

A mid term oriented research project• 20% of short term innovations• 60% of mid-term innovations• 20% of long term innovations

Some innovations already discussed in 3GPP Project finished since June 2012 More than 100 publications, ~20 deliverables All results available (for free) online at:

http://ict-artist4g.eu


Thank you !

Questions?

16

WP1 backup



Introduction ARTIST4G’s goal: improvement of user experience ubiquity LTE-A Rel. 10 is the baseline system ARTIST4G WP1: design of interference avoidance schemes for improving

user experience ubiquity

We have defined different requirements on the architecture• CP_COOP: only control is exchanged between nodes for the sake of cooperation• UP_COOP: control+data are exchanged between nodes• HETNET: New architecture enabler are need for the sake of cooperation

We are targeting• Short term deployments: No modification of LTE Rel 10• Medium term deployments: New mechanisms and architecture solutions based

on LTE Rel 10• Long term deployments: New deployments optimized for CoMP


Introduction 3D beamforming and

coordinated schedullingfor CP_COOP systems

• Short and Medium term solutions

• Improvement of the cell edge throughput

Joint transmission with interference shaping (UP_COOP)

• Long term innovation• Improvement of the

spectral efficiency and user experience ubiquity

Massive deployment of small cells (HETNET)

• Improvement of the network capacity


eNB

Low inter-cell interferenceon each of the resources represented bydifferent colors:

eNB

Low inter-cell interferenceon each of the resources represented bydifferent colors:

Advanced 3D Beamforming (1/3)


• Today:beamforming and MIMO schemes with fixed vertical antenna pattern

• Impact on signal strength according to UE location

• Constant inter-cell interference independent of served UE location

• Advanced 3D Beamforming:• additional degree of freedom for Tx signal

optimization and interference reduction

• Exploitation of dynamic vertical beamsteering per radio resource

• Serve each UE with individual Tx antenna pattern

• Verification of 3D beamforming• Field measurements in real deployment

scenarios

D1.2

Advanced Radio Interface TechnologIes for 4G SysTemsAdvanced Radio Interface TechnologIes for 4G SysTems

Methods and implementation options Implicit coordination

fixed resource allocation according tocell areas

Distributed horizontal and vertical beam coordination Resource allocation to UEs

depending on dynamic feedbackand exchange of adjacent cellworst case interference PMIsamong base stations

Multiple fixed downtilts vs. exact steering


serving cell

its cooperation set

site

information exchangeinformation exchange

serving cell

its cooperation set

site

information exchangeinformation exchange

= 9° = 13° = 17°3 DT areas:

Multiple fixed downtilts


Impact & applicability Advanced 3D beamforming shows clear benefits over the conventional

one fixed downtilt solution, even without coordination.

Advanced antenna systems suitable for implementation of advanced3D beamforming are available

3D beamforming already became part of ALUD product strategy Verification with field measurements

Field trials verified the basic characteristic of the vertical channel as assumed for the simulations


Key outcomes: Simultaneous gains for both spectral efficiency (about 15 – 20%) and

cell edge throughput (up to 50%) Coordination gain and vertical BF gain (3D BF) are almost additive Exact vertical downtilt adaptation is approximated reasonably

with 2 or 3 fixed downtilts


User centric cooperation in clustered networks:

Enlarged cooperation areas (e.g. 9 cells) increase probability to contain e.g. 3 strongest cells within cooperation area

Partial reporting limits feedback overhead

Setup of overlapping cooperation areas in different frequency subbands cover shifts

Cooperation area edge UEs are scheduled into center of best fitting cover shift

User centric clustering for partial CoMP (1/3)(Tortoise)

Advanced Radio Interface TechnologIes for 4G SysTems D1.4

789 1

23456

789 1

23456

123

456 7

897

89 1234

56

123

456 7

89

With six cover shifts all UEs having their 3 strongest cells within 3 adjacent sites are being served user centric

cover shift 1

cover shift 2


Interference floor shaping – Tortoise concept Cell edge UEs suffer from low SNR as well as

strong interference floor generated by other cells

JP CoMP gains hidden in interference floor Cover shift concept provides novel options for IF

floor shaping Cooperation area (CA) cell individual

antenna tilting and power allocation

CA center/outbound wideband beams with low/strong tilt & strong/low Tx power

CA edge UEs scheduled into other cover shift


Tortoise like shape ofRx power over location

7° tilt / 46dBm

15° tilt 40dBm


Impact & applicability User centric clustering essential for generating real world CoMP gains

Partial CoMP combines a large penetration rate of user centric served users with limited feedback overhead

‘Tortoise’ interference floor shaping boosts performance and localizes thenetwork wide to a cooperation area limited optimization problem

Interaction with WP6:

WP6 field trials verified the basic principle of the Tortoise interference floor shaping technique


Key outcomes: Penetration rate for CoMP users of about 90% Interference floor suppression of about -20dB for about 70% of UEs In combination with other means spectral efficiency up to

6..7 bit/s/Hz/cell about 100% CoMP gains


Statement: Most of the traffic is in indoor (home and offices)HeNB deployments improve the network capacity

HeNBs allows for offloading the in-home traffic HeNBs must bring high wireless performance Closed Subscriber Groups restricting the access to the HeNBs generate

interference issues in downlink The high number of HeNBs generates non-negligible levels of

interference on the macro network in uplink eNB/HeNB cooperation is limited by the architecture

Objectives: Interference avoidance mechanisms and architecture enablers for protecting the macro network while taking benefit from the huge offloading gain from a massive small cell deployment

Interference avoidance scheme for co-channeleNB/HeNB deployments

Advanced Radio Interface TechnologIes for 4G SysTems D1.4


Principle: Define a power setting strategy guarantees a target degradation in a High Interference Zone (HIRZ) around each HeNBNo eNB/HeNB cooperation is required

Based on measurements available at the HeNB onlyTarget short term deployments, no impact on the architecture

Key outcome: For an average 2b/s/Hz cell edge performance at the HeNBsNo degradation of the eNB cell edge performance for less than 125 HeNB/km²For 500 HeNBs/km² => Huge network capacity gain by offloading

35% degradation of the eNB cell-edge performance with fixed power

10% degradation only with the proposed power control strategy

Macro network protection in Downlink


• Virtualization of the “Small Cell Network”(SCN) as a single interferer in uplink and a single radio neighbor

• Architecture enablers: Group PCI and Coordination Gateway

• Shaping of the interference distribution in a semi centralized fashion (one central unit per eNBcell)

• Measurements of the eNB from the small cells• Measurements of the SCN from the eNB• Optimization and broadcast of parameters to small

cells• Selection of parameters and computation of the

Ues power control oin each small cell• Excellent trade-off between eNB protection, small

cell performance, and small cell-UEs power consumption

Macro network protection in Uplink


Conclusions We have shown three research axes of the ARTIST4G

project, based on interference avoidance The innovations that have been developed can be

associated with interference cancelation or used in the relays context

New architecture enablers have been developed in ARTIST4G WP4 for supporting these innovations• New cooperation interface for Joint Processing CoMP• Design of the cooperation protocols for Joint

Processing CoMP• Enablers for eNB/HeNB cooperation

WP2 backup



Paradigm “interference exploitation”: Don’t avoid

interference; instead make use of it through flexible interference control and advanced receivers

Advanced Receivers

Link to System Abstraction

System Level Concepts

Main Focus of Presentation


Advanced Receivers




Iterative MMSE-Soft IC Iterative receiver with equalization and channel decoding step Basis for advanced receiver enhancements (like complexity reduction and

semi-blind channel estimation) in ARTIST4G System level analysis and link adaptation require easy and fast

performance prediction.


Advanced Receivers




Advanced L2S Modelling Keep track of the coded bit average mutual information (AMI)

circulating between the MUD and the bank of outer APP decoders Demapping and decoding understood as a joint process.

Measurements

• Transport Format• Transmit power• UE Positions

Measurement

UE1

UE2

IQ-B

aseb

and

Sam

ples

TUD ML Chain

SIC Receiver

FT ML Chain

Prediction

MeasuredBLER

PredictedBLER

Parameter Value Bandwidth 20 MHz FFT size 2048 Subcarriers for tx 1200 Sampling rate 30.72 MHz Subcarrier spacing 15 kHz Transmit Time Interval 1ms OFDM symbols/TTI 14 Subcarriers per PRB 12 Total available PRBs 100 Modulation Schemes QPSK, 16QAM, 64QAM

Effective SNR

BLE

R

Regime of InterestreferencePredicted from measurements

Measurement Results

Performance predictionmethodology shows reasonable accuracy for real world measurements

Abstraction algorithmcould be used for fastlink adaptation and in resource allocation algorithms

Recommended Deliverables: D2.3, D6.4


Advanced Receivers



Flexible Interference Control

UL CoMPScheduling


Flexible Interference Control (FIC) Combine interference exploitation (receiver side) with

interference avoidance (transmitter side) in flexible ways.

A) avoid interference, where possible

B) Cope with remaining interference


Flexible Interference Control Examples

Joint Transmission Interference Cancellation Receiver (JTICR)

• based on combination of Joint Transmission and successive interference cancellation receivers for a more efficient operation of the downlink

• two modes applied in different operational conditions, mainly associated with the operation in HetNet environments

Adaptive Resource Allocation Algorithms for Multiuser MIMO Systems with Iterative MMSE-IC Joint Decoding

• resource allocation algorithms taking advantage of Iterative MMSE-IC receivers at the base station to maximize capacity in Multiuser MIMO systems


Flexible Interference ControlIn homogeneous networks, traffic is served by macro base stationsDeploying pico cells offers capacity gains by traffic offloadingAdvantages

– Better load balancing between macro and pico layer improves network capacity and user

How to achieve traffic offloading? Answer: Cell range expansion of the pico cell

– The UE should connect to the pico cell even if the macro cell is stronger

– Coverage area of the pico cell is artificially enlarged Cell range expansion

MacroPico Pico

Pico

SIR = 0 dB at cell edge of regular coverage area

SIR < 0 dB at cell edge of expanded coverage area


Flexible Interference Control UEs in close proximity to CSG femto cells that it is not allowed to connect to

• strong interference from the CSG cell

Macro-pico deployments with UEs operating in cell range expansion• Nominally, a UE associates with a base station with strong DL SINR • With cell range expansion, a UE can associate with a low power eNB

Femto is the aggressor and macro the victim

Macro is the aggressor and pico the victim

MacroFemto Femto

MacroPico Pico


eICIC (Rel-10)

1ms

12 sub

carrier o

f 15 kH

z each

RS antenna port 1 resource element

RS antenna port 2 resource element

Empty PDSCH resource element

Empty PDCCH/PHICH/PCFICH resource element

In Rel-10, almost blank subframes (ABS) have been introduced

In a ABS, no unicast PDSCH and PDCCH is transmitted

To ensure backward compatibility, following signals are transmitted

• CRS (pilot signal)• PSS/SSS (synchronization signals)• SIB1/MIB (broadcast information)

CRS/PSS/SSS/SIB1/MIB can still cause strong interference in certain PRBs/REs


eICIC (Rel-10)

Interference coordination between aggressor cell and victim cell is done by means of a bitmap sent over X2 interface

• Each bit is mapped to a single subframe and indicates an ABS subframe• Based on the data traffic demand, the pattern can change each 40ms• Cell creating strong interference controls which resources can be used by the

victim cell to serve terminals in harsh interference conditions

MacroPico

Interference bitmap transmitted over X2 backhaul

Statically assigned almost blank subframe

Semi‐statically assigned almost blank subframe

Semi‐statically assigned regular subframe

X2 backhaul link


Why IC Receivers are needed

Even in case of almost blank subframes at the aggressor nodes, the CRS/PSS/SSS/SIB1/MIB are still transmitted

This interference from aggressor node causes significant performance degradation to data and control channels of serving cell

• Therefore cancelling interference of CRS, PSS/SSS and PBCH is needed to enlarge cell range expansion

Interference cancelation algorithms for those signals/channels can be designed that achieve SIR = -18 dB


Throughput Gains by CRS IC –Colliding RS (Rel-11)

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 320

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

Serving cell C/I (dB)

Thro

ughp

ut (K

bps)

CollidingRS-TxMode3

CellID0-TxMode3CellID0-TxMode3-CellID96-TxMode3-16dBCellID0-TxMode3-CellID96-TxMode3-IC-16dB

CRS IC Gains for colliding RSin non-MBSFN ABS

9 dB Gain

Recommended Deliverables: D2.4, D2.5


Advanced Receivers



Flexible Interference Control

UL CoMP Scheduling


Interference Exploitation – UL CoMP

Distributed Schedulers

Joint Detection

Array G

ain

Multiplex

ing Gain

Multi-User Diversity

Joint Scheduler

Backhaul


Scheduling and Power Control for UL CoMP

Joint Scheduling:•Scheduler centrally assigns resources in all cells•Full CSI needed at scheduler

Cooperation Cluster

Backhaul

Joint Detection:•BSs exchange of rx-signals•Performance increase through interference exploitation

Joint SchedulerJoint Detection

• Single-antenna BSs• Single-antenna MTs• 2 user spatial multiplex (grouping)

dd-d00 d0

K Terminals (uniform distri.)

JD

Scenario


Power Control

All Terminals transmit with the same power.

Fixed Power Single-Cell Multi-Cell

Pathloss to the nearest BS is compensated to achieve an average target SNR.

Pathloss to both BSs is compensated to achieve an average target SNR (involving receive power and noise at both BSs).

0 100 200 300 400 500-30

-25

-20

-15

-10

-5

0

5

Position (m)

Pow

er (d

Bm

)

Power vs. Position (SNRsingle=30dB)

Multi-cellSingle-cellFixed

0 100 200 300 400 5000

0.5

1

1.5

2

Position (m)P

ower

(mW

)

Power vs. Position (SNRsingle=30dB)

Multi-cellSingle-cellFixed


Scheduling Algorithms

50 100 150 200 250 300 350 400 4500

1

2

3

4

5

6

Position (m) - (bin size 20m)

Ave

rage

Thr

ough

put (

bpcu

)

MultiSingleFixed

Proportional Fair

• Multi-Cell Power Control achieves good fairness (and higher throughput than for other algorithms)

• Fixed power control not as fair as Single/Multi

50 100 150 200 250 300 350 400 4500

1

2

3

4

5

6


Ave

rage

Thr

ough

put (

bpcu

)

MultiSingleFixed

Max Rate


Scheduling Algorithms

50 100 150 200 250 300 350 400 4500

1

2

3

4

5

6


Ave

rage

Thr

ough

put (

bpcu

)

MultiSingleFixed

50 100 150 200 250 300 350 400 4500

1

2

3

4

5

6


Ave

rage

Thr

ough

put (

bpcu

)

MultiSingleFixed

Max Fair Proportional Fair

• Similar fairness for all power controls • Overall higher throughputs• Fixed power control not as fair as

Single/Multi


Throughput vs. Fairness

0.5 1 1.50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Mean Throughput (bpcu)

Jain

´s In

dex

multi - max ratemulti - prop fairmulti - max fairsingle - max ratesingle - prop fairsingle - max fairfixed - max ratefixed - prop fairfixed - max fair

fair, spectrally inefficient

unfair, spectrally inefficient

fair, spectrally efficient

unfair, spectrally efficient

Recommended Deliverables: D2.5


Conclusions Advanced receivers with interference cancellation capability allow to

overcome some of the limitations of current system without requiring changes to the standards.

Flexible interference control (tuning between interference avoidance and coping with interference) opens up a new design space that allows systems to adjust to the radio propagation conditions in a more fine grained way.

IC important component for traffic offloading in LTE HetNets (62% gains for 1MB file download at 30

New abstraction methodologies were provided that allow to make use of these added degrees of freedom for system analysis and design (proven with measurements)

UL CoMP is a promising means to exploit interference across cell borders. Scheduling and Power Control have large impact on system efficiency and fairness.

WP3 Backup



Introduction ARTIST4G work on relay concepts beyond the scope of the actual

standardisation work LTE-Advanced Relays are for coverage WP3 is looking for capacity improvements and ways for introducing

advanced relays into the LTE technology roadmap ARTIST4G aims for ubiquitous user experience

• Advanced Relays needed to provide capacity on top of coverage• Improve average throughput• Improve ubiquity (more uniform QoS) (cell-edge)

WP3 work is based on features of LTE-A and related research to optimise

• CoMP • Carrier aggregation, with in- and outband operation • Channel Coding• Scheduling• Interference Management• Various types of relays


Type-1 relay nodes Non-transparent RNs

• UEs see them as eNBs

They have their own cell ID, synchronization and control channels

RNs attach to a Donor eNB (DeNB)• RN acts like a UE while attaching to DeNB

Have been standardized by 3GPP• Several issues still need to be addressed


Challenges for type-1 relay nodes

Resource partitioning between backhaul and access network (two-hop)

Backhaul optimization to avoid bottlenecks• Multi-carrier relaying with carrier aggregation

Multi-hop deployments Inter-cell interference coordination (ICIC)


Resource Partitioning (RP) between backhaul and macro-access links

In-band Relays (RN tx-rx segregation in time):• In MBSFN frames: DeNB transmits to macro UEs and to the backhaul link• In non-MBSFN frames: DeNB transmits only to macro UEs and RN transmits to

relay UEs.

Out-band Relays (RN tx-rx segregation in frequency):• On primary carrier: DeNB transmits to macro UEs and to the backhaul link• On secondary carrier: DeNB transmits only to macro UEs and RN transmits to

relay UEs.

In both cases, we need RP between backhaul and macro access


Resource Partitioning (RP) between backhaul and macro-access links

Proposals from Nomor:• Semi-static: Fair allocation of resources based on the number of users. • Dynamic: Fair allocation of resources to “uniformalize” the throughput per UE

Pros: Simple resource splitting, simple schedulers, works well in absence of QoS req Cons: Not an optimal solution especially for heterogeneous QoS constraints In real life scenarios, different applications with different QoS requirements co-exist


QoS-aware scheduling proposalUnderlying principles

Jointly tackle the problem of resource partitioning and scheduling in a QoS-aware manner

In this presentation: a unified treatment of in-band and out-band relays Slots where macro-access and backhaul share resources:

• Use a QoS-aware scheduler by aggregating backhaul users as super users (one for each RN and each QoS type)

• The latency deadline for R-UEs needs to be split

Slots where only macro access is served at DeNB and relay-access at RNs• Scheduler at DeNB “as usual”• Scheduler at RNs need knowledge of already consumed delay budget• The latency dead line needs to be adjusted


Simulation Results Relay enhanced scenario with heterogeneous traffic mix

• 3GPP case-1 (ISD 500m) with 2 RNs in macro-cell• 25UEs in total, some assigned to macro cell (called M-UEs) and

others assigned to RNs (called R-UEs)• Two types of user traffic with distinct bit rate and latency

requirements:– VoIP: 128kbps and latency deadline of 100ms– Video: 256kbps and latency deadline of 300ms

Measured the level of QoS satisfaction w.r.t. the achieved throughput and experienced packet delay

Percentage of QoS-satisfied users w .r.t. throughput and packet delay

0.020.040.060.080.0

100.0120.0

Video UEs VoIP UEs Video UEs VoIP UEs

Throughput Delay

Conventional

Proposed

Multi-Carrier RelayingMotivation:- Multi-carrier operation is necessary to

achieve IMT-A requirements- Baseline LTE relaying is single carrier

Proposed solutions:- Enable carrier aggregation on RN backhaul => higher capacity, bottleneck elimination- Utilize frequency domain ICIC => improved signal quality for RNs and UEs- Make dynamic carrier reconfiguration => resources on demand, higher flexibility, lower

power utilization

Allowed CAAllowed CAAllowed CAAllowed CANot allowed CANot allowed CA

Backhaul link is typically the

bottleneck

LTE Release 10 system configuration

Problems:- Performance of RN-attached UEs

limited by backhaul capacity- RN configuration is (semi-)static and

unable to follow dynamic conditions

Multi-Carrier Relaying Single carrier is often insufficient to support

traffic of RN backhaul (especially for multi-hop)

Carrier aggregation on RN backhaul provides:• Higher peak throughputs• Multi-carrier load diversity

Main gains:• +20% in 5%-ile throughputs• +11% in throughput Jain’s index

Carrier loadbalancing

Support for high traffic

High order carrier aggregation

High order carrier aggregation

Low order carrier aggregation

Low order carrier aggregation

Single carrier operation

Single carrier operation

Multi-Carrier Relaying Dynamic carrier reconfiguration:

• RN backhaul-access carrier reconfiguration for capacity balancing (bottleneck elimination)

• Dynamic deactivation of secondary RN access carriers in case of low traffic load (energy efficiency)

• Main gains:+6% capacity for RN UEs (+4% overall)-14% lower RN access carrier activity

Frequency domain ICIC:• Coordinated selection of RN backhaul and/or access

carriers to avoid interference• Iterative approach with several levels of decentralization

(centralized, distributed, autonomous)• Main gains: +14% capacity for RN UEs (+10% overall)


Type-2 relay nodes Transparent RNs

• RNs just expand the cell of DeNB

They replicate the cell ID of DeNB Type-2 RNs can be used for capacity and QoS

enhancement• They implicitly forward information

Have received functional definition by 3GPP but yet part of the standards


Challenges for type-2 relay nodes

How to best cooperate with eNB• Adaptive HARQ• Distributed coding between eNB and RNs

New transmission protocols enabling throughput enhancements


Adaptive HARQHigh Low

Less More

Residual FER

Resources forretransmission

OptimalThroughputperformance

Apply adjustment of the retransmission bits, considering the tradeoff between FER and resource utilization.

In case of Type 2 relay without channel decoding, further ideas of how to optimize retransmission bits are necessary

Closed form solution for Mutual Information of relayed link


Available Results/Analysis Possible gain of HARQ with memoryless Type 2

relay• Restricted to Chase combining• Inherent loss, SNR gain but no coding gain

SNR gain pays off the coding gain penalty for wide range of relay position

0 0.2 0.4 0.6 0.8 10.2

0.4

0.6

0.8

1

dSR

Nr/N

t

CC AFIR SourceCC Source


Possible saving with 16QAM

Saving of up to 45 % of resources for retransmission

EF most promising


Advanced cooperative distributed turbo coding techniques coupled with cooperative HARQ (1/2)

Proposition of novel distributed HARQ protocols for cooperative transmissions, where source and relay cooperatively construct a turbo code (DTC).

Question to be solved:• How ACK/NACK feedback can be exploited to efficiently select which agent(s) of the

cooperation scheme has to retransmit when a packet decoding error is detected at relay(s) and/or at destination ?

Selection of the node that retransmits:• Check the quality not only of the relay to the destination channel, but also of the source to

the destination one, prior to chose the node that will retransmit.• Exploit the instantaneous mutual information knowledge at both UE and relay locations Intelligent ACK/NAK feedback with additional information is exploited in order to select the

node that retransmits.

From Turbo Codes (TC) to Distributed Turbo Codes (DTC)

CC

CCDecoder CC

Destination

Interleaver

Source

Information bits + Source coded bits

Relay

DTC

TC

CC1

Decoder CC1

Interleaver

Source

CC2

Decoder CC2

Interleaver

DeinterleaverInterleaver

info

infopcc1

pcc2

pcc1

pcc2

Asymmetric case +10dB, 4 Tx max

0,00

0,05

0,10

0,15

0,20

0,25

-13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6Es/N0 (dB)

Nor

mal

ised

Thr

ough

put

ReTx Both - partner info - perfect SR link

ReTx via outage - partner info - R, S, or R+S - perfect SR link

ReTx via outage - partner info - imm.stop - R,S,R+S - perfect S-R link

idem above but S-R @ 20dB



SS

RR

DD

SS

RR

DD

N1 retransmissions,(maximum: Nmax)

TCC at first, TCC or CC1/2 after

(Nmax- N1) retransmissions

ACK@R

Yes

No

SS

RR

DD

Yes

No

Go to next packet

ACK@D

ACK@DYes

No

ACK@DYes

No

Outageafter Nmax

?

Yes

No

TCC or CC1/2

CC1/2

Phase 1

Phase 2

Phase 3


Advanced cooperative distributed turbo coding techniques coupled with cooperative HARQ (2/2)

Investigation of several HARQ schemes for cooperative networks:• Classical retransmission schemes where either both nodes retransmit the same

codeword or only one node with the highest average SNR retransmits. • Advanced schemes based on mutual information computation and outage check

Proposed strategies achieve interesting gain over classical retransmission schemes:

• The most promising schemes are “retransmission of both partner information” (lowest PER) and “retransmission of the partner information by one node via outage check” (highest link throughput).


Adaptive resource allocation with interference mitigation for in-band relay nodes (1/2)

Proposition of novel frequency reuse schemes aiming at:• Interference mitigation• Throughput and fairness increase between macro and relay nodes• Taking into account the system load and the level of interference

Frequency adaptive allocation scheme based on:• Estimation for each type of traffic of the number of PRBs required for both

relayed and macro UEs• Allocation of the minimum number of PRBs in order to make sub-bands

orthogonals as much as possible between adjacent sectors

,

, 2RB M M M

RB R R R

N NN N

Nb MCS

1m

sp mNbBits m


Adaptive resource allocation with interference mitigation for in-band relay nodes (2/2)

Adaptive frequency partitioning allows improvements w.r.t. FR1 scheme both in term of number of UEs with satisfied QoS and in term of Jain index.

2

2Jain index i ii i

x N x

Moving relay nodes


RNs mounted on top of public transportation vehicles (buses, trains etc)• Consist of inter-connected outdoor and indoor antennas

Enhance performance of vehicular UEs whose number is greatly rising• Overcome vehicular penetration loss (VPL)• Facilitate handovers and manage mobility

A study item for 3GPP

Moving relay nodes for buses


With moving relay nodes (MRNs), VPL can be reduced or even eliminated.

Users experience better signal reception and the system capacity can be increased.

Moving relay nodes for trains


Challenges for moving relay nodes


Rate adaptation for backhaul links• Channel prediction can be employed

Handover issues• Guaranteeing no blocking and dropping for vehicular UEs

Interference coordination

Conclusions


The research work of ARTIST4G in the field of relay shows significant potentials for further technical enhancements

The involved companies will continue the research and will be prepared for a coming evolution step for relays.

Documents

Ppt0000000.ppt [Lecture seule] - ITU: Committed to ... · Transversal Supporting Activities WP4: Impact of innovations on the RAN ... interference management (with 3D beam-forming)