• No results found

Closed Loop Control Scheduling in Multihop Cellular Networks

N/A
N/A
Protected

Academic year: 2021

Share "Closed Loop Control Scheduling in Multihop Cellular Networks"

Copied!
20
0
0

Loading.... (view fulltext now)

Full text

(1)

Closed Loop Control Scheduling in Multihop

Cellular Networks

VDE/ITG Workshop Contribution

Dr.-Ing. Rainer Schoenen

ComNets, RWTH Aachen

(2)

Outline

• Packet vs. resource scheduling:

• Packet scheduling

: QoS, Priorities etc.

• Resource scheduling

– Fading channel

(frequency & time varying)

– Channel State Information (CSI,CQI)

– OFDMA scheduling under fading conditions

– Dynamic Subcarrier Assignment (DSA)

– Adaptive Modulation&Coding (AMC)

– Adaptive Power Control (APC)

Closed Loop Control Resource Scheduling

(3)

Motivation

packet scheduling

Packet scheduling vs. resource scheduling:

- Packet scheduling chooses packets/bits from queue, handles QoS

- Resource scheduling allocates OFDMA subchannel,

(4)
(5)

Scheduler Performance

Delay:

Pr

{

d

>

t

}

t

t

t

t

~ a

e

-bt

FCFS

EDF

Round Robin Weighted Round Robin

~ a

e

-b(t+x)

higher rate

higher rate

1 0.1 0.01 1e-3 1e-4 1e-5 1 0.1 0.01 1e-3 1e-4 1e-5 1 0.1 0.01 1e-3 1e-4 1e-5 1 0.1 0.01 1e-3 1e-4 1e-5 100 100 100 100 0 0 0 0

(6)

QoS with priorisation

Separation of QoS classes

High channel utilisation due to unlimited best effort traffic

and flow control

typical load

on a controlled channel:

100 %

0 . 0 0 0 5

0 . 0 0 1

0 . 0 0 1 5

0 . 0 0 2

1 0 %

4 0 %

1 0 0 %

1 5 0 %

C B R 0 . 0 5 W

V B R 0 . 0 5 W

A B R 0 . 0 5 W

C B R 0 . 0 0 0 1 W

V B R 0 . 0 0 0 1 W

A B R 0 . 0 0 0 1 W

Total offered load average packet delay [s]

with CDMA CBR VBR ABR time Throughput load Last

(7)

Why is the scheduler so complicated?

Why is the scheduler so complicated?

Components of the scheduler:

Components of the scheduler:

– CQI : Channel Quality Indication

– DSA : Dynamic Subcarrier Assignment

– AMC : Adaptive Modulation & Coding

– APC : Adaptive Power Control

– Multi-Antenna: MIMO/Beamforming

– Resource Partitioning

– QoS : Priorities and Substrategies

– Buffer/Queue management

Advanced Wireless Scheduling

resource scheduling

packet scheduling

(8)

FDD Multihop Frame Scheduling

for LTE Advanced

S y n c R A C H BC H S y n c UL DL time f B S UT S y n c R A C H BC H S y n c UL DL f f UT RN S y n c R A C H BC H S y n c UL DL S y n c R A C H BC H S y n c UL DL f half duplex Switch Rx + Tx Full duplex half duplex Full duplex 10ms

(9)

Half-Duplex FDD Frame Scheduling

Map Data BS UT HalfDuplex Group 1 RN UL DL fr e q ue nc y 2 1 1 2 2 1 1 2 2 1 1 2 2 1 1 2 0 1 2 3 4 5 6 7 framenumber / time UL DL fr e q ue nc y 2 1 1 2 2 1 1 2 2 1 1 2 2 1 1 2 0 1 2 3 4 5 6 7 framenumber / time UL DL fr e q ue nc y UT UT BS BS UT UT BS BS UT UT BS BS UT UT BS BS 0 1 2 3 4 5 6 7 framenumber / time UL DL fr e q ue nc y 1 2 2 1 1 2 2 1 0 1 2 3 4 5 6 7 framenumber / time RemoteUT HalfDuplex Group 1

(10)

Multihop Resource Partitioning

(between BS and RN)

BS

1

time

fr

eq

ue

ncy

DL

UL

BS

UT

TaskPhase

RN

2

%

67

BS

1

BS

1

RN

2

%

67

BS

1

BS

1

RN

2

BS

1

BS

1

RN

2

BS

1

BS

1

RN

2

BS

1

BS

1

RN

2

BS

1

...

...

frame 0 1 2 3 4 5

...

...

UT3 UT4 UT5 RUT6 RUT7 RUT8 BS1 RN2 hop1 hop1

(11)

Half-duplex multihop

(uplink) throughput

Offered total traffic [Mbit/s]

T hr ou gh pu t pe r st at io n [b it/ s] ( * 10 7 ) UT3 UT4 UT5 RUT6 RUT7 RUT8

Offered total traffic [Mbit/s]

T hr ou gh pu t pe r st at io n [b it/ s] ( * 10 7 ) UT3 UT4 UT5 RUT6 RUT7 RUT8

Using simple stateless scheduler Using stateful scheduler: ProportionalFair

(12)
(13)

BestChannel

CQI

LinearFFirst

CQI

Dynamic Subcarrier Assignment

• DSA strategies:

LinearFFirst, BestChannel, BestCapacity …

.

.

.

.

.

.

.

(14)

Adaptive Modulation & Coding

AMC: PhyMode choice depends on:

(15)

Control loop representation of scheduler

Adaptive Power

Control (Path loss + fading)Physical channel

+

RxPower PhyMode(f,t), TxPower(f,t)

+

Interference + Noise SINR(f,t)

Prediction of path loss Prediction of interference

(Kalman Filter) normalization

Dynamic Subchannel Assignment desired RxSINR (20dB)

+

measurement noise channel quality measurement Interpolation filtering

send filtered info back to BS

Averaging the filtered info

special procedures to get downlink channel quality at BS

Adaptive Modulation and Coding

SINR Estimation with nominal txPower

RxSINR with nominal TxPower

-IL Raw PathLoss Interference   maxPower BS UT z-1

Reference: Schoenen, R. et al, “Resource Allocation and Scheduling in FDD Multihop Cellular Systems”,

(16)

pathloss on downlink

Performance of adaptive resource scheduling

Dynamic Subcarrier Assignment

DSA Strategies Evaluation: (768m)

Resource usage: old method

(17)

Performance of adaptive resource scheduling

Adaptive Modulation & Coding

• AMC Strategies Evaluation: (1600m)

QPSK-1/3 QPSK-1/2 QPSK-2/3 QAM16-1/2 QAM16-2/3 QAM16-5/6 QAM64-2/3 QAM64-5/6

PhyMode usage: method 2 (WithCQI) on downlink

QPSK-1/3 QPSK-1/2 QPSK-2/3 QAM16-1/2 QAM16-2/3 QAM16-5/6 QAM64-2/3

PhyMode usage: method 1

(WithoutCQI) on downlink SINR: method 1 (WithoutCQI) on downlink

(18)

Performance of adaptive resource scheduling

Adaptive Power Control

• APC Strategies Evaluation: (768m)

TxPower: method 1 (UseNominalTxPower) on downlink

TxPower: method 2 (FCFSMaxPhyMode) on downlink

SINR: method 1 (UseNominalTxPower) on downlink

SINR: method 2 (FCFSMaxPhyMode) on downlink

(19)

Conclusions

Packet and resource scheduling

can and must be separated

QoS distinction by priorities is sufficient in the early phase

Sub-strategies are important for further QoS differentiation

and fairness

QoS aware scheduling and optimum utilization go hand in hand

Resource Scheduling and Resource Partitioning

happen on different timescales

The wireless link is a loop (DL+UL). Delay=RTT

(round trip time)

DSA and AMC are straightforward (open loop), but

APC requires a

closed control loop

system view

CQI, DSA, AMC and APC optimally utilize the channel capacity

(20)

Thank you for your attention !

Dr.-Ing. Rainer Schoenen

[email protected]

References

Related documents

These models predict that market structure, operating costs, bank risk aversion, money market interest rates volatility, credit risk, the covariance between interest rate risk

The Note will focus on three of the six sections of the USSPM: Section 2 ‘ Ensuring Board, Management and Employee Commitment to Social Goals ’, Section 3 ‘ Treating

While significant correlations between attenders and non-attenders were found for smoking, time of the appointment and score on the MSPQ, the only significant predictors of

Development of robotic instruments for bone removal will definitely improve the application of the surgical robot in fields like spine surgery and anterior skull base surgery,

the potential temperature tendency due to the longwave radiation (shading, K/hour), and extinction coefficient at 550nm (contour) at 00UTC 22 August 2010 for (a) the experiment

We included vendors that offer one or more solutions that were available for customers to use by October 1, 2014 and that provide at least the following core predictive

The comparison group was comprised of the 5,978 insured depository institutions that did not participate in the SBLF and (i) were established prior to March 31, 2009, (ii) had

Rare isotopologues and vibrationally excited states of known astronomical species, as well as new interstellar molecules will be detected by ALMA in the upcoming years.. For the