lte-handover on high speed trains - winst.com.t · pdf fileprisma telecom testing srl ......

PRISMA Telecom Testing Srl | Private & Confidential | Vers. 1.6

LTE-Handover on High Speed Trains: In-lab experimental study on the current

and future solutions

A.Parichehreh, P.Marini, Luca Dell’Anna, Umberto Spagnolini

2016.10.19

IWPC Workshop

Evolving the Internet of Things

Turin, October 18-20, 2016

Confidential – Version 1.6

Agenda

Massive handovers in HST scenario

Introduction to HST onboard Connectivity

UeSIM (User Equipment Simulator)

Two-tier Architecture for on-board wreless connectivity

Experimental Result

In-Lab Exerimental Setup

Conclusion


Introduction to HST On-board Internet Service

• Satellite Communication • Low bandwidth and high latency • Disconnections across tunnels • Cost of communication

• GSM-R • signaling of train and

ground station • 114 kbit/second for GPRS

• HSPA/HSPA+ • 3G technology based on CDMA • 14 Mbps in DL and 5.76 Mbps in UL.

• LTE/LTE-A • 4G technology based on TDMA/FDMA • 100Mbps peak data rate in high mobility scenario (>350km/h).

HST with speed up to 500 km/h

• Challenges • Varying channel and Doppler Shift • Penetration loss • Frequent handovers • Resource limitations

High speed train (HST) is becoming one of the preferred mid-range

transportation and the on-board Internet service is a must for train operators.


Problem definition of Handover (HO) in mobility system

Low mobility scenario train setting (e.g., v=70km/h = 20m/s, carriage length=25m) • Arrival time of carriage i+1 is larger than the HO time of carriage i • All HO related to one carriage is concluded before the next carriage

Cell border

HO latency

RS

RP

[dB

m]

Ms

Mn

H

Os

On

TTT

A3 E

vent

Meas. R

eport


Problem definition of Handover (HO) in mobility system

High mobility scenario train setting (e.g., v = 350km/h = 97m/s, carriage length=25m) • Arrival time of carriage i+1 is smaller than the HO time of carriage i

Handovers (HOs) of multiple UEs (20-60UE/carriage) and carriages (8-10carriage/train) must be done within a short time to guarantee QoE over the RACH signaling with heavy signaling overload. Basic idea and motivation:

• To study and verify the performance of different manufacturer for this special challenging scenario

Cell border

HO latency Depend on the HST speed/load

RS

RP

[dB

m]

Ms

Mn

H

Os

On

TTT

A3 E

vent

Meas. R

eport


Amplify and Forward (AF)-LTE Relaying

Radio Relay

Technology General Description Advantages/Disadvantages

Type-1

Layer-1 relay

(analog repeater)

AF relay

AF of on-board UEs (transparent relay)

with full-duplex capability.

+

• Simple and inexpensive

• Minimum impact on the LTE standard

(already defined in LTE Rel.8)

• Suitable solution when carriage are shielded

-

• Noise and inter-cell interference amplification

• Frequent HOs, the same as baseline scenario (with direct link between UEs and

ground eNB)

LTE Relay LTE Relay

HO region L

TE

serv

ice

rate

Moving direction


Decode and Forward (DF)-LTE Relaying

Radio Relay

Technology General Description Advantages/Disadvantages

Type-2:

Moving Relay Node

(MRN)

Acting with the same functionalities of a

base station, non-transparent for on-

board UEs.

DF half-duplex relay.

+

• Noise and inter-cell interference mitigation capability

• Group HOs and Tracking Area Updates (TAU)

-

• Requires its own PCID and non-transparent to UEs

• UEs need mobility management for joining to MRN

• Extra resource allocation overhead for LTE backhaul link between Donner evolved

NodeB (DeNB) and MRN (Un), and MRN and UEs (Uu).

• Needs further modifications on LTE specification (LTE Rel.12)

MRN MRN

HO region L

TE

serv

ice

rate

Moving direction

Uu

Un

Uu


UeSIM Architecture

• Protocol processor and IP traffic generator. Processes the lower protocol layers for LTE/WCDMA/GSM (MAC, RLC, PDCP)

• Simulation control and Control Plane processing

• Optimized implementation of RRC/NAS

• SDR Unit provides multi-standard baseband processing and includes the RF interface • The same SDR Unit can be configured to work with any 3GPP radio access technology and any frequency band: GSM, UMTS, LTE/FDD, LTE/TDD

SDR Units S1 eNodeB

cell 1

GbE

PE-Wireshark AirMosaic

GUI

eLSU

GbE Test Manager

Workstation

eNodeB

cell 2

eNodeB

cell 3

S1

S1

ePC

Core Network

Logging workstation

tracing from Radio to

Application Layer

User Workstation controlling

test scenario, mobility,

application, radio conditions


TCP measurement of the ensemble of on-board UEs

UE 1

UE N

eN

B

Core

ne

two

rk

UE N

UE 1

UE 2 UE 2

Onboard

UE

s

Internet

Services

D=1km

0

0

0

0

0

Moving direction v=300kmph

a)

b)

c)

d)

e)

UE 1

UE 2

UE N

eNB1→ eNB2

eNB2→ eNB3

eNBK-1→ eNBK

eNB1 eNB2 eNB3 eNBK

PHY

UE N

PHY

MAC

MAC

MAC

MAC

MAC

MAC

UE 1

UE 2

UE N

UE 1

UE 2

PHY level

settings

BB Interf eNodeB

cable

Mu

ltiu

ser

traff

ic

gen

era

tor

(vid

eo,

HT

TP

, F

TP

)

eNodeB

Cell k

Cell k+1

Simulated group of UEs via UeSIM LTE eNB & core network

LTE HO procedure

On-board UEs

UE1 –CH2 UE1-CH1

UE2 –CH2 UE2-CH1

UE2 –CH2 UEN-CH1

UE1 (TCP

connection)

UE2 (TCP

connection)

UEN TCP

connection) Tra

ffic

an

aly

sis

(QoS

, th

roughput)

eNB

port 1

port 2

cable

eNB

UE1 (TCP

connection)

UE2 (TCP

connection)

UEN (TCP

connection)

PD

CP

RLC

MA

C

PD

CP

RLC

MA

C

PD

CP

RLC

MA

C

(L1-L3) PHY


Network Configuration

Network Configuration

eNB

LTE (Rel-11) band 7

downlink uplink bandwidt

h

2620MHz 2500MHz 10MHz

Cell size 1km

Off-track distance 50m [10]

UE Category Cat 4, 2×2 MIMO, 150Mbps (with 64QAM MCS)

Mobility scenario 300Km/h

Number of UEs {8,10,15,18,22,30} per carriage

HO Configuration

Hysteresis value 0dB

A3-offset 2dB

Time-To-Triger value 40ms

RACH Configuration

PRACH Configuration Index 3

Total number of dedicated preamble 56

Maximum preamble transmission 10

RA response window size 8ms


Throughput of the AF-LTE HST

,HN

)t,h,i(TThroughputAverage

t

N

1i

H

1hPDSCH

t

where N is the total UEs on-board and Ht is the non-zero hits in the throughput measured at time t.

100

Avera

ge T

hro

ughput

[Mbps] 15x4 UEs (L1 relay)

0

c)

HO transition

100

Avera

ge T

hro

ughput

[Mbps]

30x4 UEs (L1 relay)

0

f)

HO transition

8x4 UEs (L1 relay)

-5 0 +5 Time (s)

-10 +10

D=1km

0

a)

HO transition

100

Avera

ge T

hro

ughput

[Mbps]

10x4 UEs (L1 relay)

0

b)

HO transition

100

Avera

ge T

hro

ughput


0

e)

HO transition

100

Avera

ge T

hro

ughput


0

d)

HO transition

100

Avera

ge T

hro

ughput

[Mbps]

a)

-5 0 +5 Time (s)

-10 +10

D=1km

-5 0 +5 Time (s)

-10 +10

D=1km

8x4 UEs (L1 relay)

By increasing the number of UEs, not only throughput reduces, but also the HO transient time increases (due to the HO signaling

of massive UEs)


Service Time of the AF-LTE HST

0 0.01 0.02 0.03 0.04 0.05 0

0.2

0.4

0.6

0.8

1

Service Time [ms]

cdf

8x4 in cell

l cell = -5770

l HO =-848

8x4 in HO

5 10 15 20 0

0.2

0.4

0.6

0.8

1

Service Time [ms]

cdf

10x4 in cell

l cell = -5.302

l HO = -0.6992

10x4 in HO

0 5 10 15 20 0

0.2

0.4

0.6

0.8

1

Service Time [ms]

cdf

15x4 in cell

l cell = -4.378

l HO = -0.7443

15x4 in HO

0 5 10 15 20 0

0.2

0.4

0.6

0.8

1

Service Time [ms] cdf

18x4 in cell

l cell = -1.847

l HO = -0.5897

18x4 in HO

0 5 10 15 20 0

0.2

0.4

0.6

0.8

1

Service Time [ms]

cdf

22x4 in cell

l cell = -1.133

l HO = -0.5309

22x4 in HO

0 5 10 15 20 0

0.2

0.4

0.6

0.8

1

Service Time [ms]

cdf

30x4 in cell

l cell = -0.7582

l HO = -0.4583

30x4 in HO

a) b)

c) d)

e) f)


CDF of Handover Latency of the AF-LTE HST

HO latency shows an exponential behaviour with the rate parameter depending on the cell load

10 0

10 1

10 2

10 3

0

0.2

0.4

0.6

0.8

1

t= RRC complete

- RRC config

(ms)

EC

DF

Experimental 4X4 UEs

F(t; t ° )=1-e

-0.275t

Experimental 30x4 UEs

F(t; t ° )=1-e

-0.002t

0 200 400 600 800 1000 1200 0

0.2

0.4

0.6

0.8

1

t= RRC complete

-RRC config

(ms)

EC

DF

4x4 UEs

8x4 UEs

15x4 UEs

18x4 UEs

22x4 UEs

30x4 UEs

1 2 3 4 5 6

0.5

1

1 2 3 4 5 60

500

𝜏°

HO

ave

rag

e la

ten

cy

8x4 10x4 15x4 18x4 22x4 30x4


Multicell Access Scheme

HST with relay nodes (Layer-1 relay or MRN) equipped with directional antenna

Basic idea: • Manipulating the HO decisions of on-board relays with spatial adaptation of the signal.

a)

Moving direction

mobile relay (L1 or MRN)

forward

omni-directional

backward

dispatcher

b)

eNBk+1 eNBk eNBk-1


HST Outage Probability: Theoretical Analytics

HST Outage Probability Analysis:

where

so that

where

and

Note that

1

2

)()()1(N

i

N

out

i

outout

HST

out PPPP

)()),(1(),( )()()()(

th

i

k

iii

out PHHPkk

kj

k

))H(P1(1)H,( )i(

k

)i(

j

)i(

2

k

2

j

kk

jj

10

)i(

k

)i(

j /x))t(,(g

x))t(,(glog10HQ)H(P

.2

1)( 2

2

x

x

dxexQ

(1)

(2)

(3)

(4)

(5) kkk10thk

)i(

k /x))t(,(glog10Q1)(P


Computer Simulation Results: Train Outage Analysis

Outage probability anaylsis [1]

[1] A. Parichehreh, S. Savazzi, L. Goratti, U. Spagnolini: Seamless LTE connectivity in high-speed trains. Wireless Commu.and Mobile Computing

16(12): 1478-1494 (2016)


Computer Simulation Results: Train Outage Analysis

Overlapped HOs can happen due to mispositioning of eNBs along the track.

Before the HO of directional antenna

Subject to:

Directional antenna postpones its HO affirmatively in favor of omni-directional antennas to prevent HO overlapping.

(1)

(2a)

(2b)

},{minarg],[ )()1(

},{

)()1(

)()1(

N

HOHO

N PPN

max0 PPout

max

)()1(

min },{ N

directional antenna HO

Geometrical position [km]

0.5 1 1.5 2 2.5 3 3.5 25

30

35

40

45

50

Beam

wid

th (∆𝜃

)

Time

𝑇𝑜𝑓𝑓 On service eNB #2

𝑇𝑜𝑓𝑓 On service period eNB #2 On service eNB #1

On service eNB #3 Direct.

Omni.

10 -4

10 -2

10 0

Outa

ge p

robabili

ty

P out direct

P out omni

Delay in HO

#1 #2

#3


Multicell Access Scheme Experimental results: PRISMA MRN and layer-1 relay with Directional Antennas versus

Omni Antennas

0 0.5 1 1.5 20

0.2

0.4

0.6

0.8

1

Average Throughput [Mbps]

CC

DF

L1

MRM

L1+MA

MRN+MA

0 200 400 600 800 10000

0.2

0.4

0.6

0.8

1

RRCCompleted

-RRCConf ig

time (ms)E

CD

F

MRN+MA

L1 + MA

L1

(a) (b)


Conclusion

• In-Lab test has been set to validate the performance with “actual” eNB from major

manufacturers.

• In-Lab experimental results proves the QoS/QoE impairments of HO in HST

• Fixed-beam directional antenna is compatible with all the existing on-board relay

solutions and it is a promising solution to provide multi-cell access.

• Advantages:

Further reduction of the HO signaling and interruption time

Better QoS provisioning via multi-cell access mechanism

Distributing/balancing the on-board load among multiple LTE cells

No change in the current LTE specification.

Further tests for higher speeds (400-450kmph) are planned with more eNBs (from other

manufacturers), mixed UMTS/LTE technologies (inter-RAT HO)

More info on In-Lab setup: http://www.prismatelecomtesting.com/


Via Petrocchi 4 I 20127 Milano

T+39.02.26113507 F+39.02.26113597

[email protected]

www.prismatelecomtesting.com

C.F. e P.IVA IT 08876610968

Thank You!

mailto:[email protected]

http://www.prismatelecomtesting.com/

lte-handover on high speed trains - winst.com.t · pdf fileprisma telecom testing srl ......

Documents