A Train Integrity Solution based on GNSS Double-Difference Approach

A. Neri1, F. Rispoli2, P. Salvatori1, and A.M. Vegni1

1 RADIOLABS, Rome Italy{alessandro.neri, pietro.salvatori, annamaria.vegni}@radiolabs.it

2 Ansaldo STS, Genoa Italy, francesco.rispoli@ansaldo-sts.com

ION GNSS+ 2014 - Tampa (FL) - U.S.A.

ION GNSS+ 2014

Roadmap• Introduction to ERTMS/ETCS

• Train integrity issue

• Proposed solution

• Protection Level evaluation

• Simulation results

• Conclusions

ION GNSS+ 2014

ERTMS/ETCSERTMS / ETCS (European Railway Traffic Management System / European Train Control

System) is the standard for European railways. There are three levels:

Train Integrity

ETCS trainborne

DRIVER

Interlocking and

Radio Block Center

Balise (fixed message)

Level 1

Level 2

Level 3

Present

future Train must be able to evaluate its own

position and to determine whether no carriage has been decoupled

Virtual Track Circuit

Track circuit

LEU

ETCS trainborne

balise

signal

Movement authorities

Track circuit

ETCS trainborne

RBCMovement authorities

Balise

(fixed message)

Interlocking

ION GNSS+ 2014

Train IntegrityIssue:

With the term train integrity we mean the ability to determine whether allthe carriages are still coupled each others.

Goal:

Define a Virtual Track Circuitto reduce operational costand increase the line capacity.

Solution:

Use a double difference approach between a couple of GNSS receiverlocated respectively at the head and at the end of the train. In such a wayit is possible to estimate the train length with a little time delay.

END-OF-TRAIN OBU GNSS RX

HEAD-OF-TRAIN OBU GNSS RX

ION GNSS+ 2014

Why satellite technology?Main advantages are:

• Reduction of operational and maintenance cost

• Increasing of line capacity

Market perspective:

• Cost-effective solution to increase safety on low traffic lines• Increase traffic on high-speed lines

Main challenge:

• Fulfill the SIL-4 requirements in terms of THR (Tolerable Hazard Rate) imposed for railways (i.e. THR ≤ 10-9/h)

ION GNSS+ 2014

Reference architecture

EGNOSRadio Block Center

(RBC)

TALS

RS 1 RS 2 RS n…

Track Area Augmentation Network

EDAS

GNSS CONSTELLATIONSSPACE SEGMENT

LDS & Train Integrity

ION GNSS+ 2014

Double Difference approach

( )

ˆ ˆ ˆ1 , , ,

h

h

H H E E

RxSat Sat Track

i i i H i

RxSat Sat Track

i i E i

i i i i

Rx Rx Rx Rx

SD T k s T k

T k s T k

r

X X

X X

e e b e

H E

ij

Rx Rx i jDD SD SD

ˆ ˆ ˆ1 , ,H H E E

i i i i

Rx Rx Rx Rxr

e e b e

ˆ ˆ ˆ1 , ,H H E E

j j j j

Rx Rx Rx Rxr

e e b e

ˆ ˆ ˆ ˆ1 , 1 ,H H E H H E

i i i j j j

Rx Rx Rx Rx Rx Rxr r

e e e e

ˆ ˆ, .E E

i j

Rx Rx b e e

END-OF-TRAIN OBU GNSS RX

HEAD-OF-TRAIN OBU GNSS RX

HRxERx b

90

ˆH

i

Rxeˆ

be

ˆE

i

Rxe

ˆ ˆ,E

i

b Rxb e e

ˆ ˆ,H

i

b Rxb e e

iSat

Mitigation of most of the iono, tropo and clocks errors

ION GNSS+ 2014

Railway Constraint

( )hRx

mX

( 1)ˆ hRx

mX

( 1)hRx

mX

( ) ( )ˆh

m m

h bb e

We adopt a constrained positioningalgorithm to map the 3-D estimationproblem into a 1-D estimation problem

( )

( )

( )

( )

( )

( )

ˆ

ˆ

H

H

mH

E

E

mE

Rx

m

b

s s k

Rx

m

b

s s k

s

s

Xe

Xe

%

%

( ) ( ) ( ),ˆ ˆ

H E

m m m

b b

G e e

( ) ( ) ( )m m m

constrH H G g

ION GNSS+ 2014

Simulation tool

OBU

End GNSS

Rx

RIM RURIM RU

Scenario

definition

Track

& Train

DB

RIM RU

DB

EM

Environment

DB

DTM AnalyticsGNSS

records

DB

LAAS

Servers

DB

Satellite

Orbit

Generator

GNSS

Tx Signals

Generator

Propagation RIM RU

Train

Motion

GNSS

Localization

TAAS

IP

Backbone

TALS

Server

OBU

Head GNSS

Rx

Train Control

Command

Generator

GNSS

Localization

GNSS

Train Integrity

ION GNSS+ 2014

Protection Level Evaluation

True p-th

satellite

location

RxE RxH

Nominal p-th

Satellite

location

b

p

r

RxH

p eRx

H

p

r

RxE

p eRx

E

p

We can define protection as astatistical over bound of the gapestimation error. In fact

ˆ

decoupled

L

coupled

We have to link the trainintegrity issue with thesatellite integrity issue

ION GNSS+ 2014

Performance Assessment

2 21 1

1

1

21 ,

e

N NRIM RIM

TrainIntegrity

TH

enc

Dec fe Max SF

Rk erfc

N D D P P

10-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

100

0

2

4

6

8

10

12

14

THR

PL

[m

]

SLOPE=0.1

SLOPE=0.3

SLOPE=0.5

SLOPE=0.7

SLOPE=0.9

ION GNSS+ 2014

Protection Level for single faultTo have a numerical reference of the protection level let us consider:

2Maxdd

50B km

1Max

g

2.5b km

0.14SLOPE 7PL m

In such a way it is possible to derivethe Virtual Circuit Length consideringthe dynamical model of the train(dynamic coupling junctions)

B

VTC

b

PL

ION GNSS+ 2014

Simulation results

Train weight 1275 ×103 kg

Carriage weight 35 t

Locomotive weight 120 t

fR 0.02

FR 24.9×104 N

Receiver noise model N (0.1,0.8)

Results are provided for both a typical passenger train (500 m lengthtravelling at 108 km/h) and a heavy freight one (2500 m length with cruisespeed of about 80 km/h)

ION GNSS+ 2014

Train Length estimation error

Short Train

Long Train

Scenario with all coupled carriages

ION GNSS+ 2014

Intercarriage gap vs time to alarm

95 97 99 101 103 105 107 109 111 113 115495

500

505

510

515

520

525

Elapsed time from beginning of simulation [s]

Mile

ag

e b

etw

ee

n r

ece

ive

rs [m

]

Estimated mileage between receivers

Real mileage between receivers

95 97 99 101 103 105 107 109 111 113 1152495

2500

2505

2510

2515

2520

2525

Elapsed time from beginning of simulation [s]

Mile

ag

e b

etw

ee

n r

ece

ive

rs [m

]

Estimated mileage between receivers

Real mileage between receivers

We considered the following scenario:

• The train moves at constant speed• The train is a rigid block (no dynamic

coupling between carriages)• One of the carriages decouples from the

previous one• The front train section continues its

movement after the decoupling as ifnothing has been occurred

• The tail section stops only by action ofrolling resistance

• Track slope effect has been neglected

ION GNSS+ 2014

Train Length estimation error

Short Train

Long Train

Scenario with one decoupled carriage

ION GNSS+ 2014

Dynamic effect mitigation

0 200 400 600 800 1000 1200-5

0

5

Elapsed time from beginning of simulation [s]

Estim

atio

n e

rro

r o

n M

ilea

ge

be

twe

en

re

ce

ive

rs [m

] Train length L=2500 m

Median Filter

-10 -5 0 5 100

0.1

0.2

0.3

0.4

0.5

Mileage estimation error [m]

Pro

bability

-4 -3 -2 -1 0 1 2 3

0.0010.003

0.010.02

0.050.10

0.25

0.50

0.75

0.900.95

0.980.99

0.9970.999

Data

Pro

babili

ty

Normal Probability Plot

We use a median filter with window sizeequal to 10 epochs. In this way we reducethe outlier number in the error distributionby increasing the time to alert

ION GNSS+ 2014

Conclusions • Train Integrity function is key for the introduction of the ERTMS L3

system with GNSS technology

• We focused on the Virtual Track Circuit definition to estimate at the same time the train position and its length by the:– computation of Protection Level to evaluate the gap that can be

protected by Virtual Track Circuit

– verification that the theoretical model fulfils the SIL-4 (Safety Integrity Level 4) requirements

• Computer-based simulations have demonstrated the performance in terms of estimation error with different train lengths:– 250 m

– 2500 m

• A median filter approach has been introduced to minimize the outlier number in the distribution by increasing the time to alert

ION GNSS+ 2014

Thank You for your kind attention