X-Map Estimation Michaela Neuland, TUBS Mehdi Amirijoo, Ericsson - - PowerPoint PPT Presentation

FP7 ICT-SOCRATES

X-Map Estimation

Michaela Neuland, TUBS Mehdi Amirijoo, Ericsson

WWW.FP7-SOCRATES.EU

Outline

• Introduction
• Study overview
• Current status
• Challenges

2/15

WWW.FP7-SOCRATES.EU

Drive Tests

• Primary objectives of drive testing is to identify network performance using a

reasonable sample of locations in the network

• Identify performance in terms of, e.g.,

– Coverage (e.g., pilot power) – Accessibility (e.g., random access) – Retainability (e.g., hand over) – Quality (e.g., throughput and speech quality)

• Sampling carried out manually by drive/walk tests
• Automation of drive tests has been proposed by NGMN due to limitations in

drive/walk testing

3/15

WWW.FP7-SOCRATES.EU

Drive Tests Only a limited part of network can be analysed Drive/walk test are costly Drive tests only capture a snapshot of the conditions in the field

4/15

WWW.FP7-SOCRATES.EU

X-map Estimation

• Main principle:

– Connect UE event/measurements with estimated position – Gather UE reports to build map relating geo reference data and metric of interest

• A map can indicate, e.g.

– Path loss – Interference

• Used to detect, e.g.

– Coverage holes – Service quality – Traffic density (used for e.g., site planning)

5/15

WWW.FP7-SOCRATES.EU

Study

• Aim of study is not to develop positioning mechanisms

– Quite siginificant work already done – Not within the scope of network management – Positioning is being studied for LTE

• We assume that UE positioning is in place
• Model accuracy of UE positioning techniques as a function of

– RAT – radio environment – number of measured cells – etc.

• Model accuracy of UE measurements (RSRP, PL etc.)
• Use models to find map accuracy by means of simulations

6/15

WWW.FP7-SOCRATES.EU

Study Output

• What is the map accuracy that can be obtained using today’s and

tomorrow’s technology?

– WCDMA positioning – GPS – LTE R8/R10

• Understand map accuracy as a function of:

– UE positioning accuracy – UE measurement accuracy – Number of measurements taken

• Are there other ways of improving accuracy?

– Can UE measurements be combined with prediction data to obtain better

accuracy?

– Can we combine several sources from different RATs, GPS, prediction data?

7/15

WWW.FP7-SOCRATES.EU

Study Output

• What is the positioning and measurement accuracy needed to obtain maps

with sufficient accuracy?

– Determine requirements on, e.g. UEs and positioning mechanisms – Determine feasibility of map generation

• Impact on interfaces, e.g., UE-eNodeB, X2
• Impact on UEs
• Appropriate triggering of measurement reports

– Identification of time + locations in the network

8/15

WWW.FP7-SOCRATES.EU

• Small realistic scenario of

1.5 km x 1.5 km in Braunschweig

• Static and mobile users based on

a mobility model

• Network information available
• Realistic path loss information

derived from a prediction model

• Decorated user snapshots

Simulation Scenario

Source: Google Earth 5.0

9/15

WWW.FP7-SOCRATES.EU

• For LTE three different localisation methods are planned

– GPS – Observed Time Difference of Arrival (OTDOA) – Enhanced cell ID positioning method

• Model for the position error based on the Cramér-Rao lower bound found in

the literature

• This model is based on the

– Geometry of eNodeBs / satellites and the UE – Number of measured signals – Standard deviation of the measurement error

(for GPS: 33.3 ns)

Position Error Modelling

10/15

WWW.FP7-SOCRATES.EU

• At the moment successive positions are uncorrelated
• Next step: applying some kind of filter to get a "flat" route

Position Error Modelling - GPS

11/15

WWW.FP7-SOCRATES.EU

• mean error: 7.4 m
• standard deviation: 4.9 m

Position Error Modelling - GPS

position error in m distribution function

12/15

WWW.FP7-SOCRATES.EU

• How can we find realistic values for the achievable position accuracy and

reliability?

• Can we improve the position error modelling?
• What are the error sources for localisation methods and how can we

consider them?

• Can we benefit from combining different localisation methods?

Challenges

13/15