© DLR ACO TUT 2011 | Project Overview | ICL-GNSS, Tampere, Finland… · · 2015-04-27© DLR ACO TUT 2011 | Project Overview | ICL-GNSS, Tampere, Finland, ... Project Overview

© DLR ACO TUT 2011 | Project Overview | ICL-GNSS, Tampere, Finland, June 29-30 2011 Page 1 | 30.06.2011


Motivation

» New Galileo signals:» Improved accuracy, integrity and authentication» Massive wave of new applications in key

downstream markets

» Total market of upstream and downstream European GNSS based industry: About €300bn in 2020 [L.E.K. Consulting]

» Now time to build a successful European GNSS industry

» R&D in GRAMMAR: Boost Galileo downstream industry by providing IP for future Galileo mass market receivers


Galileo and GPS Radio Front-End

» Follow-up of FP6 GREAT results E1-E5a switching receiver chipset



» Custom RF FE measurements with laboratory (CW) and real signals

Gain Range

-70

-60

-50

-40

-30

-20

-10

0

10

20

30

0,E+00 5,E+06 1,E+07 2,E+07 2,E+07 3,E+07

Frequency [Hz]

Out

put P

ower

[dB

m]

Gain Range

-70

-60

-50

-40

-30

-20

-10

0

10

20

30

0,E+00 5,E+06 1,E+07 2,E+07 2,E+07 3,E+07

Frequency [Hz]

Out

put P

ower

[dB

m]



» E1-E5a Switching receiver single chip integration

» Two LNAs (L1/E1 & L5/E5a)» Reconfigurable down-conversion and IF sections» Digital control» Functional integration (no global area

optimization)



» Dual Channel low power E1-E5a receiver


Galileo and GPS Baseband:TUTGNSS Receiver Architecture

» Dual-Frequency GNSS Receiver» Web Server Debugging Interface» FPGA Altera Stratix II with Nios II softcore processor


Galileo and GPS Baseband:Navigation Software

» Dual-Frequency/Dual-System Navigation Software

» LS/EKF Algorithms Implemented


Galileo and GPS Advanced Baseband Algorithms:Tracking and Multipath Mitigation

» Multi-correlator-based structure» A huge number of correlators: 50-100 correlators» Reduced Search Space Maximum Likelihood (RSSML) algorithm

Block diagram of multi-correlator based DLL structure


Galileo and GPS Advanced Baseband Algorithms:Tracking and Multipath Mitigation

» DLR suburban (car) channel model» Comparison with traditional NEML

DLL» Correlator spacing: 0.025 chips» Infinite radio front-end BW

» RSSML » Good amount of memory» 81 complex correlators» Very good performance at the

cost of expensive hardware design


Galileo and GPS Advanced Baseband Algorithms:E5 Signal Simulink Model

» Transmitter: AltBOC(15,10) modulated E5 signal, IF signal (fIF= 26 MHz)


Galileo and GPS Advanced Baseband Algorithms:Acquisition

» Complex E5a reference signal» 1 ms integration

Example of time frequency correlation outputafter 1ms, C/N0=100dB-Hz

-1.5 -1 -0.5 0 0.5 1 1.50

0.5

1

1.5

2

2.5

3

3.5

4

4.5

x 104

chips

AC

F

Auto-Correlation Function at zero delay


Galileo and GPS Advanced Baseband Algorithms: Ionospheric Corrections and Range Estimation

» Traditional» Two pseudorange measurements for two

unknowns: Range and ionospheric delay» Linear Least Squares method (LLS)

» Minimize square difference between observed and modeled data

» Enhanced methods» Constrained Linear Least Squares (CLS):

» Impose constraints to avoid physical violations (negative ionospheric delay)

» Constrained Iterative Mean (CIM)» Constrained iterative range estimation

via measurement averaging» Simple implementation, less complex

than CLS

RMSE [m] versus standard deviation of measurement error [chips] for E1-E5a frequency combination


Hybrid Data Fusion:WLAN Positioning

Generation of radio map

Two approaches:•PDF approximations for probabilistic methods•RSS averages for pattern recognition

…

Access points

Samples

Coordinates

Calibration data (RSS)

…Probabilistic Methods:• Histograms• Coordinates

Pattern recognition:• Pattern vector• Coordinates

Summarize



Offline phase:Generation of radio map (= fingerprint database)•Received signal strengths•As a function of location

Online phase:Position estimation using• Radio map •Current RSS measurements

0 10 20 30 400

10

20

30

40

-87.56

-79.13

-70.69

-62.25

-53.81

-45.38AP#17

RSS(dBm)



Performance of positioning

Average error distance •A function of location

Average error (m)

0 10 20 30 400

10

20

30

40

2.81

5.63

8.44

11.25

14.06

16.88


Hybrid Data Fusion:Integration of GNSS/WLAN

» Positioning Handover

» Outdoor-to-Indoor or Indoor-to-Outdoor Environments


» 3GPP LTE: Cramer-Rao lower bound

Hybrid Data Fusion: 3GPP Long Term Evolution (LTE)


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Basic idea

» Time difference of arrival (TDoA)» Mobile stations and base stations not synchronized

» Communication systems not build for navigation» Multiple access interference?» Cell boundary issues?» Initial acquisition?


Hybrid Data Fusion: GNSS and 3GPP Long Term Evolution (LTE)

» 3GPP LTE OFDM receiver platform for data fusion

State-of-the-artGNSS Receiver

GRAMMARGNSS RX Platform

GRAMMAR3GPP LTE

TDOA Platform

Ground truthTime stamp

Extended Kalman Filter,Particle Filter


Introduction: Non-Line-of-Sight Problem

» Non-Line-of-Sight» Direct path blocked» Signals propagate additional distance» Adds positive bias» Causes positioning estimation error

» Thus:» Detect NLOS signals» Mitigate bias


Analytical dual frequency NLOS detection

» Analytical NLOS detection:» Derivation of

» False Alarm Probability » Detection Probability

» For dual frequency receiver» With DLR land mobile channel model


Analytical NLOS detection for LTE systems

» To distinguish between NLOS and LOS signals

Serving BS: lower threshold At cell-edge or for neighboring BSs: higher threshold


Introduction: Raytracing Urban Scenario

» Urban scenario» Building height up to 26 m» Fixed base stations and GNSS

constellation» Realization of different tracks



» NLOS Situations for GPS and Galileo + GPS

» Only GPS Satellites » GPS and Galileo Satellites



» Perfect knowledge of LOS satellites

» Positioning with GNSS» Additionally using

LTE network for positioning

0 10 20 30 40 500

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

x [m]

CD

F, P

(err

or <

x)

GPSGPS+GalileoTDOA, close to BSTDOA, cell edgeGPS+Galileo+TDOA, close to BSGPS+Galileo+TDOA, cell edgeGPS+Galileo, all scenarios


Hybrid Positioning with GNSS and 3GPP-LTE NLOS Detection

» Confidence metric, heuristic approach:

» Threshold depends on scenario

Example P

τ

P

τ

LoSNLoS

1 1

max

log log 1 ( ) logN

n c m

LoSNLoS

max 1



GNSS: 24 satellites

No NLOS outliers due to conservative threshold» often less than 4 satellites» increased RMSE



Using 3GPP-LTE in addition» decreases max. error» increases availability

GNSS + 3GPP-LTE


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Transmitters

» Generic transmitter system needed (baseband and RF) to emulate communication systems» Transmit OFDM modulated signals:

» WLAN, 3GPP-LTE, WiMAX, etc.

» 4 base stations realized with 2 FPGA boards:» 20 MHz bandwidth, 32/64 symbols» CP, Nfft, Preamble fully variable» Variable symbol mapping» MATLAB interface

FPGA Baseband TX 1Digital Upconverter


Synchronization (Sync)

60 MHz IF signal



2.4 GHz RF signal

PA

PA

PA

PA

Sync

Sync

OFDM SymbolCP OFDM Symbol OFDM SymbolCP OFDM SymbolCP....

t

Preamble Data



» Typical, generic OFDM based transmitter (not related to a specific standard):

» Our baseband solution:» Define data, framing and pilots in MATLAB» Covers most common OFDM based communication standards

Channel Coder

Scrambler / Interleaver Mapper

S/P

P/S

IFFT Cyclic Prefix Insertion

Datastream(Bitstream)

Pilot Generator

Mux

Synchronization Sequence Generator

Mux

Complex baseband signal

OFDM Frame – Memory32 or 64 OFDM Symbols

P/S

IFFT Cyclic Prefix Insertion

Digital Up-Conversion

Real IF signal



» System On Chip (SOC) design» Digital Up-Conversion

» Hardware efficient halfband interpolation filters

» Radio frequency chain with mostly COTS devices» Discrete devices due to IF signal

» Filters (custom cavity filter)» Mixers» Power amplifiers

NIOS IICPU

Timer

Graphics-Display

Text-Display

PIOs

PLLs

SGDMARx + Tx

Descriptor Memory

TSE-MACClock Crossing Bridge

DDR-2 Memory Controller

Synchronization Core

TopOfdm

UpConversion



Tx 2 IF signalTx 1 IF signalEthernet connection

to MATLAB

Board Board synchronizationnot shown here


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Receiver

» Generic receiver needed» Flexible hardware for algorithm testing and verification

» Sampling frequency up to 48 MHz at 14 bit resolution» I/Q and bandpass subsampling mode» Continous data storage up to 10 minutes» MATLAB interface to gather data blockwise

» Directly process raw data in MATLAB!

» COTS WiFi RF front end from MAXIM» Single chip solution for 2.4 GHz

Complex baseband sampling or bandpass subsampling

COTS WLAN Frontend2.4 GHz ISM-Band

AD

C I/Q

OFDM Baseband-Processor +

Timing/Position - EstimationPosition – Information

Raw TDoA estimation

Control



» COTS workstation» Additional Solid States Disk Drive (SSD) RAID» Delivers enough throughput for continous data storage» Flexible solution with all PC interfaces, like USB, Ethernet, WiFi etc.

Stratix IV GXFPGA

Altera PCIe Dev. Kit

terASIC THDB_ADA

IntelX58

ChipsetIntel i7 CPU

PCIe RAID SSD

DDR3 RAM

DMA Buffer #2

DMA Buffer #1

PCIe LinkGen1 x4



Maxim WiFi front end

FPGA board

SSD RAID


» Example: PSD of a grabbed test frame» Transmitter Tx 1 only» No data symbols used» PSCH and SSCH according to LTE

» ~ 1.25 MHz bandwidth» Scattered pilots as in LTE

» ~ 19 MHz bandwidth

» Note: » Codes as defined in LTE» Larger subcarrier spacing


Scattered pilots PSCH & SSCH



» Simple scenario: 2 transmitters» 2 different receiver positions

» Rx1:» Received power from BS1 and BS2 very

similar» Interference due to cross correlation

artifacts» Rx2:

» BS2 not detectable with one-step receiver

» Iterative receiver needed to cancel interference» 3GPP-LTE frequency reuse of 1!



» New access scheme» Implicit interference cancelation

» Detect all assumed transmitters» SIC stage to refine estimations

» Analog imperfections?» Known for communication



» Simulation results: Ideal conditions, AWGN channel

» 2 transmitters 1 TDoA» SNR of 10 dB» Different relative power ratios

» Result after acquisition and 1 SIC iteration


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Initial acquisition on real data

» Carrier Frequency Offset (CFO) isvery crucial

» Initial acquisition works on PSCH only Zadoff Chu codes

» Scattered pilots are very sensitive to CFO Can not be used

Correlation peaks ofpilots not visible


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Initial acquisition on real data

» Integer CFO and fractional CFO compensated

» The PSCH peaks changed in time» PSCH can not be used directly for

accurate timing synchronization

» Peaks of scattered pilots areclearly visible after CFO compensation

» Peak ambiguity resolved byPSCH peak detection


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Example 1 of a real channel

» Peak clearly visible and detectable


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Example 2 of a real channel

» Which peak to detect? Which is the right one?» Multipath cancelation due to narrow band signal


Hybrid Data Fusion: 3GPP Long Term Evolution (LTE): Channel sounder measurements

» Goal: Determine a real lower bound for the LTE testbed: What could we expect?» First measurements carried out:

» Fully synchronized channel sounder measurements» 2.4 GHz – 2.5 GHz ISM band» 20 MHz bandwidth with fully occupied subcarriers





» Model order estimation for SAGE how many multipath components?» PDF of all reference points 5 to 6 multipath components

Documents

© DLR ACO TUT 2011 | Project Overview | ICL-GNSS, Tampere, Finland… · · 2015-04-27© DLR ACO TUT 2011 | Project Overview | ICL-GNSS, Tampere, Finland, ... Project Overview