DATAMAN. VOR Base Stations for Indoor Positioning. Dragoş Niculescu. MOBILE COMPUTING LABORATORY

Σχετικά έγγραφα
ΕΙΣΑΓΩΓΗ ΣΤΗ ΣΤΑΤΙΣΤΙΚΗ ΑΝΑΛΥΣΗ

3.4 SUM AND DIFFERENCE FORMULAS. NOTE: cos(α+β) cos α + cos β cos(α-β) cos α -cos β

Aquinas College. Edexcel Mathematical formulae and statistics tables DO NOT WRITE ON THIS BOOKLET

Bayesian statistics. DS GA 1002 Probability and Statistics for Data Science.

the total number of electrons passing through the lamp.

Biostatistics for Health Sciences Review Sheet

Section 8.3 Trigonometric Equations

ECE 468: Digital Image Processing. Lecture 8

Statistics 104: Quantitative Methods for Economics Formula and Theorem Review

Υπολογιστική Φυσική Στοιχειωδών Σωματιδίων

Areas and Lengths in Polar Coordinates

ΚΥΠΡΙΑΚΗ ΕΤΑΙΡΕΙΑ ΠΛΗΡΟΦΟΡΙΚΗΣ CYPRUS COMPUTER SOCIETY ΠΑΓΚΥΠΡΙΟΣ ΜΑΘΗΤΙΚΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ 19/5/2007

Sampling Basics (1B) Young Won Lim 9/21/13

Numerical Analysis FMN011

Probability and Random Processes (Part II)

Main source: "Discrete-time systems and computer control" by Α. ΣΚΟΔΡΑΣ ΨΗΦΙΑΚΟΣ ΕΛΕΓΧΟΣ ΔΙΑΛΕΞΗ 4 ΔΙΑΦΑΝΕΙΑ 1

Section 8.2 Graphs of Polar Equations

«ΑΝΑΠΣΤΞΖ ΓΠ ΚΑΗ ΥΩΡΗΚΖ ΑΝΑΛΤΖ ΜΔΣΔΩΡΟΛΟΓΗΚΩΝ ΓΔΓΟΜΔΝΩΝ ΣΟΝ ΔΛΛΑΓΗΚΟ ΥΩΡΟ»

Second Order RLC Filters

The Simply Typed Lambda Calculus

Areas and Lengths in Polar Coordinates

Other Test Constructions: Likelihood Ratio & Bayes Tests

Approximation of distance between locations on earth given by latitude and longitude

ΠΕΡΙΕΧΟΜΕΝΑ. Κεφάλαιο 1: Κεφάλαιο 2: Κεφάλαιο 3:

Statistical Inference I Locally most powerful tests

ANSWERSHEET (TOPIC = DIFFERENTIAL CALCULUS) COLLECTION #2. h 0 h h 0 h h 0 ( ) g k = g 0 + g 1 + g g 2009 =?

: Monte Carlo EM 313, Louis (1982) EM, EM Newton-Raphson, /. EM, 2 Monte Carlo EM Newton-Raphson, Monte Carlo EM, Monte Carlo EM, /. 3, Monte Carlo EM

Spherical Coordinates

ΚΥΠΡΙΑΚΟΣ ΣΥΝΔΕΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ CYPRUS COMPUTER SOCIETY 21 ος ΠΑΓΚΥΠΡΙΟΣ ΜΑΘΗΤΙΚΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ Δεύτερος Γύρος - 30 Μαρτίου 2011

CHAPTER 25 SOLVING EQUATIONS BY ITERATIVE METHODS

Μηχανική Μάθηση Hypothesis Testing

EE 570: Location and Navigation


A Bonus-Malus System as a Markov Set-Chain. Małgorzata Niemiec Warsaw School of Economics Institute of Econometrics

Solar Neutrinos: Fluxes

6.003: Signals and Systems. Modulation

Section 7.6 Double and Half Angle Formulas

Homework 8 Model Solution Section

EPL 603 TOPICS IN SOFTWARE ENGINEERING. Lab 5: Component Adaptation Environment (COPE)

10/3/ revolution = 360 = 2 π radians = = x. 2π = x = 360 = : Measures of Angles and Rotations

derivation of the Laplacian from rectangular to spherical coordinates

ΠΑΝΕΠΙΣΤΗΜΙΟ ΠΑΤΡΩΝ ΤΜΗΜΑ ΗΛΕΚΤΡΟΛΟΓΩΝ ΜΗΧΑΝΙΚΩΝ ΚΑΙ ΤΕΧΝΟΛΟΓΙΑΣ ΥΠΟΛΟΓΙΣΤΩΝ ΤΟΜΕΑΣ ΣΥΣΤΗΜΑΤΩΝ ΗΛΕΚΤΡΙΚΗΣ ΕΝΕΡΓΕΙΑΣ

Graded Refractive-Index

UDZ Swirl diffuser. Product facts. Quick-selection. Swirl diffuser UDZ. Product code example:


Fourier Series. MATH 211, Calculus II. J. Robert Buchanan. Spring Department of Mathematics

Homework 3 Solutions

ST5224: Advanced Statistical Theory II

Exercises 10. Find a fundamental matrix of the given system of equations. Also find the fundamental matrix Φ(t) satisfying Φ(0) = I. 1.

Hadronic Tau Decays at BaBar

Lecture 34 Bootstrap confidence intervals

1. Ηλεκτρικό μαύρο κουτί: Αισθητήρας μετατόπισης με βάση τη χωρητικότητα

HISTOGRAMS AND PERCENTILES What is the 25 th percentile of a histogram? What is the 50 th percentile for the cigarette histogram?

Presentation Structure

Improvement of wave height forecast in deep and intermediate waters with the use of stochastic methods

Monolithic Crystal Filters (M.C.F.)

Models for Probabilistic Programs with an Adversary

ES440/ES911: CFD. Chapter 5. Solution of Linear Equation Systems

Instruction Execution Times

PARTIAL NOTES for 6.1 Trigonometric Identities

Study of In-vehicle Sound Field Creation by Simultaneous Equation Method

TMA4115 Matematikk 3

Module 5. February 14, h 0min

Second Order Partial Differential Equations

is like multiplying by the conversion factor of. Dividing by 2π gives you the

Review Test 3. MULTIPLE CHOICE. Choose the one alternative that best completes the statement or answers the question.


An Inventory of Continuous Distributions

b. Use the parametrization from (a) to compute the area of S a as S a ds. Be sure to substitute for ds!

Math 6 SL Probability Distributions Practice Test Mark Scheme

Πτυχιακή Εργασι α «Εκτι μήσή τής ποιο τήτας εικο νων με τήν χρή σή τεχνήτων νευρωνικων δικτυ ων»

Trigonometric Formula Sheet

Durbin-Levinson recursive method

ΕΛΛΗΝΙΚΗ ΔΗΜΟΚΡΑΤΙΑ ΠΑΝΕΠΙΣΤΗΜΙΟ ΚΡΗΤΗΣ. Ψηφιακή Οικονομία. Διάλεξη 10η: Basics of Game Theory part 2 Mαρίνα Μπιτσάκη Τμήμα Επιστήμης Υπολογιστών

HOMEWORK 4 = G. In order to plot the stress versus the stretch we define a normalized stretch:

상대론적고에너지중이온충돌에서 제트입자와관련된제동복사 박가영 인하대학교 윤진희교수님, 권민정교수님

Inverse trigonometric functions & General Solution of Trigonometric Equations

Πανεπιστήμιο Κρήτης, Τμήμα Επιστήμης Υπολογιστών Άνοιξη HΥ463 - Συστήματα Ανάκτησης Πληροφοριών Information Retrieval (IR) Systems

CE 530 Molecular Simulation

ΕΛΛΗΝΙΚΗ ΔΗΜΟΚΡΑΤΙΑ ΠΑΝΕΠΙΣΤΗΜΙΟ ΚΡΗΤΗΣ. Ψηφιακή Οικονομία. Διάλεξη 9η: Basics of Game Theory Mαρίνα Μπιτσάκη Τμήμα Επιστήμης Υπολογιστών

Buried Markov Model Pairwise

5.4 The Poisson Distribution.

ΠΑΡΑΜΕΤΡΟΙ ΕΠΗΡΕΑΣΜΟΥ ΤΗΣ ΑΝΑΓΝΩΣΗΣ- ΑΠΟΚΩΔΙΚΟΠΟΙΗΣΗΣ ΤΗΣ BRAILLE ΑΠΟ ΑΤΟΜΑ ΜΕ ΤΥΦΛΩΣΗ

VBA Microsoft Excel. J. Comput. Chem. Jpn., Vol. 5, No. 1, pp (2006)

APPENDICES APPENDIX A. STATISTICAL TABLES AND CHARTS 651 APPENDIX B. BIBLIOGRAPHY 677 APPENDIX C. ANSWERS TO SELECTED EXERCISES 679

Strain gauge and rosettes

Lifting Entry (continued)

Potential Dividers. 46 minutes. 46 marks. Page 1 of 11

(1) Describe the process by which mercury atoms become excited in a fluorescent tube (3)

DESIGN OF MACHINERY SOLUTION MANUAL h in h 4 0.

The Global Positioning System

Computing the Gradient

Shenzhen Lys Technology Co., Ltd

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

Ακριβής 3Δ Προσδιορισμός Θέσης των Σημείων του Κεντρικού Τομέα του Δικτύου LVD με τη μέθοδο του Σχετικού Στατικού Εντοπισμού

RF series Ultra High Q & Low ESR capacitor series

2 Composition. Invertible Mappings

9.09. # 1. Area inside the oval limaçon r = cos θ. To graph, start with θ = 0 so r = 6. Compute dr

[1] P Q. Fig. 3.1

NMBTC.COM /

Transcript:

DATAMAN MOBILE COMPUTING LABOATOY VO Base Stations for Indoor 80. Positioning Dragoş Niculescu dnicules@cs.rutgers.edu

indoor positioning no GPS indoors! classifications of existing systems infrastructure existing 80. base stations specialized beacons measurement medium F (radio frequency) infrared ultrasound actual positioning method triangulation trilateration signal strength map Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

positioning basics. range based - trilateration. angle based - triangulation 3. signal strength map - nearest neighbor, Bayes Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

trilateration N α N A M β B γ N C (x M x A ) + (y M y A ) = MA (x M x B ) + (y M y B ) = MB solve for (x M, y M ) (x M x C ) + (y M y C ) = MC MA, MB, MC are affected by errors several methods available Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

example: Cricket (MIT) mobile pros: cons: measures ranges to ceiling beacons TDOA between F and ultrasound triangulates good accuracy.3m extensive infrastructure line of sight to the beacons Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

triangulation N α N A M β B γ N C (x M x A ) sin α = (y M y A ) cos α (x M x B ) sin β = (y M y B ) cos β solve for (x M, y M ) (x M x C ) sin γ = (y M y C ) cos γ α, β, γ -affected by errors (Gaussian) several methods available Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

signal strength map build SS map: for each point, measure SS to all 5 BS query: measure SS to 5 BS best match in the map Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

example: ADA (Microsoft) use nearest neighbor to decide for the best match pros: cons: no additional infrastructure requires high resolution sampling people, furniture, BS position affect the SS map median position error 3m Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

example: LANDMAC (MSU) replaces training points with FID tags BS with FID readers mobile uses an FID tag nearest neighbor, etc. pros: cons: no remapping updates map on the fly additional infrastructure (FID) many readers, tags Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

VOBA - VO BAse stations goals: no signal strength map less infrastructure move complexity to the 80. base station use: angles ranges angles and ranges Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

VOBA prototype 90 I receiver I sender 50 0 60 30 antenna 80 0 30 0 0 3 0 0 80. card 0 330 prototype 0 300 70 directional antenna pattern Dragoş Niculescu VO Base Stations for Indoor 80. Positioning

basic measurements ents signal strength [dbm] 6 6 66 68 70 7 7 76 mean SS 3 78 3 discrete angles angle distribution range SS(α) Dragoş Niculescu VO Base Stations for Indoor 80. Positioning 3

experiments frag replacements 3 3 measurement points 5 + base stations N/E/S/W measurements of 3- revolutions each Dragoş Niculescu VO Base Stations for Indoor 80. Positioning 3

best peak distribution 0.6 histogram of number of peaks 0.6 Histogram of SS rank mean =.5 peaks of best peak ents 3 0 probability 60% 30% 3 5 6 7 8 0 3 5 6 7 8 number of peaks peak rank.5 peaks on average best peak is first/second 90% of the time Dragoş Niculescu VO Base Stations for Indoor 80. Positioning 3

other peak distribution ments 3 true direction 3 true direction 5% 33% other peaks point away from true direction Dragoş Niculescu VO Base Stations for Indoor 80. Positioning 3

discrete angle positioning replacements 3 cumulative probability 0.75 0.5 0.5 using best angle using first two angles angle distribution 0 0 6 8 0 6 error in meters 3.5m median position error 3m if we knew the best peak Dragoş Niculescu VO Base Stations for Indoor 80. Positioning 3

ln ents triangulation analysis 3 3 3 ln σ a 0.00 0.00 simulation lower bound 0 0. 0.5 0. 0.5 0.3 0.35 0. 0.5 0.5 0.55 0.6 0.65 ln 0.0 0.0 simulation lower bound 0 0 0.05 0. 0.5 0. 0.5 0.3 0.35 0. 0.5 0.5 σ a σ a 0.0 0.0 simulation lower bound 0 0 0.6..8. 3 3.6..8 ln V ar[x] > σ a ln m V ar[x] - standard dev. of positioning error - density of basestations / m to improve positioning:. decrease measurement error σ a. use more basestations 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln quantized angles ents 3 ln σ a 00 00 000 800 600 00 00 0 00 00 500 0 500 000 500 000 500 500 o + + ++ + 8x78 55x908 8x358 650x79 776x30 00 00 0 00 00 600 800 000 00 00 600 800 000 00 00 500 000 500 0 ln 3 σ a 500 0 500 000 500 000 500 500 00 00 000 800 600 00 00 0 00 00 o + + 8x78 55x908 8x358 650x79 776x30 00 00 0 00 00 600 800 000 00 00 600 800 000 00 00 500 000 500 0 measurements rounded to the nearest 5 simulation 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln quantized angles replacements 3 ln cumulative probability σ a 0.75 0.5 0.5 0 best angle (non quantized) quantization 5 quantization.5 quantization 90 0 6 8 0 error in meters little degradation for 6 directions (.5 ) 8 directions (5 ) 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln angle distribution placements 3 ln σ a 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln range inference open space attenuation: d(ss) SS[dBm] = SS 0 [dbm] log 0 ( d d 0 ) n obtained through fitting known to be unreliable we obtain it from integration of SS(α) 5-fold cross validation corridor basestations - waveguide effect median range error.8m 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln positioning w. ranges frag replacements 3 ln cumulative probability 0.75 0.5 0.5 σ a 0 0 6 8 0 error in meters trilateration 5 base stations median position error.5m 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln ranges and angles frag replacements ln 3 σ a α N A M β N B γ N C x M = x A + MA cos α = x B + MB cos β = x C + MC cos γ y M = y A + MA sin α = y B + MB sin β = y C + MC sin γ one base station is theoretically enough α, β, γ, MA, MB, MC - affected by errors 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln position uncertainty ln σ r σ y σ a A σ a r M σ x approximate uncertainty as an ellipse error ellipse increases with distance σ a = 0. radians σ r = 0.r 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning σa

ln σ a uncertainty w. ranges & angles Sfrag replacements how to combine 3 several readings? Kalman filter. ln ρ α BS σ a BS BS 3 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning α ρ BS BS BS3

ln σ a positioning w. ranges & angles 3 ln σ a cumulative probability α ρ BS BS BS 3 0.75 0.5 0.5 7 BS 5 BS 3 BS 0 BS 0 6 8 0 6 error in meters more base stations better positions.m median position error (all 7 BS) 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning α ρ BS BS BS3

ln σ a discussion triangulation with large outliers use more than two angles? no correlation between angle error and distance angle error and SS corridors waveguides revolving signal at the mobile? data performance? 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning α ρ BS BS BS3

ln σ a summary VOBA = VO base station complexity into the base station less infrastructure no SS map revolving basestation measures SS(α) to derive discrete angles angle distributions ranges works with quantized angles as well can achieve.m - m median error 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning α ρ BS BS BS3

ln σ a index indoor positioning trilateration Cricket triangulation SS map ADA LANDMAC VO BAse station prototype basic measurements experiment setup peak distribution angle positioning discrete angles quantized angles angle distributions range positioning range+angle pos. uncertainty performance discussion summary 3 Dragoş Niculescu VO Base Stations for Indoor 80. Positioning α ρ BS BS BS3