Fingerprinting-based Indoor Positioning Dr. R. Montoliu, Dr J. - - PowerPoint PPT Presentation

▶
fingerprinting based indoor positioning
SMART_READER_LITE
LIVE PREVIEW

Fingerprinting-based Indoor Positioning Dr. R. Montoliu, Dr J. - - PowerPoint PPT Presentation

Jan 22, 2024 143 likes •1.7k views

Fingerprinting-based Indoor Positioning Dr. R. Montoliu, Dr J. Torres-Sospedra, Dr. A. Prez-Navarro, Dr. J. Conesa, Dr. O. Belmonte The main objective of this tutorial is: The students will learn how to develop a fingerprint-based

slide-1
SLIDE 1

Fingerprinting-based Indoor Positioning

  • Dr. R. Montoliu, Dr J. Torres-Sospedra,
  • Dr. A. Pérez-Navarro, Dr. J. Conesa, Dr. O. Belmonte
slide-2
SLIDE 2

The main objective of this tutorial is:

  • The students will learn how to develop a fingerprint-based localization

algorithm from zero avoiding the same mistakes we faced when we started.

slide-3
SLIDE 3

Professors:

  • Dr. R. Montoliu
  • Dr. J. Torres-Sospedra
  • Dr. A. Pérez-Navarro
  • Dr. J. Conesa
  • Dr. O. Belmonte

Phd student G. Mendoza

slide-4
SLIDE 4

Table of contents

Part Time Content 1 17:00-17:15 Introduction to fingerprinting 2 17:15-17:30 Theoretical background 3 17:30-17:45 The training step 4 17:45-18:15 Time to perform the training step 5 18:15-18:30 The operational step 6 18:30-18:45 Time to play with operational source code 7 18:45-19:00 How to improve the ILS 8 19:00-19:15 Time to improve the ILS 9 19:15-19:30 Awards

slide-5
SLIDE 5

Part 1: Introduction to fingerprinting

slide-6
SLIDE 6

Introduction to fingerprinting

  • Four types of indoor localization algorithms:

○ Deploy beacons ○ Use existing beacons and the position of the beacons is known ○ Use existing beacons and the position of the beacons is unknown ○ No use beacons

slide-7
SLIDE 7
  • 1. Deploy beacons
slide-8
SLIDE 8
  • 1. Deploy beacons

https://www.kinvey.com/wp-content/uploads/2014/05/beacon.jpg

slide-9
SLIDE 9
  • 1. Deploy beacons

Beacon ID Localization 1 [lat, long] ... [lat, long] N [lat, long]

slide-10
SLIDE 10
  • 1. Deploy beacons

http://www.clipartkid.com/images/679/pedestrian-20clipart-clipart-panda-free-clipart-images-gz3PIn-clipart.png

slide-11
SLIDE 11
  • 1. Deploy beacons

RSSI: Received Signal Strength Indication

slide-12
SLIDE 12
  • 1. Deploy beacons
  • Simplest solution:

○ The desired location is the one of the closest beacon

slide-13
SLIDE 13
  • 1. Deploy beacons
  • Better solution:

○ Apply trilateration

slide-14
SLIDE 14
  • 1. Deploy beacons
  • Better solution:

○ Apply trilateration

slide-15
SLIDE 15
  • 1. Deploy beacons
  • The position of the beacons is known

○ Trilateration techniques can be applied

  • High accuracy can be obtained

https://pixabay.com/static/uploads/photo/2013/07/13/13/24/fist-160957_640.png https://pixabay.com/static/uploads/photo/2013/07/13/13/24/fist-160958_640.png

  • Expensive
  • A lot of beacons can be needed in big scenarios
slide-16
SLIDE 16
  • 2. Use existing beacons (knowing the position)
  • Apply trilateration
slide-17
SLIDE 17
  • 2. Use existing beacons (knowing the position)
  • Apply trilateration

Note that the existing beacons could be deployed for providing some services, but not localization.

slide-18
SLIDE 18
  • 2. Use existing beacons (knowing the position)
  • The position of the beacons is known

○ Trilateration techniques can be applied

  • High accuracy can be obtained
  • Cheap, since we are using already deployed devices
  • High dependence of the existing beacons

○ Most of the times, without localization purpose

  • Low accuracy if there are a few number of beacons
slide-19
SLIDE 19
  • 3. Use existing beacons without knowing the position
  • Fingerprinting
slide-20
SLIDE 20
  • 3. Use existing beacons without knowing the position
  • Fingerprinting
slide-21
SLIDE 21
  • 3. Use existing beacons without knowing the position
  • Cheap, since we are using already deployed devices
  • High dependence of the existing beacons
  • Low accuracy if there are a few number of beacons
  • Less accuracy than previous cases
  • The position of the beacons is unknown

○

Trilateration techniques can not be applied

slide-22
SLIDE 22
  • 4. Without using beacons
  • Magnetic field based
slide-23
SLIDE 23
  • 4. Without using beacons
  • Magnetic field disturbances are constant
  • No devices are needed
  • The cheapest solution
  • Less discriminative power than WIFI fingerprinting
  • Not easy solution

○ Algorithms are in “work in progress” state

slide-24
SLIDE 24

A brief introduction to Fingerprinting based methods

  • Two main steps:

○ Training step ○ Operational step

  • Beacons are WIFI AP

○ Position is unknown

slide-25
SLIDE 25

The training step

slide-26
SLIDE 26

The training step

MAC RSII

xx-xx-xx-xx-xx-xx

  • 30db

...

  • 80db

xx-xx-xx-xx-xx-xx

  • 45db

Latitude Longitude

slide-27
SLIDE 27

The training step

MAC RSII

xx-xx-xx-xx-xx-xx

  • 30db

...

  • 80db

xx-xx-xx-xx-xx-xx

  • 45db

Latitude Longitude MAC RSII

xx-xx-xx-xx-xx-xx

  • 80db

...

  • 30db

xx-xx-xx-xx-xx-xx

  • 95db

Latitude Longitude

slide-28
SLIDE 28

The training step

Training Fingerprints database

slide-29
SLIDE 29

The training step

  • UJIIndorloc database:

○ https://archive.ics.uci.edu/ml/datasets/UJIIndoorLoc

  • Some data:

○ 21048 fingerprints ○ 520 different MACs ○ 4 multifloor building

  • Platform to share results
slide-30
SLIDE 30

The training step

  • UJIIndorloc database:

○ Joaquín Torres-Sospedra, Raúl Montoliu, Adolfo Martínez-Usó, Tomar J. Arnau, Joan P. Avariento, Mauri Benedito-Bordonau, Joaquín Huerta “UJIIndoorLoc: A New Multi-building and Multi-floor Database for WLAN Fingerprint-based Indoor Localization Problems” In Proceedings of the Fifth International Conference on Indoor Positioning and Indoor Navigation, 2014.

MAC001 MAC002 ... MAC520 Longitude Latitude Floor Building User Phone Timestamp

slide-31
SLIDE 31

The operational step

slide-32
SLIDE 32

The operational step

MAC RSII

xx-xx-xx-xx-xx-xx

  • 30db

...

  • 80db

xx-xx-xx-xx-xx-xx

  • 45db
slide-33
SLIDE 33

The operational step

Training Fingerprints database Indoor Localization System MAC RSII

xx-xx-xx-xx-xx-xx

  • 30db

...

  • 80db

xx-xx-xx-xx-xx-xx

  • 45db
slide-34
SLIDE 34

The operational step

  • A kNN based algorithm is used to obtain the localization
  • It will be explained after in this tutorial
slide-35
SLIDE 35

Discrete vs continuous positioning

  • This tutorial only cover discrete localization

○ The system only use the information captured in a particular time moment to estimate the location.

  • Continuos positioning:

○ The system use the last information captured and some historic data. ○ Tracking.

slide-36
SLIDE 36

Part 2: Theoretical background

slide-37
SLIDE 37

Waves

  • Wifi are electromagnetic waves and behave like them
  • They are affected by the following phenomena:

○ Reflection ○ Refraction ○ Diffraction ○ Absorption

slide-38
SLIDE 38

Electromagnetic waves

By Витольд Муратов (Own work) [Public domain], via Wikimedia Commons

slide-39
SLIDE 39

Electromagnetic Spectrum

Per Inductiveload, NASA [GFDL (http://www.gnu.org/copyleft/fdl.html) o CC-BY-SA-3.0 (http://creativecommons.org/licenses/by-sa/3.0/)], via la Wikimedia Commons

slide-40
SLIDE 40

Electromagnetic Spectrum

Per Inductiveload, NASA [GFDL (http://www.gnu.org/copyleft/fdl.html) o CC-BY-SA-3.0 (http://creativecommons.org/licenses/by-sa/3.0/)], via la Wikimedia Commons

slide-41
SLIDE 41

Electromagnetic Spectrum

Per Inductiveload, NASA [GFDL (http://www.gnu.org/copyleft/fdl.html) o CC-BY-SA-3.0 (http://creativecommons.org/licenses/by-sa/3.0/)], via la Wikimedia Commons

WiFi: 2,4 GHz 5 GHz

slide-42
SLIDE 42

Reflection and refraction

θi θr θt Separation interface n1 n2

slide-43
SLIDE 43

Reflection and refraction

θi θr θt Separation interface n1 n2

slide-44
SLIDE 44

Fresnel coefficients

slide-45
SLIDE 45

Fresnel coefficients

(for parallel polarization)

slide-46
SLIDE 46

Stationary waves

At the end will always be a minimum.

slide-47
SLIDE 47

Geometrical attenuation

r1 r2

slide-48
SLIDE 48

Difraction

Per Bbkkk (Treball propi) [GFDL (http://www.gnu.org/copyleft/fdl.html) o CC BY-SA 4.0-3.0-2.5-2.0-1.0 (http://creativecommons.org/licenses/by-sa/4.0-3.0-2.5-2.0-1.0)], via la Wikimedia Commons

slide-49
SLIDE 49

Absorption

Intensity Attenuation coefficient Penetration depth frequency magnetic permeability conductivity

slide-50
SLIDE 50

Example of real situation

Absorption A.P.

฀

slide-51
SLIDE 51

Example of real situation

Difraction A.P.

฀

slide-52
SLIDE 52

Example of real situation

Reflection A.P.

฀

slide-53
SLIDE 53

Number of people within a room

slide-54
SLIDE 54

Number of people within a room

slide-55
SLIDE 55

Bodies absorb WiFi radiation

Specific Absorption Rate (SAR) Power absorved mass Also exists the Whole Body SAR (WBSAR)

slide-56
SLIDE 56

Absorption cross section

Absorption Cross Section Power absorved Power density in the incident wave Silhouette area of a perfectly-absorbing surface that would absorb the same power as the loading object under discussion

slide-57
SLIDE 57

Dependence on tissue

Source: S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties of biological tissues:

  • III. parametric models for the dielectric spectrum of tissues,” Physics in Medicine

and Biology, vol. 41, no. 11, p. 2271, 1996.

Frequency of 1 cm penetration Dry Skin 5.2 GHz Infiltrated Fat 9.5 GHz Muscle 4.7 GHz

Source: Melia, Gregory (2013) Electromagnetic Absorption by the Human Body from 1 - 15 GHz. PhD thesis, University of York.

slide-58
SLIDE 58

Dependence on position

Source: Melia, Gregory (2013) Electromagnetic Absorption by the Human Body from 1 - 15 GHz. PhD thesis, University of York.

slide-59
SLIDE 59

Dependence on clothes

Source: Melia, Gregory (2013) Electromagnetic Absorption by the Human Body from 1 - 15 GHz. PhD thesis, University of York.

slide-60
SLIDE 60

Results

Source: Melia, Gregory (2013) Electromagnetic Absorption by the Human Body from 1 - 15 GHz. PhD thesis, University of York.

slide-61
SLIDE 61

All body mean approximation

Source: S. Garcia-Villalonga and A. Perez-Navarro, "Influence of human absorption of Wi-Fi signal in indoor positioning with Wi-Fi fingerprinting," Indoor Positioning and Indoor Navigation (IPIN), 2015 International Conference on, Banff, AB, 2015, pp. 1-10. doi: 10.1109/IPIN.2015.7346778

slide-62
SLIDE 62

Example

  • 83 points
  • 8.23 m of distance

between them

  • Few-bodies time (9 AM)
  • 115 points
  • 42 inside shops
  • 73 in corridors
  • 5.56 m of minimum distance

RadioMap Test

slide-63
SLIDE 63

Example

9 A.M. 4 P.M.

slide-64
SLIDE 64

Example

  • Results with

Airplace

  • w-kNN algorithm
slide-65
SLIDE 65

Summary

  • WiFi is an electromagnetic wave: microwaves.
  • It is affected by:

○ Reflection ○ Refraction ○ Absorption ○ Diffraction Changes in the environment can affect values of WiFi measured

  • Life tissues absorb microwaves.
  • Different tissues have different absorptions

Changes in the number of people affect values of WiFi measured

  • Different frequencies are affected in different ways.
slide-66
SLIDE 66

Part 3: The training step

slide-67
SLIDE 67

https://www.flickr.com/photos/shonk/57302289

https://www.flickr.com/photos/shonk/57302289

slide-68
SLIDE 68
slide-69
SLIDE 69
slide-70
SLIDE 70

Reference Data

slide-71
SLIDE 71

Reference Data

slide-72
SLIDE 72

Reference Data

slide-73
SLIDE 73

Reference Data

RSSI1, RSSI2, RSSI3, RSSI4, RSSI5

slide-74
SLIDE 74

Reference Data

Strong, Medium, Medium, Medium, Weak Medium, Strong, Strong, Medium, Medium Weak, Medium, Medium, Strong, Medium Weak, Weak, Medium, Medium, Strong

slide-75
SLIDE 75

Reference Data

Medium, Strong, Strong, Medium, Medium Medium, Strong, Strong, Medium, N/A Medium, Strong, Medium, Medium, Medium Strong, Strong, Strong, Medium, Medium

slide-76
SLIDE 76

Reference Data

https://pixabay.com/es/persona-icono-salida-de-emergencia-1332793/

Medium, Strong, Strong, Medium, N/A

slide-77
SLIDE 77

Reference Data

N/A, Weak, Strong, N/A, N/A Very extreme case

slide-78
SLIDE 78

Reference Data

Medium, Strong, Strong, Medium, Medium Medium, Strong, Strong, Medium, Medium Medium, Strong, Strong, Medium, Medium Medium, Strong, Strong, Medium, Medium

slide-79
SLIDE 79

Reference Data

https://commons.wikimedia.org/wiki/File:Mobile_phone_font_awesome.svg https://www.goodfreephotos.com/vector-images/mobile-cellphone-vector-clipart.png.php https://pixabay.com/es/smartphone-tel%C3%A9fono-m%C3%B3vil-tel%C3%A9fono-1132675/ https://commons.wikimedia.org/wiki/File:Mobile_phone.svg

Medium, Strong, Strong, Medium, Medium Medium, Medium, Strong, Medium, Medium Medium, Strong, Strong, Medium, Medium-Weak Medium-Weak, Strong, Strong, Medium, Medium Strong ~ -40dBm Strong ~ -50dBm Strong ~ -30dBm Strong ~ -35dBm

slide-80
SLIDE 80

https://i.ytimg.com/vi/OyRW9uFSmh0/maxresdefault.jpg

slide-81
SLIDE 81

Reference Data

Strong, Medium, Medium, Medium, Weak Medium, Strong, Strong, Medium, Medium Weak, Medium, Medium, Strong, Medium Weak, Weak, Medium, Medium, Strong

slide-82
SLIDE 82

Reference Data

https://pixabay.com/p-303768/?no_redirect

Cover all the environment Consider spatial density Consider temporal density Consider device heterogeneity Consider dynamics of the environment

slide-83
SLIDE 83

Reference Data

Now I have the training data... I have a perfect Indoor Location System

https://c2.staticflickr.com/4/3228/2373073659_d231a0cc65.jpg

slide-84
SLIDE 84

Reference Data

You need independent data to fine tune and validate your system

http://www.relatably.com/m/img/valid-memes/78c0b38fecebd3c736c8123b34fc69059aedce91ada224ee82677ba7707e14f9.jpg

slide-85
SLIDE 85

Validation data

Avoid using Training Data! It may provide slanted information. N consecutive fps: (N-1) Training and 1 for validation ? No!

slide-86
SLIDE 86

Validation data

slide-87
SLIDE 87

Validation data

slide-88
SLIDE 88

Validation data

slide-89
SLIDE 89

Validation Data

Necessary to: 1. Provide an estimation of the IPSs error: geometric error, hit detection rates,... 2. Calibrate your System: kNN algorithms and variants 3. Filter APs 4. Among many other useful operations :-)

slide-90
SLIDE 90

Validation Data

What happens if there is no Validation Data ? 1. k-fold Cross-Validation of training data 2. Consider groups of samples: ref point, user, device, day, among others to increase diversity and independence of the sets

~ 60-80% training, 40-20% validation

slide-91
SLIDE 91

Validation Data

Now I have the validation data... I have a perfect Indoor Location System

https://c2.staticflickr.com/4/3228/2373073659_d231a0cc65.jpg

slide-92
SLIDE 92

Validation Data

Your system may work fine All the contexts have not been considered

slide-93
SLIDE 93

Operational Data vs. Testing Data

Depending on the main objective of the IPS, you may have

1. Operational Data Fingerprints from working system + Feedback from users 2. Testing Data Fingerprints explicitly collected for testing Research: Training + Validation + Test Better if Test Data is Blind UJIIndoorLoc has T + V + Blind TS !!!

slide-94
SLIDE 94

Mapping Strategies

Slow procedure! High precision in reference points High precision on the fingerprint measures Consecutive / Cumulative values Dense radio map Simple

slide-95
SLIDE 95

Mapping Strategies

Slow procedure! High precision in reference points High precision on the fingerprint measures Consecutive / Cumulative values Dense radio map Simple

slide-96
SLIDE 96

Mapping Strategies

Slow procedure! High precision in reference points High precision on the fingerprint measures Consecutive / Cumulative values Dense radio map Simple

slide-97
SLIDE 97

Mapping Strategies

Fast procedure! You have a few reference or calibration points Depends on user’s velocity Fingerprint readings may be close or far to ref. points Fingerprint attached to a segment of the path The reference point may be displaced to the real path Light radio map Complex

slide-98
SLIDE 98

https://c1.staticflickr.com/9/8429/7755469546_61f6d51490_b.jpg

slide-99
SLIDE 99

https://pixabay.com/p-1546436/

slide-100
SLIDE 100

Storing Data

  • Raw data
  • Database: mySQL, HADOOP, mongoDB,...

Record as maximum information as possible!

mac, rssi, channel, bssid, .... position (XYZ), room, area, floor, building, ...

slide-101
SLIDE 101

Storing Data

  • CSV Files + Document (UJIIndoorLoc)

RSSI1 RSSI2 RSSI3 ... RSSIn X Y Z ... Others

  • 99

+100

  • 88
  • 55

...

Use of default value for non-detected signal +100 Use of a documented coordinate system XYZ Use of additional location info: office, area, floor, building, campus, city, ….

slide-102
SLIDE 102

https://cdn.meme.am/instances/500x/58606836.jpg

slide-103
SLIDE 103

Part 4: Time to training

http://www.soulseeds.com/wp-content/uploads/2011/10/take-ownership.jpg

slide-104
SLIDE 104

The App for training

Create new file to record fingerprints

slide-105
SLIDE 105

The App for training

Select number

  • f consecutive

fingerprints to record

slide-106
SLIDE 106

The App for training

Just Select the reference point, ...

slide-107
SLIDE 107

The App for training

Just Select the reference point, .... and Capture Data ….

slide-108
SLIDE 108

The App for training

.... and Capture More Data ….

slide-109
SLIDE 109

The App for training

DON’T FORGET TO UPDATE THE REFERENCE POINT !!!

slide-110
SLIDE 110

The App for training

Capture Even More Data

DON’T FORGET TO UPDATE THE REFERENCE POINT !!!

slide-111
SLIDE 111

The App for training

Finally, send the database to us by e-mail

slide-112
SLIDE 112

Part 5: The operational step

slide-113
SLIDE 113

The kNN algorithm

slide-114
SLIDE 114

The kNN algorithm

slide-115
SLIDE 115

The kNN algorithm

slide-116
SLIDE 116

The kNN algorithm

slide-117
SLIDE 117

The kNN algorithm

Algorithm with k=1

  • INPUT:

○ Training database (samples and labels of each sample) ○ Test sample

  • OUTPUT:

○ Label of the test sample

  • BEGIN

○ Estimate the distances between the test sample and all training ones ○ Return the label of the training sample with less distance

  • END
slide-118
SLIDE 118

The kNN algorithm

Algorithm with k>1

  • INPUT:

○ Training database (samples and labels of each sample) ○ Test sample

  • OUTPUT:

○ Label of the test sample

  • BEGIN

○ Estimate the distances between the test sample and all training ones ○ Get the labels of the k-th training samples with less distance ○ Return the majority label

  • END
slide-119
SLIDE 119

The kNN algorithm

slide-120
SLIDE 120

The kNN algorithm

slide-121
SLIDE 121

The weighted-kNN algorithm

  • k = 3
  • It is blue by simple voting
  • It is red by weighted voting
slide-122
SLIDE 122

The kNN algorithm in indoor localization

  • In the classical classification problems, each sample has a label
  • In indoor localization each sample (fingerprint) has two continuous values as

label: longitude and latitude.

slide-123
SLIDE 123

The kNN algorithm in indoor localization

  • With k=1

○ The localization of the test sample, is the localization of the closest sample in the training dataset. ■ “closest” in the feature space

slide-124
SLIDE 124

The kNN algorithm in indoor localization

  • Feature space is not the same than real space
slide-125
SLIDE 125

The kNN algorithm in indoor localization

  • Feature space is not the same than real space
slide-126
SLIDE 126

The kNN algorithm in indoor localization

  • Feature space is not the same than real space
slide-127
SLIDE 127

The kNN algorithm in indoor localization

  • With k>1

○ The localization of the test sample, is the centroid of the localizations of the k-th closest samples in the training dataset.

slide-128
SLIDE 128

The kNN algorithm in indoor localization

  • With k>1

○ The localization of the test sample, is the centroid of the localizations of the k-th closest samples in the training dataset.

slide-129
SLIDE 129

Source code explanation

function IPIN2016_Tutorial_TrainYourSystem conf = SetMyConfiguration(); data = ReadAllData(conf); data = ChangeDataRepresentation(data); for i=1:conf.experiment_repetitions folds = DivideInFolds(number_of_samples,conf); vmean_error_in_meters(i) = KnnWithCrossValidation(data, folds, conf); end mean_error_in_meters = mean(vmean_error_in_meters); fprintf('\n The method has obtained an error of %f meters.\n',mean_error_in_meters); end

slide-130
SLIDE 130

Source code explanation

function mean_error_in_meters = KnnWithCrossValidation(data, folds, conf) M = conf.number_of_macs; for i=1:conf.number_of_folds for j=1:conf.number_of_folds if (i==j) test_data = data(folds{i},1:M); test_labels = data(folds{i},M+1:M+2); else train_data = [train_data; data(folds{j},1:M)]; train_labels = [train_labels; data(folds{j},M+1:M+2)]; end end est_labels = ApplyKnn(train_data,train_labels,test_data,conf); verror(i) = EstimateMeanErrorMeters(est_labels,test_labels); end mean_error_in_meters = mean(verror); end

slide-131
SLIDE 131

Source code explanation

function est_labels = ApplyKnn(train_data,train_labels,test_data,conf) distance_matrix = GetDistanceMatrix(train_data,test_data,conf); for i=1:N_test vpos = GetPositionsOfTheMinimums(distance_matrix,conf); lat = 0; long = 0; for j=1:conf.k lat = lat + train_labels(vpos(j),2); long = long + train_labels(vpos(j),1); end est_labels(i,:) = [ long/conf.k, lat/conf.k ]; end end

slide-132
SLIDE 132

Part 6: Time to play with operation source code

slide-133
SLIDE 133

Competition rules (first phase)

  • Read SetMyConfiguration().
  • Test with using different parameter configuration.
  • Write in the competition paper the best solution.
slide-134
SLIDE 134

Competition rules (first phase)

  • You have only 20 minutes.
  • If you do not hand the paper in time,

your results will not be estimated.

slide-135
SLIDE 135

Part 7: How to improve the ILS

slide-136
SLIDE 136

How to improve the ILS

Presence of Noisy WAPs 1. SSID ‘iPhone of …’ 2. Very weak signal 3. Located at distant places 4. High variability in the same reference point

FEATURE SELECTION

slide-137
SLIDE 137

How to improve the ILS

Huge workload of the kNN algorithm: 1. Simplify reference dataset

a. Calculate means b. Remove noisy fingerprints in ref. point c. Remove repeated fingerprints

2. Apply clustering pre-stage

a. Group similar fingerprints - Representative FP

3. Reduce reference dataset on-the-fly

a. Common macs b. Strongest signal

CONDENSE & FILTER

slide-138
SLIDE 138

How to improve the ILS

kNN Centroid

slide-139
SLIDE 139

How to improve the ILS

kNN Centroid

FS = 10 FS = 40 FS = 120

slide-140
SLIDE 140

How to improve the ILS

kNN Centroid

FS = 10 FS = 40 FS = 120

slide-141
SLIDE 141

How to improve the ILS

kNN Centroid

FS = 10 FS = 40 FS = 120 Weight = 17 Weight = 4 Weight = 1

slide-142
SLIDE 142

How to improve the ILS

kNN Centroid

FS = 10 FS = 40 FS = 120 Weight = 12 Weight = 3 Weight = 1

slide-143
SLIDE 143

How to improve the ILS

kNN Centroid

FS = 10 FS = 40 FS = 120 Weight = 144 Weight = 9 Weight = 1

slide-144
SLIDE 144

How to improve the ILS

kNN Centroid

FS = 10 FS = 40 FS = 120

slide-145
SLIDE 145

How to improve the ILS

  • Use knowledge on Signal Propagation to develop better distance metrics

10 dBm of difference in both cases Very different meaning!!!

  • 45
  • 80

X

  • 55
  • 80

X

  • 55
  • 90

X vs

slide-146
SLIDE 146

How to improve the ILS

  • Use knowledge on Signal Propagation to develop better distance metrics

Perfect match in both cases The second case is less representative!

  • 50
  • 40
  • 60
  • 50
  • 40
  • 60

N/A

  • 55

N/A N/A

  • 55

N/A vs vs

slide-147
SLIDE 147

How to improve the ILS

  • Use knowledge on Signal Propagation to develop better distance metrics

Continuity

problems!

  • 50
  • 45
  • 70

N/A

  • 50
  • 70

N/A

  • 45

N/A vs

  • 50
  • 45

N/A

  • 55
  • 45
  • 60
  • 50
  • 40

N/A

slide-148
SLIDE 148

Part 8: Time to improve the ILS

slide-149
SLIDE 149

The best configuration is:

slide-150
SLIDE 150

Competition rules (second phase)

  • You are allowed to modified the function:

○ ApplyKnn ○ ChangeDataRepresentation

  • But, you can not modified the function:

○ TestKnn

slide-151
SLIDE 151

Part 9: Awards

slide-152
SLIDE 152

The “Best training award” goes to:

slide-153
SLIDE 153

The “Best ILS award” goes to: