SlideShare a Scribd company logo
1 of 24
Download to read offline
Global Map Matching using BLE
Beacons for Indoor Route and
Stay Estimation
Daisuke Yamamoto*, Ryosuke Tanaka, Shinsuke Kajioka,
Hiroshi Matsuo and Naohisa Takahashi.
Nagoya Institute of Technology, Japan.
Background (1/2)
• Location based services
• GPS, BLE(Bluetooth Low Energy), Web Map Service
• Services using BLE beacons to estimate the location of users
• A lot of smartphones and devices equipped with BLE function.
• Compact, battery-drive (over 5 years), inexpensive, easy to install,
and can communicate with smartphones.
• Location can be estimated by using RSSI (received signal strength indication)
• RSSI is correlated to the distance between the transmitter and the receiver
2Nagoya Insititue of Technology
Background (2/2)
• BLE beacon-based Attendance
Management System
• 1600 beacons in all classrooms
and corridors.
• This system allows students to
report attendance through
a smartphone application by
accepting the BLE beacons.
• In the classroom, 4 to 8 beacons
are installed.
3
BLE beacon installation example in our university
(Bld. 52, floor 2)
Nagoya Insititue of Technology
Motivation
• Realize a highly accurate indoor route / stay estimation of users
using BLE beacons.
• Route Estimation : estimating a travel route in a path network
• Stay Estimation : estimating the place and the time (start and end)
staying at the node in a path network.
4Nagoya Insititue of Technology
Related work
• Indoor location estimation using BLE beacon
• Proximity method, Particle filtering method, Fingerprint method, etc…
• Outdoor route estimation using GPS
• Global map matching
• Road matching is a method of estimating a travel route in a road network
• There is little research on “route” estimation using ”BLE beacons”.
• We believe that it is important to estimate the route as well as its location.
5Nagoya Insititue of Technology
Related work
- Global map matching (for GPS)
• ST-matching[1]
• Global map matching – Find the global optimum route from the GPS data.
• High accuracy even when the GPS reception interval is rough.
• Estimates the path using the Viterbi algorithm, considering spatial and
temporal features.
6
GPS points
[1] Y. Lou and et al. Map-matching for low-sampling-rate GPS trajectories. In ACM SIGSPATIAL GIS, 2009.
Candidate points
Estimated route
(Incremental map matching)
Estimated route
(Global map matching)
Road
Nagoya Insititue of Technology
Approach
• We apply the idea of global map matching for GPS to indoor route
estimation based on BLE beacons.
• The global optimum route in beacon network can be estimated from the radio
field intensities of the BLE beacon directly.
7
BLE
beacon
Global Map
Matching
for GPS
Global optimum
Indoor route estimation
High accuracy
Nagoya Insititue of Technology
Problems (1/2)
• Problem 1 – The principles of
BLE and GPS are different.
• GPS log is recorded as latitude and
longitude coordinates.
• BLE beacon log is recorded as its
ID and RSSI.
•  BLE beacon log does not contain
position information.
• Targets and algorithm are also
different.
GPS BLE
Input data
Latitude,
longitude
Beacon ID,
RSSI
Estimation
target
Route Location
Algorithm
Map
matching
Fingerprint
Indoor
/outdoor
outdoor indoor
8Nagoya Insititue of Technology
Problems (2/2)
• Problem 2 – BLE beacons might not be
placed on the nodes of path network.
• Restrictions involving the location where
the BLE beacons are installed
• Multi-pass problem occurs when
the distance from the beacon increases.
• The distance between nodes and beacons should be shorter.
• Approaches
• BST-Matching method to realize global map matching for BLE beacons.
• Based on ST-Matching, which is one of the global map matching for GPS.
• Two layer network model.
9Nagoya Insititue of Technology
Two layer network model
 Path Network (PN) Layer
• Represent the route that users
move
• ( Logical layer )
 Beacon Network (BN) Layer
• Represents the installation
relationship of the BLE beacons
• One beacon has one BN node,
their positions are the same.
• ( Physical layer)
10Nagoya Insititue of Technology
Outline – Route estimation
11
Sequence of
candidate
nodes on BN
BN node
sequence
(route)
PN node
sequence
(route)
PN link
sequence
(route)
Path Network
Beacon Network
Twolayernetwork
Path
network
(PN)
Beacon
Network
(BN)
Conversion
(1st)
BST-
Matching
(core)
Conversion
(2nd)
Route
Interpolation
Beacon
log
1 2
3 4
Route
Nagoya Insititue of Technology
Beacon log
• Beacon logs can be corrected by
user’s smartphone
• Format
• Beacon log
• Sequence of Message
in chronological order
• Beacon Message
• time t
• Set of Beacon data
• Beacon Data
• {BEACON_ID, RSSI}
• RSSI < 0
12
time t1
{BEACON_4, -70}
{BEACON_6, -50}
time t2
{BEACON_2, -80}
{BEACON_4, -60}
{BEACON_6, -55}
time t3
{BEACON_2, -60}
…
Beacon log
Beacon Message
Nagoya Insititue of Technology
❶ Convert to sequence of candidate nodes
13
time t1
{BEACON_18, -70}
{BEACON_19, -60}
{BEACON_20, -50}
time t2
{BEACON_18, -80}
{BEACON_20, -60}
{BEACON_21, -55}
…
• Filtering by the threshold R
• delete if RSSI < R
• Convert BEACON_ID to NODE_ID
• Because the position of the
beacon and its BN node are
the same, it is simple and
non-deteriorating conversion.
• Extract top k RSSI for each time
Sequence of candidate nodes
for BST-matching
Candidate nodes:A set of
{NODE_id, RSSI}(k=3)
Beacon log
Apply for each beacon message
k NODE k on the beacon network
20
19
18
21
20
15
12
18
time t1 time t2 time tn
❷ BST-matching
• Find the optimal path with the highest score from all combinations of
candidate nodes.
• Viterbi algorithm
• N(c): score of node c, F(c1c2): score of link c1c2
14
C11
C12
C13
C21
C22
C31
C32
C33
Cn1
Cn2
n3
time t1 time t2 time t3 time tn
F(C11C21)
N c1
1
N c1
2
N c1
3
Optimal path
❷ BST-matching – Scoring
• Spatial Score
• Represent the proximity between the candidate node and the BLE beacon
• 𝑁 𝑐 =
1
10
1
20(RSSImax−𝐶_𝑅𝑆𝑆𝐼)
• Temporal Score
• Indicates whether the traveling speed on the route is valid.
• Ft(c1c2) = 1 (if valid)
less than 1 (if not valid)
• Integrated Score
• F(c1c2) = N(c2) Ft(c1c2)
15
Detailed function of Ft(c1c2)
Nagoya Insititue of Technology
❸&❹ Convert and Interpolation
• Conversion (BN  PN)
• It converts to the route on the path network, referring to the relation between the
beacon network and the path network.
• Route Interpolation
• The PN node sequence may lack some nodes on the route.
• So that we connect these nodes using the Dijkstra’s method.
16
BN node
sequence
PN node
sequence
PN link
sequence
BST-
Matching
(core)
Conversion
(2nd)
Route
Interpolation
2 3 4
Route
Nagoya Insititue of Technology
Stay Estimation
• Stay estimation
• Estimate the place where a user stayed, and its time.
• When the PN node x is estimated continuously for T seconds or more, it is
determined that the user stayed in x.
17
BN node
sequence
PN node
sequence
PN link
sequence
BST-
Matching
Conversion
(2nd)
Route
Interpolation
(Dijkstra)
2 3
Route
Stay
estimation
Stay
4
Stay node
time
Nagoya Insititue of Technology
Experiment – Route estimation
• Condition
• Method
• BST-matching (global map matching, proposed method)
• Inc-matching ( Incremental map matching )
• Simplified method of the BST-matching
• Experimental data – Beacon log
• 3 routes (route 1, route 2, route 3)
• Correct 10 beacon logs for each route using smartphone.
• 3 seconds interval.
• Evaluation
• Compare the estimated route with the correct route.
Route 1
Route 3
Floor 2
Floor 3
Floor 2
18Nagoya Insititue of Technology
Experiment – Inc-matching
• One of the Incremental map
matching
• Simplified method of the BST.
• Use only 1 candidate node for
each timestamp.
• As a result, the route can be
incrementally estimated
• Other conditions are the same
as the BST-matching
• Two layer network
• Conversion, Interpolation
BST-matching Inc-Matching
Optimum Global Local
Input data
Beacon ID,
RSSI
Beacon ID,
RSSI
Candidate
nodes
greater than 1 1
Network
Two layer
network
Two layer
network
Algorithm
Viterbi
algorithm
---
Conversion,
Interpolation
〇 〇
19Nagoya Insititue of Technology
Result – Route estimation
• Result ( BST vs Inc )
• Both BST and Inc are high accuracy
• F-measure is 0.95 in the best threshold.
• BST is robust against the threshold.
• BST is suitable to practical use
• it is not easy to find the best threshold.
• RSSI tends to change according to
environments
• BST is not only high accuracy but also
robust against the threshold.
20
0.82
0.84
0.86
0.88
0.9
0.92
0.94
0.96
0.98
F-measure
Threshold (corridor, classroom)
BST
Inc
* *
*
*
*
* P=0.05
Nagoya Insititue of Technology
Result – Stay estimation
• Evaluation metrics
• Compare the estimated time and the correct time for specific node.
• Result
• BST-Matching was up to 29% better than the incremental map matching.
• The lower the threshold, the smaller the error.
21
Method Threshold Stay estimation error (sec) Standard deviation
BST-matching
Low 15.679 9.0095
High 21.239 9.7209
Inc-matching
Low 27.317 12.622
High 26.925 11.448
Nagoya Insititue of Technology
Result – Processing time
• Result
• In incremental map matching, the processing time is fast.
• BST-Matching takes time because it is necessary to calculate the optimal
solution between the candidate nodes.
22
Method Threshold Processing time in Route 3 (sec)
BST-matching
Low 9777.6
High 2554.0
Inc-matching
Low 19.8
High 103.6
Nagoya Insititue of Technology
Conclusion
• Global map matching using BLE beacons
• For Indoor route and stay estimation.
• Very good accuracy ( F-measure is 0.95 ) of route estimation.
• Robust against the threshold.
• Future work
• Solve the processing cost problem
• Route estimation for both indoor and outdoor environment
• Using both BLE beacon and GPS
• Evaluation based on massive data.
23Nagoya Insititue of Technology
Thank you for your attention.
Any questions? Please speak slowly 
Nagoya Insititue of Technology 24

More Related Content

What's hot

Adaptive analog beamforming
Adaptive analog beamformingAdaptive analog beamforming
Adaptive analog beamformingKhalid Hussain
 
A Review on Comparison of the Geographic Routing Protocols in MANET
A Review on Comparison of the Geographic Routing Protocols in MANETA Review on Comparison of the Geographic Routing Protocols in MANET
A Review on Comparison of the Geographic Routing Protocols in MANETEditor IJCATR
 
Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...
Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...
Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...IOSR Journals
 
Prediction and planning for self driving at waymo
Prediction and planning for self driving at waymoPrediction and planning for self driving at waymo
Prediction and planning for self driving at waymoYu Huang
 
Higher Order Feature Set For Underwater Noise Classification
Higher Order Feature Set For Underwater Noise ClassificationHigher Order Feature Set For Underwater Noise Classification
Higher Order Feature Set For Underwater Noise ClassificationCSCJournals
 
Remote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 ProjectsRemote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 ProjectsVijay Karan
 
Remote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 ProjectsRemote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 ProjectsVijay Karan
 
Area, Delay and Power Comparison of Adder Topologies
Area, Delay and Power Comparison of Adder TopologiesArea, Delay and Power Comparison of Adder Topologies
Area, Delay and Power Comparison of Adder TopologiesVLSICS Design
 
TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013Clay Stanek
 
MIMO System Performance Evaluation for High Data Rate Wireless Networks usin...
MIMO System Performance Evaluation for High Data Rate  Wireless Networks usin...MIMO System Performance Evaluation for High Data Rate  Wireless Networks usin...
MIMO System Performance Evaluation for High Data Rate Wireless Networks usin...IJMER
 
PVANet - PR033
PVANet - PR033PVANet - PR033
PVANet - PR033Jinwon Lee
 
Mimo detection-bp-p mrf-yoon-160706a
Mimo detection-bp-p mrf-yoon-160706aMimo detection-bp-p mrf-yoon-160706a
Mimo detection-bp-p mrf-yoon-160706aSeokhyun Yoon
 
MLS An Efficient Location Service for Mobile Ad Hoc Networks
MLS An Efficient Location Service for Mobile Ad Hoc NetworksMLS An Efficient Location Service for Mobile Ad Hoc Networks
MLS An Efficient Location Service for Mobile Ad Hoc Networkskga185
 
2 - Generation of PSK signal using non linear devices via MATLAB (presented i...
2 - Generation of PSK signal using non linear devices via MATLAB (presented i...2 - Generation of PSK signal using non linear devices via MATLAB (presented i...
2 - Generation of PSK signal using non linear devices via MATLAB (presented i...Youness Lahdili
 
MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...
MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...
MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...VLSICS Design
 
Outdoor path loss models for ieee 802.16
Outdoor path loss models for ieee 802.16Outdoor path loss models for ieee 802.16
Outdoor path loss models for ieee 802.16Nguyen Minh Thu
 
Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...
Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...
Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...Alex Klibisz
 

What's hot (18)

Adaptive analog beamforming
Adaptive analog beamformingAdaptive analog beamforming
Adaptive analog beamforming
 
A Review on Comparison of the Geographic Routing Protocols in MANET
A Review on Comparison of the Geographic Routing Protocols in MANETA Review on Comparison of the Geographic Routing Protocols in MANET
A Review on Comparison of the Geographic Routing Protocols in MANET
 
Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...
Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...
Enhanced Mobile Node Tracking With Received Signal Strength in Wireless Senso...
 
Prediction and planning for self driving at waymo
Prediction and planning for self driving at waymoPrediction and planning for self driving at waymo
Prediction and planning for self driving at waymo
 
Higher Order Feature Set For Underwater Noise Classification
Higher Order Feature Set For Underwater Noise ClassificationHigher Order Feature Set For Underwater Noise Classification
Higher Order Feature Set For Underwater Noise Classification
 
Remote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 ProjectsRemote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 Projects
 
Remote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 ProjectsRemote Sensing IEEE 2015 Projects
Remote Sensing IEEE 2015 Projects
 
Area, Delay and Power Comparison of Adder Topologies
Area, Delay and Power Comparison of Adder TopologiesArea, Delay and Power Comparison of Adder Topologies
Area, Delay and Power Comparison of Adder Topologies
 
TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013TurnerBottoneStanekNIPS2013
TurnerBottoneStanekNIPS2013
 
MIMO System Performance Evaluation for High Data Rate Wireless Networks usin...
MIMO System Performance Evaluation for High Data Rate  Wireless Networks usin...MIMO System Performance Evaluation for High Data Rate  Wireless Networks usin...
MIMO System Performance Evaluation for High Data Rate Wireless Networks usin...
 
PVANet - PR033
PVANet - PR033PVANet - PR033
PVANet - PR033
 
Mimo detection-bp-p mrf-yoon-160706a
Mimo detection-bp-p mrf-yoon-160706aMimo detection-bp-p mrf-yoon-160706a
Mimo detection-bp-p mrf-yoon-160706a
 
D017522833
D017522833D017522833
D017522833
 
MLS An Efficient Location Service for Mobile Ad Hoc Networks
MLS An Efficient Location Service for Mobile Ad Hoc NetworksMLS An Efficient Location Service for Mobile Ad Hoc Networks
MLS An Efficient Location Service for Mobile Ad Hoc Networks
 
2 - Generation of PSK signal using non linear devices via MATLAB (presented i...
2 - Generation of PSK signal using non linear devices via MATLAB (presented i...2 - Generation of PSK signal using non linear devices via MATLAB (presented i...
2 - Generation of PSK signal using non linear devices via MATLAB (presented i...
 
MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...
MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...
MINIMIZATION OF HANDOFF LATENCY BY CO-ORDINATE EVALUATION METHOD USING GPS BA...
 
Outdoor path loss models for ieee 802.16
Outdoor path loss models for ieee 802.16Outdoor path loss models for ieee 802.16
Outdoor path loss models for ieee 802.16
 
Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...
Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...
Research Summary: Scalable Algorithms for Nearest-Neighbor Joins on Big Traje...
 

Similar to Global Map Matching using BLE Beacons for Indoor Route and Stay Estimation

Positioning techniques in 3 g networks (1)
Positioning techniques in 3 g networks (1)Positioning techniques in 3 g networks (1)
Positioning techniques in 3 g networks (1)kike2005
 
Review On Different Feature Extraction Algorithms
Review On Different Feature Extraction AlgorithmsReview On Different Feature Extraction Algorithms
Review On Different Feature Extraction AlgorithmsIRJET Journal
 
Dynamic time warping and PIC 16F676 for control of devices
Dynamic time warping and PIC 16F676 for control of devicesDynamic time warping and PIC 16F676 for control of devices
Dynamic time warping and PIC 16F676 for control of devicesRoger Gomes
 
High Efficiency Video Codec
High Efficiency Video CodecHigh Efficiency Video Codec
High Efficiency Video CodecTejus Adiga M
 
Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...
Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...
Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...Anxo Tato Arias
 
VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...
VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...
VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...IRJET Journal
 
Design and analysis of dual-mode numerically controlled oscillators based co...
Design and analysis of dual-mode numerically controlled  oscillators based co...Design and analysis of dual-mode numerically controlled  oscillators based co...
Design and analysis of dual-mode numerically controlled oscillators based co...IJECEIAES
 
Data quality evaluation &amp; orbit identification from scatterometer
Data quality evaluation &amp; orbit identification from scatterometerData quality evaluation &amp; orbit identification from scatterometer
Data quality evaluation &amp; orbit identification from scatterometerMudit Dholakia
 
Analyzing the performance of the dynamic
Analyzing the performance of the dynamicAnalyzing the performance of the dynamic
Analyzing the performance of the dynamicIJCNCJournal
 
5G physical layer
5G physical layer 5G physical layer
5G physical layer Ali Nikfal
 
Video Annotation for Visual Tracking via Selection and Refinement_tran.pptx
Video Annotation for Visual Tracking via Selection and Refinement_tran.pptxVideo Annotation for Visual Tracking via Selection and Refinement_tran.pptx
Video Annotation for Visual Tracking via Selection and Refinement_tran.pptxAlyaaMachi
 
Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...
Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...
Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...TSC University of Mondragon
 
IRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo Method
IRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo MethodIRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo Method
IRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo MethodIRJET Journal
 
Digital signal transmission in ofc
Digital signal transmission in ofcDigital signal transmission in ofc
Digital signal transmission in ofcAnkith Shetty
 
Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...
Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...
Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...theijes
 
Noise Tolerant and Faster On Chip Communication Using Binoc Model
Noise Tolerant and Faster On Chip Communication Using Binoc ModelNoise Tolerant and Faster On Chip Communication Using Binoc Model
Noise Tolerant and Faster On Chip Communication Using Binoc ModelIJMER
 

Similar to Global Map Matching using BLE Beacons for Indoor Route and Stay Estimation (20)

LTE-A Virtual Drive Testing for Vehicular Environments
LTE-A Virtual Drive Testing for Vehicular Environments LTE-A Virtual Drive Testing for Vehicular Environments
LTE-A Virtual Drive Testing for Vehicular Environments
 
Positioning techniques in 3 g networks (1)
Positioning techniques in 3 g networks (1)Positioning techniques in 3 g networks (1)
Positioning techniques in 3 g networks (1)
 
Review On Different Feature Extraction Algorithms
Review On Different Feature Extraction AlgorithmsReview On Different Feature Extraction Algorithms
Review On Different Feature Extraction Algorithms
 
Dynamic time warping and PIC 16F676 for control of devices
Dynamic time warping and PIC 16F676 for control of devicesDynamic time warping and PIC 16F676 for control of devices
Dynamic time warping and PIC 16F676 for control of devices
 
High Efficiency Video Codec
High Efficiency Video CodecHigh Efficiency Video Codec
High Efficiency Video Codec
 
3G Radio Network Planning
3G Radio Network Planning3G Radio Network Planning
3G Radio Network Planning
 
Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...
Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...
Link Adapatation in Mobile Satellite Links: Field Trial Results using SDR and...
 
VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...
VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...
VLSI Architecture for Cyclostationary Feature Detection Based Spectrum Sensin...
 
Design and analysis of dual-mode numerically controlled oscillators based co...
Design and analysis of dual-mode numerically controlled  oscillators based co...Design and analysis of dual-mode numerically controlled  oscillators based co...
Design and analysis of dual-mode numerically controlled oscillators based co...
 
Network on Chip
Network on ChipNetwork on Chip
Network on Chip
 
Data quality evaluation &amp; orbit identification from scatterometer
Data quality evaluation &amp; orbit identification from scatterometerData quality evaluation &amp; orbit identification from scatterometer
Data quality evaluation &amp; orbit identification from scatterometer
 
Analyzing the performance of the dynamic
Analyzing the performance of the dynamicAnalyzing the performance of the dynamic
Analyzing the performance of the dynamic
 
5G physical layer
5G physical layer 5G physical layer
5G physical layer
 
Video Annotation for Visual Tracking via Selection and Refinement_tran.pptx
Video Annotation for Visual Tracking via Selection and Refinement_tran.pptxVideo Annotation for Visual Tracking via Selection and Refinement_tran.pptx
Video Annotation for Visual Tracking via Selection and Refinement_tran.pptx
 
Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...
Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...
Real-time Implementation of Sphere Decoder-based MIMO Wireless System (EUSIPC...
 
IRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo Method
IRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo MethodIRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo Method
IRJET- Synchronization Scheme of MIMO-OFDM using Monte Carlo Method
 
Digital signal transmission in ofc
Digital signal transmission in ofcDigital signal transmission in ofc
Digital signal transmission in ofc
 
Lambda Data Grid
Lambda Data GridLambda Data Grid
Lambda Data Grid
 
Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...
Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...
Evaluation of STBC and Convolutional Code Performance for Wireless Communicat...
 
Noise Tolerant and Faster On Chip Communication Using Binoc Model
Noise Tolerant and Faster On Chip Communication Using Binoc ModelNoise Tolerant and Faster On Chip Communication Using Binoc Model
Noise Tolerant and Faster On Chip Communication Using Binoc Model
 

Recently uploaded

Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...
Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...
Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...Amil baba
 
ADM100 Running Book for sap basis domain study
ADM100 Running Book for sap basis domain studyADM100 Running Book for sap basis domain study
ADM100 Running Book for sap basis domain studydhruvamdhruvil123
 
Structural Integrity Assessment Standards in Nigeria by Engr Nimot Muili
Structural Integrity Assessment Standards in Nigeria by Engr Nimot MuiliStructural Integrity Assessment Standards in Nigeria by Engr Nimot Muili
Structural Integrity Assessment Standards in Nigeria by Engr Nimot MuiliNimot Muili
 
Curve setting (Basic Mine Surveying)_MI10412MI.pptx
Curve setting (Basic Mine Surveying)_MI10412MI.pptxCurve setting (Basic Mine Surveying)_MI10412MI.pptx
Curve setting (Basic Mine Surveying)_MI10412MI.pptxRomil Mishra
 
Katarzyna Lipka-Sidor - BIM School Course
Katarzyna Lipka-Sidor - BIM School CourseKatarzyna Lipka-Sidor - BIM School Course
Katarzyna Lipka-Sidor - BIM School Coursebim.edu.pl
 
SOFTWARE ESTIMATION COCOMO AND FP CALCULATION
SOFTWARE ESTIMATION COCOMO AND FP CALCULATIONSOFTWARE ESTIMATION COCOMO AND FP CALCULATION
SOFTWARE ESTIMATION COCOMO AND FP CALCULATIONSneha Padhiar
 
Module-1-(Building Acoustics) Noise Control (Unit-3). pdf
Module-1-(Building Acoustics) Noise Control (Unit-3). pdfModule-1-(Building Acoustics) Noise Control (Unit-3). pdf
Module-1-(Building Acoustics) Noise Control (Unit-3). pdfManish Kumar
 
Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...
Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...
Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...shreenathji26
 
High Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMS
High Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMSHigh Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMS
High Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMSsandhya757531
 
Javier_Fernandez_CARS_workshop_presentation.pptx
Javier_Fernandez_CARS_workshop_presentation.pptxJavier_Fernandez_CARS_workshop_presentation.pptx
Javier_Fernandez_CARS_workshop_presentation.pptxJavier Fernández Muñoz
 
Cost estimation approach: FP to COCOMO scenario based question
Cost estimation approach: FP to COCOMO scenario based questionCost estimation approach: FP to COCOMO scenario based question
Cost estimation approach: FP to COCOMO scenario based questionSneha Padhiar
 
STATE TRANSITION DIAGRAM in psoc subject
STATE TRANSITION DIAGRAM in psoc subjectSTATE TRANSITION DIAGRAM in psoc subject
STATE TRANSITION DIAGRAM in psoc subjectGayathriM270621
 
multiple access in wireless communication
multiple access in wireless communicationmultiple access in wireless communication
multiple access in wireless communicationpanditadesh123
 
priority interrupt computer organization
priority interrupt computer organizationpriority interrupt computer organization
priority interrupt computer organizationchnrketan
 
Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...
Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...
Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...Sumanth A
 
Python Programming for basic beginners.pptx
Python Programming for basic beginners.pptxPython Programming for basic beginners.pptx
Python Programming for basic beginners.pptxmohitesoham12
 
March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...
March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...
March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...gerogepatton
 
Robotics Group 10 (Control Schemes) cse.pdf
Robotics Group 10  (Control Schemes) cse.pdfRobotics Group 10  (Control Schemes) cse.pdf
Robotics Group 10 (Control Schemes) cse.pdfsahilsajad201
 

Recently uploaded (20)

Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...
Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...
Uk-NO1 kala jadu karne wale ka contact number kala jadu karne wale baba kala ...
 
ADM100 Running Book for sap basis domain study
ADM100 Running Book for sap basis domain studyADM100 Running Book for sap basis domain study
ADM100 Running Book for sap basis domain study
 
Structural Integrity Assessment Standards in Nigeria by Engr Nimot Muili
Structural Integrity Assessment Standards in Nigeria by Engr Nimot MuiliStructural Integrity Assessment Standards in Nigeria by Engr Nimot Muili
Structural Integrity Assessment Standards in Nigeria by Engr Nimot Muili
 
Curve setting (Basic Mine Surveying)_MI10412MI.pptx
Curve setting (Basic Mine Surveying)_MI10412MI.pptxCurve setting (Basic Mine Surveying)_MI10412MI.pptx
Curve setting (Basic Mine Surveying)_MI10412MI.pptx
 
Katarzyna Lipka-Sidor - BIM School Course
Katarzyna Lipka-Sidor - BIM School CourseKatarzyna Lipka-Sidor - BIM School Course
Katarzyna Lipka-Sidor - BIM School Course
 
SOFTWARE ESTIMATION COCOMO AND FP CALCULATION
SOFTWARE ESTIMATION COCOMO AND FP CALCULATIONSOFTWARE ESTIMATION COCOMO AND FP CALCULATION
SOFTWARE ESTIMATION COCOMO AND FP CALCULATION
 
Module-1-(Building Acoustics) Noise Control (Unit-3). pdf
Module-1-(Building Acoustics) Noise Control (Unit-3). pdfModule-1-(Building Acoustics) Noise Control (Unit-3). pdf
Module-1-(Building Acoustics) Noise Control (Unit-3). pdf
 
Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...
Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...
Introduction to Artificial Intelligence: Intelligent Agents, State Space Sear...
 
High Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMS
High Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMSHigh Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMS
High Voltage Engineering- OVER VOLTAGES IN ELECTRICAL POWER SYSTEMS
 
Versatile Engineering Construction Firms
Versatile Engineering Construction FirmsVersatile Engineering Construction Firms
Versatile Engineering Construction Firms
 
Javier_Fernandez_CARS_workshop_presentation.pptx
Javier_Fernandez_CARS_workshop_presentation.pptxJavier_Fernandez_CARS_workshop_presentation.pptx
Javier_Fernandez_CARS_workshop_presentation.pptx
 
Cost estimation approach: FP to COCOMO scenario based question
Cost estimation approach: FP to COCOMO scenario based questionCost estimation approach: FP to COCOMO scenario based question
Cost estimation approach: FP to COCOMO scenario based question
 
STATE TRANSITION DIAGRAM in psoc subject
STATE TRANSITION DIAGRAM in psoc subjectSTATE TRANSITION DIAGRAM in psoc subject
STATE TRANSITION DIAGRAM in psoc subject
 
multiple access in wireless communication
multiple access in wireless communicationmultiple access in wireless communication
multiple access in wireless communication
 
ASME-B31.4-2019-estandar para diseño de ductos
ASME-B31.4-2019-estandar para diseño de ductosASME-B31.4-2019-estandar para diseño de ductos
ASME-B31.4-2019-estandar para diseño de ductos
 
priority interrupt computer organization
priority interrupt computer organizationpriority interrupt computer organization
priority interrupt computer organization
 
Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...
Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...
Robotics-Asimov's Laws, Mechanical Subsystems, Robot Kinematics, Robot Dynami...
 
Python Programming for basic beginners.pptx
Python Programming for basic beginners.pptxPython Programming for basic beginners.pptx
Python Programming for basic beginners.pptx
 
March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...
March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...
March 2024 - Top 10 Read Articles in Artificial Intelligence and Applications...
 
Robotics Group 10 (Control Schemes) cse.pdf
Robotics Group 10  (Control Schemes) cse.pdfRobotics Group 10  (Control Schemes) cse.pdf
Robotics Group 10 (Control Schemes) cse.pdf
 

Global Map Matching using BLE Beacons for Indoor Route and Stay Estimation

  • 1. Global Map Matching using BLE Beacons for Indoor Route and Stay Estimation Daisuke Yamamoto*, Ryosuke Tanaka, Shinsuke Kajioka, Hiroshi Matsuo and Naohisa Takahashi. Nagoya Institute of Technology, Japan.
  • 2. Background (1/2) • Location based services • GPS, BLE(Bluetooth Low Energy), Web Map Service • Services using BLE beacons to estimate the location of users • A lot of smartphones and devices equipped with BLE function. • Compact, battery-drive (over 5 years), inexpensive, easy to install, and can communicate with smartphones. • Location can be estimated by using RSSI (received signal strength indication) • RSSI is correlated to the distance between the transmitter and the receiver 2Nagoya Insititue of Technology
  • 3. Background (2/2) • BLE beacon-based Attendance Management System • 1600 beacons in all classrooms and corridors. • This system allows students to report attendance through a smartphone application by accepting the BLE beacons. • In the classroom, 4 to 8 beacons are installed. 3 BLE beacon installation example in our university (Bld. 52, floor 2) Nagoya Insititue of Technology
  • 4. Motivation • Realize a highly accurate indoor route / stay estimation of users using BLE beacons. • Route Estimation : estimating a travel route in a path network • Stay Estimation : estimating the place and the time (start and end) staying at the node in a path network. 4Nagoya Insititue of Technology
  • 5. Related work • Indoor location estimation using BLE beacon • Proximity method, Particle filtering method, Fingerprint method, etc… • Outdoor route estimation using GPS • Global map matching • Road matching is a method of estimating a travel route in a road network • There is little research on “route” estimation using ”BLE beacons”. • We believe that it is important to estimate the route as well as its location. 5Nagoya Insititue of Technology
  • 6. Related work - Global map matching (for GPS) • ST-matching[1] • Global map matching – Find the global optimum route from the GPS data. • High accuracy even when the GPS reception interval is rough. • Estimates the path using the Viterbi algorithm, considering spatial and temporal features. 6 GPS points [1] Y. Lou and et al. Map-matching for low-sampling-rate GPS trajectories. In ACM SIGSPATIAL GIS, 2009. Candidate points Estimated route (Incremental map matching) Estimated route (Global map matching) Road Nagoya Insititue of Technology
  • 7. Approach • We apply the idea of global map matching for GPS to indoor route estimation based on BLE beacons. • The global optimum route in beacon network can be estimated from the radio field intensities of the BLE beacon directly. 7 BLE beacon Global Map Matching for GPS Global optimum Indoor route estimation High accuracy Nagoya Insititue of Technology
  • 8. Problems (1/2) • Problem 1 – The principles of BLE and GPS are different. • GPS log is recorded as latitude and longitude coordinates. • BLE beacon log is recorded as its ID and RSSI. •  BLE beacon log does not contain position information. • Targets and algorithm are also different. GPS BLE Input data Latitude, longitude Beacon ID, RSSI Estimation target Route Location Algorithm Map matching Fingerprint Indoor /outdoor outdoor indoor 8Nagoya Insititue of Technology
  • 9. Problems (2/2) • Problem 2 – BLE beacons might not be placed on the nodes of path network. • Restrictions involving the location where the BLE beacons are installed • Multi-pass problem occurs when the distance from the beacon increases. • The distance between nodes and beacons should be shorter. • Approaches • BST-Matching method to realize global map matching for BLE beacons. • Based on ST-Matching, which is one of the global map matching for GPS. • Two layer network model. 9Nagoya Insititue of Technology
  • 10. Two layer network model  Path Network (PN) Layer • Represent the route that users move • ( Logical layer )  Beacon Network (BN) Layer • Represents the installation relationship of the BLE beacons • One beacon has one BN node, their positions are the same. • ( Physical layer) 10Nagoya Insititue of Technology
  • 11. Outline – Route estimation 11 Sequence of candidate nodes on BN BN node sequence (route) PN node sequence (route) PN link sequence (route) Path Network Beacon Network Twolayernetwork Path network (PN) Beacon Network (BN) Conversion (1st) BST- Matching (core) Conversion (2nd) Route Interpolation Beacon log 1 2 3 4 Route Nagoya Insititue of Technology
  • 12. Beacon log • Beacon logs can be corrected by user’s smartphone • Format • Beacon log • Sequence of Message in chronological order • Beacon Message • time t • Set of Beacon data • Beacon Data • {BEACON_ID, RSSI} • RSSI < 0 12 time t1 {BEACON_4, -70} {BEACON_6, -50} time t2 {BEACON_2, -80} {BEACON_4, -60} {BEACON_6, -55} time t3 {BEACON_2, -60} … Beacon log Beacon Message Nagoya Insititue of Technology
  • 13. ❶ Convert to sequence of candidate nodes 13 time t1 {BEACON_18, -70} {BEACON_19, -60} {BEACON_20, -50} time t2 {BEACON_18, -80} {BEACON_20, -60} {BEACON_21, -55} … • Filtering by the threshold R • delete if RSSI < R • Convert BEACON_ID to NODE_ID • Because the position of the beacon and its BN node are the same, it is simple and non-deteriorating conversion. • Extract top k RSSI for each time Sequence of candidate nodes for BST-matching Candidate nodes:A set of {NODE_id, RSSI}(k=3) Beacon log Apply for each beacon message k NODE k on the beacon network 20 19 18 21 20 15 12 18 time t1 time t2 time tn
  • 14. ❷ BST-matching • Find the optimal path with the highest score from all combinations of candidate nodes. • Viterbi algorithm • N(c): score of node c, F(c1c2): score of link c1c2 14 C11 C12 C13 C21 C22 C31 C32 C33 Cn1 Cn2 n3 time t1 time t2 time t3 time tn F(C11C21) N c1 1 N c1 2 N c1 3 Optimal path
  • 15. ❷ BST-matching – Scoring • Spatial Score • Represent the proximity between the candidate node and the BLE beacon • 𝑁 𝑐 = 1 10 1 20(RSSImax−𝐶_𝑅𝑆𝑆𝐼) • Temporal Score • Indicates whether the traveling speed on the route is valid. • Ft(c1c2) = 1 (if valid) less than 1 (if not valid) • Integrated Score • F(c1c2) = N(c2) Ft(c1c2) 15 Detailed function of Ft(c1c2) Nagoya Insititue of Technology
  • 16. ❸&❹ Convert and Interpolation • Conversion (BN  PN) • It converts to the route on the path network, referring to the relation between the beacon network and the path network. • Route Interpolation • The PN node sequence may lack some nodes on the route. • So that we connect these nodes using the Dijkstra’s method. 16 BN node sequence PN node sequence PN link sequence BST- Matching (core) Conversion (2nd) Route Interpolation 2 3 4 Route Nagoya Insititue of Technology
  • 17. Stay Estimation • Stay estimation • Estimate the place where a user stayed, and its time. • When the PN node x is estimated continuously for T seconds or more, it is determined that the user stayed in x. 17 BN node sequence PN node sequence PN link sequence BST- Matching Conversion (2nd) Route Interpolation (Dijkstra) 2 3 Route Stay estimation Stay 4 Stay node time Nagoya Insititue of Technology
  • 18. Experiment – Route estimation • Condition • Method • BST-matching (global map matching, proposed method) • Inc-matching ( Incremental map matching ) • Simplified method of the BST-matching • Experimental data – Beacon log • 3 routes (route 1, route 2, route 3) • Correct 10 beacon logs for each route using smartphone. • 3 seconds interval. • Evaluation • Compare the estimated route with the correct route. Route 1 Route 3 Floor 2 Floor 3 Floor 2 18Nagoya Insititue of Technology
  • 19. Experiment – Inc-matching • One of the Incremental map matching • Simplified method of the BST. • Use only 1 candidate node for each timestamp. • As a result, the route can be incrementally estimated • Other conditions are the same as the BST-matching • Two layer network • Conversion, Interpolation BST-matching Inc-Matching Optimum Global Local Input data Beacon ID, RSSI Beacon ID, RSSI Candidate nodes greater than 1 1 Network Two layer network Two layer network Algorithm Viterbi algorithm --- Conversion, Interpolation 〇 〇 19Nagoya Insititue of Technology
  • 20. Result – Route estimation • Result ( BST vs Inc ) • Both BST and Inc are high accuracy • F-measure is 0.95 in the best threshold. • BST is robust against the threshold. • BST is suitable to practical use • it is not easy to find the best threshold. • RSSI tends to change according to environments • BST is not only high accuracy but also robust against the threshold. 20 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 F-measure Threshold (corridor, classroom) BST Inc * * * * * * P=0.05 Nagoya Insititue of Technology
  • 21. Result – Stay estimation • Evaluation metrics • Compare the estimated time and the correct time for specific node. • Result • BST-Matching was up to 29% better than the incremental map matching. • The lower the threshold, the smaller the error. 21 Method Threshold Stay estimation error (sec) Standard deviation BST-matching Low 15.679 9.0095 High 21.239 9.7209 Inc-matching Low 27.317 12.622 High 26.925 11.448 Nagoya Insititue of Technology
  • 22. Result – Processing time • Result • In incremental map matching, the processing time is fast. • BST-Matching takes time because it is necessary to calculate the optimal solution between the candidate nodes. 22 Method Threshold Processing time in Route 3 (sec) BST-matching Low 9777.6 High 2554.0 Inc-matching Low 19.8 High 103.6 Nagoya Insititue of Technology
  • 23. Conclusion • Global map matching using BLE beacons • For Indoor route and stay estimation. • Very good accuracy ( F-measure is 0.95 ) of route estimation. • Robust against the threshold. • Future work • Solve the processing cost problem • Route estimation for both indoor and outdoor environment • Using both BLE beacon and GPS • Evaluation based on massive data. 23Nagoya Insititue of Technology
  • 24. Thank you for your attention. Any questions? Please speak slowly  Nagoya Insititue of Technology 24

Editor's Notes

  1. Good morning, everyone. My name is Daisuke Yamamoto. I’m from Nagoya institute of technology japan Today I’d like to talk about our research entitled “Global Map Matching using BLE Beacons for Indoor Route and Stay Estimation”.
  2. In recent years, there are many smartphones equipped with BLE function. And, many services using BLE beacons to estimate the location of users have been proposed. BLE beacon is compact, 5 years battery life, inexpensive, easy to install, and can communicate with smartphones. Since RSSI is correlated to the distance between the transmitter and the receiver, Location can be estimated by using RSSI. Because of such features, BLE beacons are widely used for indoor location estimation studies.
  3. We have also developed an attendance management system using 1600 BLE beacons in all classrooms and corridors of our university. This system allows students to report attendance through a smartphone application by accepting the signal of the BLE beacons. 4 to 8 beacons are installed in a classroom. This figure shows the example of BLE beacon’s installation map in our university. BLE beacons are set on the wall like this.
  4. The purpose of this study is to realize a highly accurate indoor route / stay estimation of users using BLE beacons. Here, route estimation means the estimation of a travel route in a path network composed of links. Stay estimation means the estimation of the place and the time staying at the node in a path network.
  5. There are many researches to estimate indoor location using BLE beacons, such as proximity method, particle filtering method, and fingerprint method. On the other hand, in the field of GPS, many studies on road matching have been proposed. Road matching is a method of estimating a travel route in a road network based on the GPS. There are also many researches including global map matching. However, unlike GPS, there is little research to estimate the route using BLE beacons. we believe that it is important to estimate the route as well as its location.
  6. Global map matching is a method of finding the global optimum route from the GPS data. ST-Matching is one of the global map matching methods for GPS. ST-Matching can perform map-matching with high accuracy even when the GPS interval is rough. This method estimates the path using the Viterbi algorithm, considering spatial and temporal features. First, we demonstrate the example of incremental map matching. In this example, one candidate point is estimated from one GPS point. We can get the route by connecting these candidate points. Next, we demonstrate the example of global map matching. In this example, some candidate points are estimated from one GPS point. We can get the global optimum route using the Viterbi algorithm.
  7. We apply the idea of global map matching for GPS, to indoor route estimation based on BLE beacons. Thus, the global optimum route can be estimated from the BLE beacon directly. We expect that proposed method can perform map-matching with high accuracy for BLE beacons as well as GPS.
  8. To realize this approach, we have the following problems. First problem is that it is difficult to apply the global map matching method for GPS directly to the BLE beacon. Because the structure of BLE beacon and GPS are different. For example, a GPS log is recorded as a sequence of coordinates, but a BLE beacon log is recorded as a sequence of beacon ID and RSSI. In other words, the BLE beacon log does not contain positional information.
  9. Second problem is that BLE beacons might not be placed directly above the nodes of the path network. For example, as shown in this Figure, the position of the BLE beacon and the node on the path network may be far away each other. Multi-pass problem occurs when the distance from the beacon increases. So that the distance between nodes and beacons should be shorter. Our approaches are here. First, we propose a BST-Matching method to realize global map matching for BLE beacons. This method is a novel technique based on ST-Matching. Next, we propose a two-layer network model.
  10. This figure shows the two layer network. This network is consisted from the path network layer and the beacon network layer. The path network is a graph that shows the route in which users actually move. It is worked as a logical layer. A beacon network is a graph that shows the beacon placement relations. One beacon has one BN node, and their positions are the same. It is worked as a physical layer.
  11. An outline of the proposed method is shown in Here. The input data is beacon log. In the first stage, the sequence of candidate nodes can be converted from the beacon log. In the second stage, BST-matching estimates the BN node sequence. In the third stage, it converts the BN nodes to the PN node. In the last stage, we can get the route after applying the route interpolation.
  12. Let’s talk about the beacon log. The beacon logs can be corrected by user’s smartphone. The beacon log is defined as a sequence of beacon messages in chronological order. The beacon message contains the timestamp of the beacon and 0 or more beacon data recognized at that time. The beacon data is consisted from beacon_ID and RSSI.
  13. In the first stage, we state how to convert the beacon log into the candidate nodes for the BST-Matching. It applies the following method for each message. 1. First, the beacon data is filtered by the threshold R. Here, the threshold in the classroom can be different from the threshold in the corridor. 2. Next, convert BEACON_ID to NODE_ID referring the beacon network. Because the position of the beacon and its BN node are the same, it is simple conversion. 3. Then, extract top K RSSI for each time. As a result, candidate nodes can be generated for each time.
  14. In the second stage, We propose the BST-Matching algorithm. This algorithm finds the optimal path with the highest score from all combinations of candidate nodes based on the Viterbi algorithm. Each node has score N(c), and each link has score F(c1c2) In this figure, this algorithm can find the optimal red path from the blue lines. When red line is optimal path, C11, C21, C33, Cn1 nodes are the elements of estimated routes.
  15. Next, we define the scoring functions for BST-matching. It has two type scoring functions, one is spatial score, the other is temporal score. Spatial score represents the proximity between the candidate node and the BLE beacon. This function can convert a RSSI value to a distance. Temporal score indicates whether the traveling speed on the route is valid. If the speed is valid, the function outputs 1. If the speed is not valid, the function outputs less than 1. Last is the integrated score. The integrated score is simply multiplied spatial score by temporal score.
  16. BST-Matching generates a BN node sequence on the beacon network. However, the beacon network is different from the path network. So, In the third stage, we convert the BN node sequence to the PN node sequence, referring the relationship between the beacon network and the path network. In the last stage, The PN node sequence may lack some nodes on the route. So that we connect these nodes using the Dijkstra’s method.
  17. Stay estimation is a function for estimating the place and the time where a user stayed. The process of the stay estimation is the same as the route estimation except for the last step. When the PN node x is estimated continuously for T seconds or more, it is determined that the user stayed in x.
  18. We evaluate the proposed method. The experimental conditions are here. We compare the following methods: first is BST-matching, the other is Inc-matching. BST-matching is the proposed method. Inc-matching is simplified method of the BST-matching, and is worked as incremental map matching. The walking route used in the experiment is Route 1, 2, 3, as shown in Here. We collected 10 beacon logs for each route using a smartphone. We evaluate the proposed method by comparing the estimated route with the correct route.
  19. Inc-matching is simplified method of the BST-matching. This method uses only 1 candidate node for each timestamp. As a result, the route can be incrementally estimated. Other conditions such as conversion and interpolation are the same as BST-Matching.
  20. Next we talk about the route estimation result. First, lets compare the BST-matching with the Inc-matching. As shown in this figure, both BST-matching and Inc-matching are high accuracy. Both F-measures are 0.95 in the best threshold. There was no difference in F-measure between BST-matching and Inc-matching at the best thresholds. However, in the area where the threshold is low, the F-measure of BST-Matching is improved. This result suggests that BST-Matching is less sensitive to threshold. In general, the RSSI of the BLE beacon is easily varied because of the battery level or the crowd. So that it is difficult to set an optimum threshold value in a real environment. Therefore, the proposed method is an excellent method, especially in the real environment.
  21. Next, we state the result of the stay estimation. We compare the estimated time and the correct time for specific nodes. The BST-Matching was up to 29% better than the inc matching. These results suggest that the BST-method is also better for stay estimation.
  22. Next, we state the processing time of each method. In incremental map matching, the processing time is fast, while BST-Matching takes time. Because it is necessary to calculate the optimal path between the candidate nodes. We have to solve the processing cost problem.
  23. Conclusion. we propose a global map matching method using BLE beacons for indoor route and stay estimation. The evaluation experiment shows that the proposed method has very good F-measure of route estimation. The proposed method will be suitable for practical use that it is not easily influenced by the threshold value. In future work, We have to solve the processing cost problem. And we want to propose the route estimation method for both indoor and outdoor environment Since the evaluation of our study is limited, we will evaluate based on massive data.