Sensor Networks and Software Platform Development for Seamless Traffic Monitoring A. Marousis, T. Economopoulos, A. Panagopoulos School of Electrical and Computer Engineering National Technical University of Athens, Greece Efficient Management of PARking under COnstraints 7July 2015
Outline Identification of the target environments Sensor selection, system assembly, system programming Dimensioning of the environment Algorithms for position calculation and kinematic characteristics Case Studies Test scenarios Collection of raw data and quality control Database development and real time feeding Web based applications For PCs For Smart phones Efficient Management of PARking under COnstraints July 7, 2015 2
Target Environments + Add ons Outdoor parking environments Indoor parking environments User Requirements Small Velocities High density Various characteristics o Metallic structures o Building blocks o Lights o Multilevel structures Seamless monitoring Increased accuracy in position/speed/direction Systems that does not rely on extra user effort. Efficient Management of PARking under COnstraints July 7, 2015 3
Sensor Selection, System Assembly, System Programming Sensors the are capable of monitoring the radio communication devices (without direct user involvement) like smart phones, tablets, Bluetooth devices, wireless earphones, wireless audio cellular car systems, etc. The idea is based on the identification of the characteristic beacons Read out of all the public information that is capable of accurate positioning Periodic monitoring Different types of communication protocols have to be identified Grouping of the devices Efficient Management of PARking under COnstraints July 7, 2015 4
Sensor Selection, System Assembly, System Programming Sensors the are capable of monitoring the radio communication devices like smart phones, tablets, Bluetooth devices, wireless earphones, wireless audio cellular car systems, etc. Wi Fi sensors of the clients (not only the APs) Bluetooth sensors Efficient Management of PARking under COnstraints July 7, 2015 5
Sensor Selection, System Assembly, System Programming Sensors the are capable of monitoring the radio communication devices like smart phones, tablets, Bluetooth devices, wireless earphones, wireless audio cellular car systems, etc. Efficient Management of PARking under COnstraints July 7, 2015 6
Sensor Selection, System Assembly, System Programming Sensors the are capable of monitoring the radio communication devices like smart phones, tablets, Bluetooth devices, wireless earphones, wireless audio cellular car systems, etc. WiFi connection WiFi Sensor (50 200m) LoRa Connection (Long Range com.) Bluetooth Sensor (20 30m) Efficient Management of PARking under COnstraints July 7, 2015 7
Sensor Selection, System Assembly, System Programming Interface Programming (Power, Scanning Period per Protocol, Connectivity, routing, etc.) Real Time (Raw) Data Monitoring per Interface Efficient Management of PARking under COnstraints July 7, 2015 8
Sensor Selection, System Assembly, System Programming The information read from each user contains: The MAC address of the wireless interface, which allows to identify it uniquely. The strength of the signal (RSSI), which gives us the average distance of the device from the scanning point. The vendor of the Smartphone (Apple, Nokia, etc) The Class of Device (CoD) in case of Bluetooth which allows us to differentiate the type of device (Smartphone, Hands-free, Computer, LAN/Network AP). With this parameter we can differentiate among pedestrians and vehicles. Example of information monitored by the WiFi Scanner: DB ID Timestamp MAC AP RSSI Vendor 53483 2012-04-24 07:56:25 C4:2C:03:96:0E:4A 69 Apple 53482 2012-04-24 09:11:26 D8:2A:7E:10:1E:63 libelium_wsn1 60 Nokia Example of information monitored by the Bluetooth Scanner: DB ID Timestamp MAC ID RSSI CoD Vendor 45400 2012-05-16 16:18:12 00:26:7E:5F:3C:18 mycar -72 Handsfree PARROT SA 78005 2012-04-20 12:59:27 D8:2A:7E:0E:C3:10 Tropic -85 Smartphone Nokia Efficient Management of PARking under COnstraints July 7, 2015 9
Dimensioning of the Environment The amount of sensors equipment needed for a specific region to be monitored efficiently is determined by The propagation characteristics (building blocks, walls, metallic surfaces, human bodies movements, cars, etc.) The required coverage area The required accuracy with respect to position and the velocity The estimated amount of devices to be identified. The coverage areas may be modified by changing the power transmission (Tx, Gtx) of the radio interfaces allowing the creation of different scanning zones from a few meters (in order to study an specific point) to dozens of meters (to study the whole street or even the entire floor of a shopping mall). Efficient Management of PARking under COnstraints July 7, 2015 10
Algorithms for position calculation and kinematic characteristics Μέτρα Απόδοσης (Performance Metrics) Συστημάτων Εντοπισμού εσωτερικού χώρου Μέση ακρίβεια (Accuracy): Μέση απόσταση σφάλματος (mean distance error) Κατανομή ακρίβειας (Precision): Συνάρτηση κατανομής πιθανότητας (Cumulative Probability Function (CDF)) της απόστασης σφάλματος. Πολυπλοκότητα (Complexity): Η πολυπλοκότητα λογισμικού επηρεάζει το χρόνο επεξεργασίας. Κρίσιμη για αργούς επεξεργαστές, συσκευές με μπαταρίες, κινούμενα αντικείμενα άγνωστης θέσης. Ευρωστία (Robustness): Καθορίζει το αν το σύστημα συνεχίζει την ομαλή λειτουργία του, όταν μετρήσει μία καινούργια τιμή ή όταν ένα αναμενόμενο σήμα παύει να είναι διαθέσιμο. Εξελιξιμότητα (Scalability): Καθορίζει τη δυνατότητα ομαλής λειτουργίας του συστήματος όταν καλείται να λειτουργήσει σε μεγαλύτερη έκταση. Σχετίζεται με τις διαστάσεις 2 D ή 3 D. Σχετίζεται και με τον κορεσμό ασύρματων πόρων. Efficient Management of PARking under COnstraints July 7, 2015 11
Algorithms for position calculation and kinematic characteristics A) Τριγωνισμός (Triangulation) I. Τεχνικές υπολογισμού απόστασης (lateration techniques) Μετρήσεις TOA (time of arrival) Μετρήσεις TDOA (time difference of arrival) Μετρήσεις RSS (received signal strength) Μετρήσεις RTOF (Roundtrip Time Of Flight) Μετρήσεις φάσης (Received Signal Phase Method) II. Μέτρηση γωνιών κατεύθυνσης σημάτων (angulation techniques) Directional antennas, Antenna beam tracking B) Ανάλυση τοποθεσίας (scene analysis) Πιθανοτικές μέθοδοι (probabilistic methods) k πλησιέστεροι γείτονες (knn: k-nearest-neighbor) Νευρονικά δίκτυα (neural networks) Μηχανή υποστήριξης διανυσμάτων (SVM: support vector machine) Ελάχιστο M-γωνο (SMP: smallest m-vertex polygon) C) Αλγόριθμοι εγγύτητας (proximity algorithms) Efficient Management of PARking under COnstraints July 7, 2015 12
Case Study 1 Application of the whole system in the laboratory environment (movement of students with large amount of Wi Fi, Bluetooth devices). 2 m D1 D1 D1 Test Positions using 4 specific smart phones D4 D2 D2 Glass Wall Fire Unit D3 Enwalled Closets D3 D3 Aluminum Closets Glass Wall D2 D2 Network Equipment D4 D4 D4 D1 D1 D1 D1 D1 Scanner Positions (m1, m2) Workbenches Whiteboard Partition 1 Desks Partition 2 Wooden Closets Workbenches Efficient Management of PARking under COnstraints July 7, 2015 13
Case Study 1 Application of the whole system in the laboratory environment (movement of students with large amount of Wi Fi, Bluetooth devices). 2 m D1 D1 D1 D4 D4 D2 D2 Glass Wall Fire Unit D3 Enwalled Closets D3 D3 Aluminum Closets Glass Wall D2 D2 Network Equipment Scanner Positions (m1, m2) D4 D1 D1 D1 D1 D1 D4 Workbenches Whiteboard Partition 1 Desks Partition 2 Wooden Closets Workbenches Efficient Management of PARking under COnstraints July 7, 2015 14
Case Study 2 Application of the whole system in car parking during peak hours. Efficient Management of PARking under COnstraints July 7, 2015 15
Case Study 2 Application of the whole system in car parking during peak hours. Efficient Management of PARking under COnstraints July 7, 2015 16
Case Study 2 Application of the whole system in car parking during peak hours. Efficient Management of PARking under COnstraints July 7, 2015 17
Connectivity Architecture Efficient Management of PARking under COnstraints July 7, 2015 18
Connectivity Architecture Efficient Management of PARking under COnstraints July 7, 2015 19
Connectivity Architecture Efficient Management of PARking under COnstraints July 7, 2015 20
Data Base Development Efficient Management of PARking under COnstraints July 7, 2015 21
Data Base Development Monitor Libelium interfaces in (near) real time [scanning period ~40 50 sec] Bluetooth interface Wi Fi interface Web interface Access through any web browser Mobile interface Access from any Android Smart phone or Tablet Raw monitoring data MAC addresses of connected devices Time of detection (timestamp) Device ID RSSI Vendor Processed monitoring data Last detection time Estimated active time per device Total number of detected/active devices Efficient Management of PARking under COnstraints July 7, 2015 22
Data Base Development Database Tables Table Name bluetooth_table1 bluetooth_table2 wifi_table1 wifi_table2 Description Store raw data from the Bluetooth interface of the 1 st Libelium device Store raw data from the Bluetooth interface of the 2 nd Libelium device Store raw data from the Wi Fi interface of the 1 st Libelium device Store raw data from the Wi Fi interface of the 2 nd Libelium device Efficient Management of PARking under COnstraints July 7, 2015 23
Data Base Development Application Server Efficient Management of PARking under COnstraints July 7, 2015 24
Web Applications Android App Main Menu Options Menu Efficient Management of PARking under COnstraints July 7, 2015 25
Web Applications Android App Detected Devices Viewer Efficient Management of PARking under COnstraints July 7, 2015 26
Web Applications Android App Detected Devices Viewer Efficient Management of PARking under COnstraints July 7, 2015 27
Web Applications Android Studio Efficient Management of PARking under COnstraints July 7, 2015 28
Web Applications Web Browser Efficient Management of PARking under COnstraints July 7, 2015 29
Web Applications Web Browser Total Devices: 19 Active Devices: 9 Efficient Management of PARking under COnstraints July 7, 2015 30
Next Steps Future Work Evaluation of indoor positioning algorithms based on real data Incorporation of visualized environment for the area of interest. Hybrid schemes of sensor networks Use of UWB sensors Use of RFIDs Gyroscopes, magnetometers, proximity sensors Development of data fusion algorithms System installation & evaluation in real indoor parking environments Technology familiarization Mobile apps enhancing the accuracy Efficient Management of PARking under COnstraints July 7, 2015 31
THANK YOU! Efficient Management of PARking under COnstraints July 7, 2015 32
Sensor Networks and Software Platform Development for Seamless Traffic Monitoring A. Marousis, T. Economopoulos, A. Panagopoulos School of Electrical and Computer Engineering National Technical University of Athens, Greece Efficient Management of PARking under COnstraints 7July 2015