TIU Tracking SystemSponsor IntelAdvisor Prof. Robert DaaschTeam Daniel Ferguson
Man HoangLynh PhamTri TruongDung Le
2
Introduction
• Problem statement• Key requirements
Size Power Accuracy
3
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
4
System Overview
Controller+
Location Engine(s)
Detector
Users
Admin
Web App
Tag
Front-endBack-end
Proxy
Database
Wi-Fi
5
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
6
Hardware
• Deviceso Tago Detectoro Proxy
• Componentso RF12 transceivero ATmega328p MCUo WiFly 802.11b/g transceiver
• Schematic and layout by Eagle CAD• PCB by Sunstone Circuits
7
Hardware
• Tago RF12 transceivero ATmega328p MCUo Size: 1” x 1” x 1”o 240mAh coin cell battery
Product
Layout
8
Hardware
• Detectoro RF12 transceivero ATmega328p MCUo Status LEDo Size: 3.5” x 1”
Product
Layout
9
Hardware
• Proxyo RF12 transceivero ATmega328p MCUo WiFly 802.11b/g transceivero Breadboard prototypeo Ceramic antenna
10
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
11
Firmware
• Tag broadcasts• Detectors relay• Proxy forwards to Controller
Detector
Tag
Front-end
Proxy
12
Firmware
• Tago Mostly in low power stateo Periodically wakes up to broadcast
• Detectoro Listen for messages from tags and other detectorso Controlled floodingo Collision avoidance via time division
• Proxyo Listens for messages from detectorso Forwards messages to Controller
• Generallyo All speak a common message format which includes
Battery Information Infrastructure for a more sophisticated routing protocol Fixed length
13
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
14
Software
Controller+
Location Engine(s)
Web App
Back-end
ProxyMySQL Database
15
Software
• Controllero Collects data from Proxyo Feed data to Location Engineo Two operating modes
Calibrating Collects RF signatures at calibration points Stores calibration data in a local database
Locating Passes calibration data to Location Engine Collects and sorts data into packets Feeds the packets to Location Engine Stores results in MySQL Database
16
Software
Location Engine
• Theoryo Each location has a
unique & consistent RSSI pattern
o Euclidean distance
• Reduce aliasing byo Referencing the nearest
detectoro Interpolating between
two closest locations
Fingerprint Matching Algorithm
Input Filter
Output Filter
Calibration Data
RSSI Data Tag’s Location
17
Software
• Web Appo Functionalities
Visualize tags’ and detectors’ locations Add, modify, and remove tags and detectors Configure the tracking area
o Design Goals Fast Simple Easy to use
o Technologies PHP HTML5, CSS3, JavaScript
o Why Web?
18
Software
• Key points
19
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
20
Deployment
• Detector placement• Calibration density• Each tag has a unique ID with respect to other tags• Each detector has a unique ID with respect to other
detectors• Proxy possible configurations
o Proxy must have access to LANo Proxy must create an Ad-Hoc Access point for Controller to
connect to
21
Deployment
22
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
23
Results
• Size: 1” x 1” x 1”
24
Results
• Powero 240mAh coin cell batteryo 30mA transmit currento 40µA sleep currento 1sec broadcast intervalo 3ms transmit windowo 0.3% duty cycle
25
Results
• Accuracyo Average 2mo Less than 1m at calibration points
26
Agenda
• System Overview• Hardware• Firmware• Software• Deployment• Results• Conclusions
27
Conclusions
• Antenna design• More testing
o Calibration densityo Detector placement
• Improve testability• Different algorithms• Environment & signal strength
28
Acknowledgement
• Professor Robert Daasch• Alfonso Pereira & Sameer Ruiwale