UNIVERSITY OF CRETE - Hellas€¦ · QoWater clients on mobile devices • Collect ....
Transcript of UNIVERSITY OF CRETE - Hellas€¦ · QoWater clients on mobile devices • Collect ....
QoWater - A crowd-sourcing approach for assessing the water quality
Nikolaos Rapousis, Michalis Katsarakis and Maria Papadopouli
UNIVERSITY OF CRETE Department of Computer Science
Department of Computer Science, University of Crete Institute of Computer Science, Foundation for Research & Technology – Hellas (FORTH)
http://www.ics.forth.gr/mobile [email protected]
This research has been funded by a GSRT Research Excellence grant (2012-2015) , a Google Faculty Award (2013-2014), and EU Hydrobionets
13-Apr-2015 CySWater 1
Resources WHH-WASH Sector Report, FAO, WHO, UN
CySWater 3 13-Apr-2015
Roadmap • Introduction & Motivation
• Related work
• QoWater system
• Proof-of-concept & pilot testbed
• Preliminary field study & evaluation results
• Conclusions and future work
CySWater 2 13-Apr-2015
Contamination events • Pressure loss or change may result in backflow incidents during which
contaminated soil water enters the water distribution network (WDN) through pipe breaks or leaking joints
• Corrosion of iron, copper, and lead parts of the WDN (e.g., due to free chlorine for disinfection)
• Bioterrorism
4/21/2015 CYSWATER 4
Motivation • Efficient monitoring & management of the infrastructure, including the last-mile
access network
• Incorporation of customer feedback
• Increased customer awareness about the water quality & querying mechanisms
• Fast & efficient warning/alerting in the case of contamination
• More accurate models for assessing the quality of the water
CySWater 7 13-Apr-2015
Roadmap
• Introduction & Motivation
• Related work
• QoWater
• Proof-of-concept
• Evaluation
• Conclusions and future work
CySWater 4 13-Apr-2015
Related activities in water contamination 1. Mobile apps for querying specific sensors & providing feedback e.g., [Jonoski’13, Delhi Jar board, 311]
2. Sensor placement for supporting warming systems e.g., [Murray’10, Perelman’13]
3. Detection algorithms for drinking water contamination e.g., [Murray’10, Mix’12]
CySWater 5 13-Apr-2015
4/21/2015 CYSWATER 9
4/21/2015 CYSWATER 10
Roadmap
• Introduction
• Related work
• QoWater system
• Proof-of-concept
• Evaluation
• Conclusions future work
CySWater 6 13-Apr-2015
Health Organization sensors
Citizens/Customers with QoWater client
Water Distribution Network
QoWater client
QoWater server
Regulators, administrators, scientists
sensors sensors
QoWater system: client-to-server architecture
QoWater clients on mobile devices • Collect customer/scientist sentiments & store them locally • Upload data to the QoWater server • Query the QoWater server to acquire quality related info for various regions
QoWater server • Collects client and sensor data in spatio-temporal geo-DB • Responds to queries sent by users, providers & regulators
CySWater 8 13-Apr-2015
Sensor measurements pH, temperature, conductivity oxidation-reduction potential & Ion Chlorine
Main screen
CySWater 14 13-Apr-2015
Query based on clients score or sensor measurements
Results for query area, based on clients option eg, customer score
CySWater 15 13-Apr-2015
CySWater 16 13-Apr-2015
Color: red/brown, green/blue, white, black Taste: bitter, salty, sweet Odor: sewer, chlorine, gasoline, chemical Appearance: floating particles, sand, milky, rusty, stain, animal, plant Pressure: no water, low, high
CySWater 17 13-Apr-2015
Taste: bitter, salty, sweet Odor: sewer, chlorine, gasoline, chemical Appearance: floating particles, sand, milky, rusty, stain, animal, plant Pressure: no water, low, high
Chemical: turbidity, Br, Cl, Na, K, Mg2, biochemical oxygen demand, acidity, disssolved oxygen, etc Biological: ephemeroptera, plecoptera, excherichia coli, coliform bacteria, etc
QoWater Architecture QoWater server
Secure sensor readings upload
Certificate Manager
Memcached service
Certificate Authority
Data receiver
Query handler
Access control geoDB
Event detector
Data validator
Analyzer
HTTP service
PHP application
Upload data
QoWater sensor node Broadcast sensor readings
Monitor
Temperature
Conductivity
ORP
PH
Chlorine ions
Back-end interface
WiFi interface
QoWater client
Android device
Posi
tion
Ener
gy
Que
ries
Save
dat
a
Data Recorder
Cust
omer
Fo
rm
Scie
ntist
Fo
rm
Que
ryin
g
GUI
Sensor reading
GPS
Camera
Battery
Feedback
CSV
Measurement location
Back-end interface
Save sensor readings
SQLite
CySWater 18 13-Apr-2015
QoWater server
Secure sensor readings upload
Certificate Manager
Memcached service
Certificate Authority
Data receiver
Query handler
Access control geoDB
Event detector
Data validator
Analyzer
HTTP service
PHP application
Upload data
QoWater sensor node Broadcast sensor readings
Monitor
Temperature
Conductivity
ORP
PH
Chlorine ions
Back-end interface
WiFi interface
QoWater client
Android device
Que
ries
Save
dat
a
Data Recorder
Cust
omer
Fo
rm
Scie
ntist
Fo
rm
Que
ryin
g
GUI
Sensor reading
GPS
Camera
Battery
Feedback
CSV
Measurement location
Back-end interface
Save sensor readings
SQLite
CySWater 19 13-Apr-2015
Rely on the end-to-end security that protects the integrity & confidentiality by leveraging standard technologies (eg public-private key pairs, TLS)
Use Hadoop Distributed FS & Hbase for higher aggregate I/O throughput
Preliminary field study Three sources of drinking water: tap, purified, bottled
Objective measurements by sensor node • Duration of sampling: 36 minutes • One data point every 10 seconds • Collection of 215 data points (omit the first 50 data points)
Subjective measurements (QoE score) by 44 subjects • Three cups containing tap, purified & bottled water with no indication of the source of the water • Each subject inspects, smells, tastes the water from each cup and provides its score immediately
CySWater 20 13-Apr-2015
QoWater Testbed Sensor • Equipped with a micro-controller that operates at 14MHz frequency • 8 KB RAM memory • 2 GB SD card • IEEE802.11b/g WiFi Server • VM with 2 cores at 2.4 GHz • 4 GB RAM • 27 GB storage • Ubuntu 14.04 OS
Client • Android 2.1
CySWater 21 13-Apr-2015
CySWater 22 13-Apr-2015
The correlation of QoE with conductivity, ORP, & pH is negative, while with ion Chlorine is positive
CySWater 23 13-Apr-2015
Divalent metal ions (i.e., Ca++, Mg++, and Fe++) influence the conductivity & hardness Conductivity and hardness impacts negatively the perceived water quality
Official Chemical Analysis
CySWater 24 13-Apr-2015
Divalent metal ions (i.e., Ca++, Mg++, and Fe++) influence the conductivity & hardness Conductivity and hardness impacts negatively the perceived water quality
Official Chemical Analysis
CySWater 25 13-Apr-2015
Official Chemical Analysis
Main outcome
• The QoWater, a crowd-sourcing/sensing system, provides collector & querying mechanism
• The QoWater can inform customers & regulators for contamination events
• Preliminary field study indicates that users can distinguish different sources of water
CySWater 22 13-Apr-2015
Future work
• Implementation of the data validation & access control modules • Deployment of a larger-scale testbed & perform empirical studies
• Better understanding of the user sensitivity regarding the water quality
• Detection of contamination events
CySWater 23 13-Apr-2015