Nikolaos Rapousis 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 Speaker: Nancy Panousopoulou 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 Performance Analysis of a User-centric Crowd-sensing Water Quality Assessment System
33
Embed
Performance Analysis of a User-centric Crowd-sensing Water ... · Performance Analysis of a User-centric Crowd-sensing Water Quality Assessment System . Resources WHH-WASH Sector
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Nikolaos Rapousis 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
Speaker: Nancy Panousopoulou
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
Performance Analysis of a User-centric Crowd-sensing Water Quality Assessment System
• 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
4/20/2016 CYSWATER 4
Motivation • Efficient monitoring & management of the infrastructure, including the last-mile access
network
• User in the loop: user engagement
• 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., [Jonoski13], Delhi Jar board, 311)
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 21 13-Apr-2015
QoWater Testbed Sensor • 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 OS • 512 MB RAM • 3.7 inches (480 x 800)
CySWater 22 13-Apr-2015
Emulated QoWater clients • VM with 12 cores at 2.4 GHz • 32 GB RAM, 150 GB storage • Windows7
• Each scenario was executed 20 times • QoWater sensor uploading is the most time consuming (background does not effect response) • Note that u-map users different dataset size and response processing
Power Consumption
Start AppScope
End AppScope
Launch application
AppScope Inter Process Communication
Execute scenario
Exit application
Start AppScope
End AppScope
Collect measurements
Create report
• AppScope for the energy measurements • QoWater client is an HTC Nexus One smartphone • Compare the QoWater with the u-map & popular apps (e.g., Skype, YouTube)
• 100 repetitions of each scenario • The 3G has been omitted • Display is the most energy demanding in all scenarios
Scalability
Response Delay
Served vs. Time-out
V-Trickle vs Periodic
V-Trickle extends the battery lifetime approximately 4 times
Conclusions
• The QoWater is a monitoring & querying mechanism that can engage citizens, improve the transparency of the monitoring process and provide alerts in case of contamination events
• Relatively low response delay and power consumption • Most demanding scenario consumes less than a 20 sec Skype call • Current server cannot adequately support large-scale regions – possible solutions through
cloud computing • Preliminary earlier field study indicates that users can distinguish different sources of water
CySWater 22 13-Apr-2015
Future work
• Analysis of the user sensitivity regarding the water quality • Prediction models of contamination events based on extensive field studies • Analysis of the impact of incentives to enhance the participation of citizens on systems