Вы находитесь на странице: 1из 28

Design and Implementation of a Low-Cost

and Resilient Wireless Mesh Network for


First-Response Communications
Nestor Michael C. Tiglao, PhD
Angelie A. Dela Cruz
Miguel Luis A. Parabuac
Ubiquitous Computing Laboratory - EEEI
University of the Philippines Diliman

Presentation Outline
-

Introduction
Review of Related Works
Problem Statement and Objectives
Methodology
Results and Discussion
Conclusions
Recommendations

Outline | 01

Introduction
- Disaster relief and response in
the Philippines is lacking and
unreliable

Introduction | 02

Introduction

Introduction | 03

Review of Related Works


-

Network Technology: WiFi 802.11n/s


Routing Protocol: Batman-adv
Chat Messaging App: Android-based
Messaging Protocol: MQTT

Review of Related Works | 04

Problem Statement

Lack of proper communication in disaster scenarios


Consider when text messaging is unavailable
A reliable and cheap communication platform is needed
Retrieve messages sent while the user was away (caching
feature)

Problem Statement | 05

Objectives
Design and implement a wireless mesh network comprised
of Raspberry Pis
Develop a chat messaging Android app
Integrate the infrastructure network with the application
Implement caching

Objectives | 06

Methodology

Network Infrastructure

Figure 1: A fully-working wireless mesh network

Methodology | 07

Android Chatting Application


- Android Studio
- PAHO MQTT Library
- Android 2.1+

Methodology | 08

Testing Phase
- Functionality
- Messaging
- Caching Feature
- Resilience
- Self-healing
- Self-configuration
- Internet sharing

Methodology | 08

Results and Discussion

Ad-Hoc Network
R Pi 2

R Pi 4

R Pi 1

iwconfig
wlan1 IEEE 802.11bgn ESSID:RasPi
Mode:Ad-HOc
Frequency:2.427GHz
Cell:02:12:34:56:78:9A
Tx-POwer=20 dBm
Retry short limit:7
RTS thr:off
Fragment thr:off
Power Management:off

R Pi 3

Figure 2: RasPi ad-hoc network

Results and Discussion | 09

Mosquitto Brokers

type / receiver / sender / time_stamp / node

Results and Discussion | 10

Mosquitto Bridges
RPi 1
+/+/+/+/internet

+/+/+/+/node1

Figure 3: Topic matching for the


mosquitto bridges

+/+/+/+/node1

+/+/+/+/node2
+/+/+/+/internet

RPi 2

RPi 4
+/+/+/+/node2

+/+/+/+/node2
+/+/+/+/node3
+/+/+/+/node3
+/+/+/+/node3
+/+/+/+/node1

+/+/+/+/internet

RPi 3

Results and Discussion | 11

Messages To/From Internet

+/+/+/+/node1
+/+/+/+/node2
+/+/+/+/node3

RPi 4

Internet
+/+/+/+/internet

Figure 4: Topic matching for internet


messaging

Results and Discussion | 12

Rerouting Using Batman-adv


BATMAN-adv nodes:
c8:3a:35:c8:97:31
c8:3a:35:c0:3a:1d
c8:3a:35:c1:44:90
c8:3a:35:ca:68:70

Results and Discussion | 13

Rerouting Using Batman-adv

RPi1

RPi2

RPi3

Results and Discussion | 13

Rerouting Using Batman-adv


RPi2

RPi4

RPi3

RPi1

Results and Discussion | 14

Rerouting Using Batman-adv


RPi2

RPi4

RPi3

RPi1

Results and Discussion | 15

Rerouting Using Batman-adv


RPi2

RPi4

RPi3

RPi1

Results and Discussion | 16

Rerouting Using Batman-adv


RPi2

RPi4

RPi3

RPi1

Results and Discussion | 17

Crisis Chat Mobile App


- Two Modes
- Internet (as Rescuer)
- AdHoc (local)
- Messaging
- MQTT QoS2
- Caching Feature
- Receive messages even if user is
offline
- AdHoc Mode
- Detects nearest Raspberry Pi
Results and Discussion | 18

Conclusion and
Recommendations

Conclusion
- We have proven that we can make a low-powered, reliable,
and cheap wireless mesh network using Raspberry Pi nodes.
- The messages in the network are cached for future use.
- We devised a way to tag messages for the bridge
configuration
- Established bidirectional communication with the internet
- The project is 100% complete.

Conclusion and Recommendations | 19

Problems Encountered
- Linux driver problems
- Android problems
- Does not support ad-hoc connectivity
- Does not instantly return WiFi scan results
- Testing phase

Conclusion and Recommendations | 20

Recommendations
- More features and material design on chat messaging app
- Create/port an iOS equivalent (Android <-> iOS
communication)
- Subject the network to heavier load
- Deploy in an actual community

Conclusion and Recommendations | 21

Thank You!

Вам также может понравиться