A UBIQUITOUS PUBLISH/SUBSCRIBE PLATFORM FOR WIRELESS SENSOR NETWORKS WITH MOBILE MULES

1
A UBIQUITOUS PUBLISH/SUBSCRIBE PLATFORM FOR
WIRELESS SENSOR NETWORKS WITH MOBILE MULES
Xiaoyu Tong and Edith C.-H. Ngai Dept. of Information Technology, Uppsala University, Sweden
2

Problems of Ubiquitous Sensing

• Stationary sensors are more likely to be sparsely
  deployed in most scenarios.
• But it is cost prohibitive to connect each of
  them to the Internet at every site.
• As a result, their deployments and usages are
  heavily limited.

Sparsely Deployed Temperature Sensors in a Forest
3

Problems of Ubiquitous Sensing

• On the other hand, applications of mobile sensors
  become incredibly popular.
• But they are constrained by sensor types and
  abilities.
• Our goal
• Easier and more flexible deployments of
  stationary sensors.
• Enable ubiquitous sensing services from both
  stationary sensors and mobile sensors.

4

System Model

• Publish/subscribe paradigm is employed to support
  asynchronous communications.
• Smart phones are utilized as both mobile sensors
  and data mules.
• Built-in sensors GPS, Camera, etc.
• Short range communication with wireless sensors,
  IEEE802.15.4, Bluetooth, or ANT.
• Encounter and communicate with wireless sensors
  opportunistically.
• Stationary sensors are sparsely deployed.

5

Key Challenges

• Deployments of both stationary sensors and mobile
  sensors are dynamic.
• Communication overhead will draw power quickly
  from both mobile phones and stationary sensors.
• Mobile users who carry the smart phones have
  individual and uncontrollable mobility.

6

Related Work

• MQTT-S has been proposed to enable topic-based
  publish/subscribe in WSNs.
• Huang et al. described how to distribute the
  system across multiple computers or mobile
  devices to distribute load and cope with
  failures, message loss and disconnections.
• The above works focus on adoption of
  publish/subscribe in WSNs and publish/subscribe
  system with mobile environments, while the
  sensing context, heterogeneity of sensing
  devices, mobility of nodes and limited
  connectivity of wireless sensors remain to be
  further explored.

7

Network Model

• Stationary Sensor SSensor.
• Mobile Sensor MSensor.
• Platform Server Broker.
• Mobile Phone Mobile Broker.
• Sparse Sensor Networks SSNs.

8

Publish/Subscribe Platform Design

• Area-based Subscriptions

9

Publish/Subscribe Platform Design

• Adaptive Location Updates
• How often should a MB update its locations?
• Default periodically with constant time interval
• Reduce unnecessary location updates

10

Publish/Subscribe Platform Design

• Full Installation
• Best-effort Subscription Delivery
• Number of MSensors in a target area is higher
  dynamic.
• Installations of subscriptions to SSensors are
  opportunistic.
• Keep deployments of sensors transparent to
  platform.

11

Publish/Subscribe Platform Design

• Event Data Collection
• ECF flag avoiding unnecessary collection.
• Push-based event collection for MSensors.
• Pull-based event collection for Ssensors.

12

Implementation

• Broker
• Restful Http Server.
• Mobile Broker
• Android Usb cable IEEE 802.15.4.
• ANT or Bluetooth 4 in the future
• Stationary Sensor
• Contiki Rime Coffee IEEE802.15.4.

13

Experimentation

• Scenario and Settings

14

Experimentation

• Installation of Stationary Sensors along Hiking
  Trail.

15

Experimentation

• Sensing Data from Stationary Sensors

16

Experimentation

• Sensing Data from Mobile Sensors

17

Experimentation

• Sensing Data from Mobile Sensors

18

Experimentation

• Data Delivery Delay

19

Experimentation

• Number of Received Data

20

Experimentation

• Communication Overhead

21

Summary

• We proposed and implemented a novel
  publish/subscribe platform for ubiquitous data
  access from both wireless sensors and mobile
  phones without relying on any network
  infrastructure.
• Mobile phones are utilized as mobile mules to
  relay subscriptions and published data between
  the Internet users and the remote sensors.
• We implemented our platform on real hardwares and
  experimented it in a hiking trail application.
• In the future, we plan to study the coordination
  among heterogeneous mobile and sensing devices to
  further reduce the energy consumption and
  communication overhead.