Transcript CS230 Project Mobility in Energy Harvesting Wireless Sensor Network

CS230 Project
Mobility in Energy Harvesting
Wireless Sensor Network
Nga Dang, Henry Nguyen, Xiujuan Yi
What is our project?
Motivation:
Wireless Sensor Network
- Sensor nodes are powered
by batteries
- High maintenance cost
- Unreliability: network is disconnected
when nodes are out of battery
Energy Harvesting WSN
- Powered by a centralized energy
harvesting source whose energy is
delivered to sensor nodes by robot
- Advantage:
+ Green computing
+ Autonomous system
+ Low maintenance cost
Battery System
Model
Energy Harvesting
System Model
What have other groups done?
• Energy-Efficient Approaches in WSN
– Hardware layer: energy-efficient circuit,
redundant deployment
_ Network layer: energy-efficient routing protocol
and network topology
_ Operating system: dynamic voltage scheduling,
duty cycling
_ Application layer: energy-efficient quality-aware
data collection, multi-version applications
• Use robot mobility as data collector
– Robot is scheduled to visit sensor nodes,
collecting data in close range
– Goal: prolong system’s lifetime
• reduce transmission energy for sensor nodes
(shorter range)
• Find a shortest path to minimize travelling energy
• Avoid buffer flow at sensor node’s data buffer,
deliver data in time
• Usually modeled as Travelling Salesman Problem
with additional constraints
Application
Operating system
Network layer
Hardware layer
What have other groups done? (cont.)
• Use robot mobility as energy deliverer
– Robot is equipped with a large capacity battery
– Sensors’ nodes batteries are monitoring periodically
– Every hour k nodes with least remaining energy are chosen and robot
will visit and charge these nodes through wireless transfer
– Prolong system lifetime by charging extra battery
– Disadvantage:
• System lifetime extension is limited by robot’s battery capacity
• Maintenance cost: changing robot battery
How does our system work?
Execute plan:
Visit nodes and
recharge batteries
Sensor
Sensor
Nodes
Sensor
Nodes
Nodes
Send Energy Requests
Report
charging
status
Robot
Send schedule to
robot
Base Station
Collect Energy Requests
&Run algorithm to
schedule charging activity
The charging algorithm
Input: Energy request queue:
sensor deadline
Find a starting
time satisfy both
energy and timing
constraints
Input: Robot charging status
Robot speed & power consumption
& energy harvesting profile
Input to TSP:
D[i]: Deadline of each sensor node
C[i,j]: Time to travel from node i to node j
W[i]: Waiting time at each node i to
charge
Output:
A sequence of sensor nodes
which robot had to visit
Travelling
Salesman Problem
If the robot can’t visit all the node.
- It should find the maximum subset of
nodes it can visit and give the shortest
path of that subset.
Charging algorithm example
12:00
7:30
1.5 hours
13:15
21:00
2 hours
2 hours
0.5 hour
charging
23:00
2 hours
1 hour
1 hour
6:10
0.5 hour
charging
8:00
leave base station at 5:00
get back at 23:40
1 hour
9:30