Maximizing the lifetime of WSN using VBS

Post on 03-Jan-2016

27 views 0 download

description

Maximizing the lifetime of WSN using VBS. Yaxiong Zhao and Jie Wu Computer and Information Sciences Temple University. Road map. Introduction and background Centralized scheduling STG-based approach VSG-based approach Distributed implementation Iterative local replacement - PowerPoint PPT Presentation

Transcript of Maximizing the lifetime of WSN using VBS

Maximizing the lifetime of WSN using VBS

Yaxiong Zhao and Jie WuComputer and Information Sciences

Temple University

Road map

Introduction and background Centralized scheduling

STG-based approach VSG-based approach

Distributed implementation Iterative local replacement

Conclusion and future work

Road map Introduction and background Centralized scheduling

STG-based approach VSG-based approach

Distributed implementation Iterative local replacement

Conclusion and future work

Introduction

The need of reducing energy consumption and extending the network lifetime The most important challenge

We have only one general technique Duty-cycling To exploit the redundancy in sensors

Traffic is low Letting sensors work all the time is redundant for

transmitting data

The redundancy in the network level

Usually there are more-than-enough sensors deployed in the network For reliability and QoS

The same degree of redundancy is not necessary for communication Low traffic Static network 99.8% delivery ratio

Our idea

Scheduling multiple backbones to maintain the connectivity

Backbone sensors use duty-cycling to further reduce energy consumption

Turn off other sensors' radios The independent backbones is not

optimal In the example overlapped backbones help

further extend network lifetime

0 1

2 3 4

sink

0 1

2 3 4

sink

Maximum lifetime backbone scheduling

An example {Sink, 0, 1} work for 1 unit {Sink, 0, 3} work for 1 unit {Sink, 1, 3} work for 2 units Total network lifetime of 4 units of time

Find a schedule <b0, t0> … <bi, ti>

A backbone bi works for ti round(s) Has the longest network lifetime

NP-hard Reduce from the maximum set cover (MSC)

problem

0 1

2 3 4

sink

Road map

Introduction and background Centralized scheduling

STG-based approach VSG-based approach

Distributed implementation Iterative local replacement

Conclusion and future work

Scheduling Transition Graph

The time is divided into multiple rounds A backbone is selected at each round

The residual energy of each sensor is recorded with each backbone at each round

A fixed amount of energy is consumed in each round

Enumerate candidate backbones Form a graph representing the schedule

STG (cont'd)

{B1, E1}

{B2, E2}

{B3, E3}

{Bp, Ep}

{B1, E1}

{B2, E2}

{B3, E3}

{Bp, Ep}

{B1, E1}

{B2, E2}

{B3, E3}

{Bp, Ep}

Round 1 Round 2 Round i ……

Backbone transition

Initial

Round 0 {B, E} are: The backbone The associated residual

energy of all the sensors in the network

A path in the STG represents a schedule

Path ends when at least one sensor depletes energy

The purpose of our algorithm is to find the longest path

Road map Introduction and background Centralized scheduling

STG-based approach VSG-based approach

Distributed implementation Iterative local replacement

Conclusion and future work

Virtual Scheduling Graph

Transform a sensor into multiple virtual nodes Each virtual node represents a fixed amount of energy

And has a virtual ID The energy consumed in each round

Virtual nodes are connected based on several rules The virtual nodes of the same sensor form a clique The virtual nodes of the neighboring sensors connect

correspondingly with increasing order

virtual node of C

virtual node of A

virtual node of B

0

0

1

0

1CB

A

2

VSG (cont’d)

VSG works by sequentially finding the CDS Then remove the selected nodes Until a sensors' virtual nodes have all been removed

Road map Introduction and background Centralized scheduling

STG-based approach VSG-based approach

Distributed implementation Iterative local replacement

Conclusion and future work

Iterative local replacement

Let each sensor find replacements locally Sensors that have less energy should have a

higher chance to switch than those that have more energy Ec is the energy consumed since the last time

working as a backbone Er is the current residual energy

Experiment results

Conclusion and future work

A new scheduling method Two centralized approximation algorithms A distributed implementation

More theoretical inquires are needed Testbed implementation