DATA PRESERVATION IN INTERMITTENTLY CONNECTTED SENSOR NETWORK WITH DATA PRIORITY - PowerPoint PPT Presentation

Loading...

PPT – DATA PRESERVATION IN INTERMITTENTLY CONNECTTED SENSOR NETWORK WITH DATA PRIORITY PowerPoint presentation | free to download - id: 6c9d94-NjViM



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

DATA PRESERVATION IN INTERMITTENTLY CONNECTTED SENSOR NETWORK WITH DATA PRIORITY

Description:

DATA PRESERVATION IN INTERMITTENTLY CONNECTTED SENSOR NETWORK WITH DATA PRIORITY Bin Tang Department of Computer Science California State University Dominguez Hills, CA * – PowerPoint PPT presentation

Number of Views:14
Avg rating:3.0/5.0
Date added: 29 October 2019
Slides: 30
Provided by: csudhEdu
Learn more at: http://csc.csudh.edu
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: DATA PRESERVATION IN INTERMITTENTLY CONNECTTED SENSOR NETWORK WITH DATA PRIORITY


1
DATA PRESERVATION IN INTERMITTENTLY CONNECTTED
SENSOR NETWORK WITH DATA PRIORITY
  • Bin Tang
  • Department of Computer Science
  • California State University Dominguez Hills, CA

2
Outline of My Talk
  • Introduction
  • Data Preservation in Sensor Networks
  • Motivation
  • Problem Formulation
  • Centralized Algorithm
  • Distributed Algorithm and Protocol
  • Simulation Analysis
  • Conclusions and Future Work

3
System Architecture of a Sensor Node
  • TelosB Sensor
  • 16- bit RISC Processor
  • 10 KB RAM
  • IEEE 802.15.4 at 250 Mbps
  • Two AA batteries or USB
  • Sensing transducer, micro-controller unit (MCU),
    wireless radio transceiver, memory, and battery
  • Very limited resources and capabilities

4
Sensor in a Nutshell
5
Wireless Sensor Networks
  • Multi-hop wireless networks consisting of small
    sensor nodes equipped with sensing, communication
    and processing capabilities
  • Base Station
  • Multi-hop, many-to-one communication

6
Applications
  • Battle Field Surveillance
  • Structural Health Monitoring

7
Ecological Monitoring
Source http//www.networkworld.com/
8
Community Seismic Network
Source http//csn.caltech.edu/
9
My Current Sensor Network Research
  • Energy-efficient algorithms and protocols
  • Tailored to application requirements and
    constraints
  • Application
  • Network
  • Transport
  • Link
  • Physical (Hardware)

10
Intermittently Connected Sensor Networks
X
  • Underwater/ocean seismic sensor networks,
  • volcano eruption/glacial melting monitoring
  • Not feasible to install base station in field
  • Data generated and stored in the network,
  • periodically uploaded via data mules or
  • satellite links
  • Data Priority data generated may have
  • different importance (seismic,
  • infrasonic, temperature)

Source http//fiji.eecs.harvard.edu/Volcano
11
Data Preservation In Intermittently Connected
Sensor Networks
  • Non-uniform data generation and limited storage
    capacity
  • Data generators (DGs) storage-depleted
  • Data preservation offload overflow data from
    DGs to nodes with available storage
  • Data from different DGs are of different
    importance

Data Generator
Sensor node
DG1
DG2
v
DG3
priority of DG1 priority of DG2 gt priority of
DG3
12
Challenge in Data Preservation
  • Limited battery power
  • Data preservation consumes battery power
  • When not all the data can be preserved inside the
    network, how to ensure data preservation with
    maximum total priorities data preservation with
    data priorities (DPP)

Data Generator
Sensor node
DG1
DG2
v
DG3
priority of DG1 priority of DG2 gt priority of
DG3
13
Data Preservation With Priority (DPP)
  •  aaaaaafdafasdfasdf

14
Objective of DPP
  • Select a subset of data items to offload to
    maximize their total priorities

Priority of node 1 2 Priority of node 3
1 Total preserved priorities 3
Fig.1. Illustration of the DPP problem.
15
Maximum Flow Problem
  • Given a flow network G with source s and sink t,
    find maximum amount of flow from s to t

12
12
v3
v1
11
20
16
15
1
s
4
t
7
9
7
4
13
4
v4
v2
8
4
14
11
16
Multiple Sources and/or Sinks
s1
s1
s2
t1
s2
t1
?
s3
t2
s
t
s3
t2
s4
t3
s4
t3
s5
s5
17
Maximum Weighted Flow (MWF)
  •  

s1
s2
t1
 
s
t
s3
t2
s4
t3
Find a flow with maximum total weight from s to
t
s5
18
Priority-based Algorithm (PBA) for MWF
  • Find maximum flow (using Edmonds-Karp) in
    non-ascending order of source nodes priority
  • Optimality proof of PBA (omitted)
  • Maximum weighted flow is a maximum flow, but not
    vice versa
  • Time complexity O(knm2)

19
Optimal Algorithm for DPP
  • Transform G into G'

G'
G
PBA on G is an optimal algorithm for DPP on G.
Priority of node 1 2 Priority of node 3
1 Total preserved priorities 3
20
A Heuristic Algorithm of DPP
  • Offload data in non-ascending order of DGs
    priority
  • for each DG
  • while (It can still off a data item from
    it to a non-DG node)
  • Offload it to the closest non-DG node
  • Time complexity O(km kdn), d is average
    number of data items of each DG

21
Performance Evaluation
  • Visual Performance Comparison
  • Grid network 10 10, 20 20
  • Initial energy level 30 units
  • Each DG has 30 data items to offload (50 data
    items in 20 20 grid)

Shape Coordinate Priority
DG1 (4, 6) 8
DG2 (7, 6) 6
DG3 (4, 5) 4
DG4 (7, 5) 2
22
Data Preservation Blocked by Storage Constraint
23
Data Preservation Blocked by Energy Constraint
24
Push-Relabel Algorithm
  • active node - a node with excess flow
  • Relabel increase the height of the active
    node to push excess flow
  • Push send the excess flow to the neighbors
  • Terminates when no active nodes left

25
Distributed Data Preservation with Data Priority
Theorem 3
The distributed data preservation algorithm
  • Each DG broadcasts its priority to the network
  • for each DG in the non-ascending order of its
    priority
  • s pushes maximum allowable data to this DG
  • while (there exists a node u with positive
    excess)
  • Push-Relabel(u)
  • end while
  • end for

The distributed algorithm preserves maximum total
priority. It runs in O(kn2) time and uses O(n2m)
messages.
preserves maximum total priority. It runs in
O
(
kn
2
)
time and
uses
O
(
n
2
m
)
messages.
26
Distributed Algorithm Comparison
27
Conclusions
  • Data preservation in sensor networks by
    considering data priorities (DPP)
  • Maximum weighted flow (MWF), generalizing maximum
    flow problem
  • Distributed algorithm

28
Topics for Senior Project or M.S. Projects/Theses
  • DGs of low priority discard their locally
    generated data
  • General energy model
  • Combining data preservation and data retrieving
  • Data Resiliency With Data Aggregation
  • Qualification
  • Good Algorithms Knowledge
  • Good Programming Skills
  • Motivation and hardworking

29
THANKS! btang_at_csudh.edu
About PowerShow.com