presentation l`aquila new

42
Senso LAB Prof Orhan Gemikonakli Dr Enver Ever Dr Leonardo Mostarda Krishna Doddapaneni SENSO LAB is one of the most advanced wireless sensor network labs., hosting hundreds of heterogeneous motes that are deployed around Middlesex University. More than 20 members www.eis.mdx.ac.uk/staffpages/leonardom/ WSN/index.php

Upload: ikrrish

Post on 14-Jun-2015

127 views

Category:

Technology


0 download

TRANSCRIPT

Page 1: Presentation l`aquila new

Senso LAB

Prof Orhan GemikonakliDr Enver EverDr Leonardo MostardaKrishna Doddapaneni

SENSO LAB is one of the most advanced wireless sensor network labs., hosting hundreds of heterogeneous motes that are deployed around Middlesex University.

More than 20 members

www.eis.mdx.ac.uk/staffpages/leonardom/WSN/index.php

Page 2: Presentation l`aquila new

ENERGY AWARE PERFORMANCE EVALUATION OF WSNS

Krishna Doddapaneni

School of Science and TechnologyKrishna Doddapaneni

School of Engineering and Information Sciences

Page 3: Presentation l`aquila new

Overview

• Wireless sensor networks

• Evaluation methods

• Modeling Framework

• Unequal clustering algorithm – UHEED

• IDS in WSN

• PlaceLife

• Present work

• Future work

• Castalia

Page 4: Presentation l`aquila new

What are they and where are they used

• A large number of small-sensing self-powered nodes gathering information

• Communicating in a wireless fashion with an end-goal to hand their processed data to a base-station

• Key elements: Sensing, Processing & Communication

• Main area of focus - Body Sensors (Hospitals), Bio-Medical, Large-Scale Sport Fields, Industrial Automation

Page 5: Presentation l`aquila new
Page 6: Presentation l`aquila new

Oil spills

• Gulf oil spill

• No monitoring (use of faulty concrete plugs)

• Mesh-based wireless sensor networks for constant monitoring of the rigs

• Nigerian government funding• WSN for monitoring oil spills• The project went through several stages• The notification is in November• Equipment-provided

Page 7: Presentation l`aquila new

and more…

• Bridge Monitoring

• In California, 13% of the 23,000 bridges have been deemed structurally deficient, while 12% of the nation's 600,000 bridges share the same rating

• Structural health monitoring (SHM) is a sensor-based pre-emptive approach

• New York may be the first state with a 24/7 wireless bridge monitoring system

• Another application in India: Bri-Mon (Monitoring Railway bridges)

Page 8: Presentation l`aquila new

Crucial Factors

• Life time of sensor node:

1. Microprocessor

2. Sensing module

3. Wireless transmitter/receiver.

Existing studies consider these modules for best deployment, topology, protocol selection, etc.

Page 9: Presentation l`aquila new

Energy consumption in a sensor node can be attributed to either “useful” or “wasteful”

sources.

Useful energy consumption:

Wasteful energy consumption:

• Transmitting/receiving data.

• Processing query requests.

• Forwarding queries/ data to neighbouring nodes.

• Idle listening.

• Retransmissions.

• Overhearing.

• Generating/handling control packets.

Page 10: Presentation l`aquila new

Evaluation Methods

Page 11: Presentation l`aquila new

Modeling Framework

Page 12: Presentation l`aquila new

Path loss

• Attenuation in power density of an electromagnetic wave as it propagates.

• Path loss is effected by free-space loss, refraction, diffraction, reflection, coupling loss, absorption, propagation medium…..

• Path loss effects should be considered for a more realistic evaluation

Page 13: Presentation l`aquila new

Case study

Page 14: Presentation l`aquila new

Simulation

Where,

Lp is path loss between 2 points

Lo is path loss in Open space

mtype is number of objects of same

type

wtype is loss in decibels attributed to

that object

d is distance between the points

Path loss calculation

Page 15: Presentation l`aquila new

Energy: with and without path loss.

Page 16: Presentation l`aquila new

Clustering

• Achieve high energy efficiency

• Increase the network scalability

• Each cluster has a coordinator(CH) & number of nodes

• Nodes only communicate to their CH.

• Data aggregation, rotation of CH

• Distribution of load across all nodes

Cluster head CH

Member node

Advantages

Why clustering

Page 17: Presentation l`aquila new

Thank you !

Multihop Communication

bc

a

• Data travels from the source to the destination node via more than two hops.

• Increase the range of the network by a significant margin

d

Page 18: Presentation l`aquila new

It is ineffective to balance loads among cluster heads to avoid hot

spots problem, if the cluster heads are uniformly distributed,

like in HEED.

Uniformly distributed

Page 19: Presentation l`aquila new

Unequal clustering algorithm (UHEED)

• Clustering, Multihop Communication,

• Mitigates Hotspots !

• UHEED combine HEED and EEUC • The leader election is performed according to HEED• The radius size is calculated according to EEUC

• Improves network life time.

Page 20: Presentation l`aquila new

Equal sized clusters

Unequal sized clusters

Page 21: Presentation l`aquila new

UHEED mechanism

• Unequal sized clusters are based on the distance from a cluster head to the base station and energy level.

• The further a cluster head is located from the BS, the larger its competition radius is, and hence the size of the cluster.

• Unequal sized clusters reduce intra-cluster traffic for CH nearer to BS.

Page 22: Presentation l`aquila new

Where,

is maximum competition radius, predefined.

and are max. and min. distances.

C is constant coefficient between 0 and 1.

The life time of CH closer to BS is more critical, the clusters further away have larger sizes compared to

closer ones.

Competition radius

Page 23: Presentation l`aquila new

IDS in WSN

- Watchdog

- Promiscuous mode – radio continuously on, to check the correct behaviour of other nodes.

- Hence, lifetime decreases.

- Not really suitable for eventtriggered sensing.

- Agreement Based

- Monitoring based on pre-defined agreement.

- Byzantine oral solution/ signed messages algorithms.

- It is also expensive in terms of no of messages sent and time.

Enhances security

Hence, a need for new approaches, improve Agreement based IDS

Page 24: Presentation l`aquila new

The Byzantine Generals Problem

Attack!

Wait…

Attack!

Attack! No, wait! Surrender!

Wait…

Page 25: Presentation l`aquila new

Oral Message Algorithm

Page 26: Presentation l`aquila new

Simulation and Results

10

9

1

8

7

6

5

4

3

2

Case study considered

Page 27: Presentation l`aquila new

Radio always On (promiscuous mode)

Byzantine oral message solution

No IDS

Page 28: Presentation l`aquila new

Our Frame work

Page 29: Presentation l`aquila new

PlaceLife

Software Architecture Modelling Language• Set of components that exchange messages• Components have variables manipulated by the behaviour• Behaviour is represented by a list of events, conditions and

actions

Page 30: Presentation l`aquila new

PlaceLife

Node Modelling Language • Operating system, implemented MAC protocols, routing

protocols• Hardware specification

Page 31: Presentation l`aquila new

PlaceLife

Environment Modelling Language• The physical environment in which the WSN nodes are

deployed • Obstacles, material….

Page 32: Presentation l`aquila new

Weaving models• Mapping Modelling Language• Deployment Modelling Language

This approach provides a clear separation between software components, WSN nodes and the physical environments, thus promoting the reuse of models.

PlaceLife

Page 33: Presentation l`aquila new

PlaceLife

Page 34: Presentation l`aquila new

Present work!

• Expressiveness of Languages : Precision of our abstraction

• Improvise the path loss model

Where,

The one we used in our earlier work

Now,

Page 35: Presentation l`aquila new

Path loss data

• With the formula, we calculate the path loss between two nodes.

• With this data, we explicitly set our path loss map. (its like a matrix, representing the path loss values between the nodes on the network).

• This is done through the SN.wirelessChannel.pathLossMapFile parameter, in Castalia

• Example : 0>1:56,2:40,3:59,4:54,5:58

• This means that when node 0 is transmitting, node 1 is experiencing 56dB path loss, node 2 is experiencing 40dB loss, node 3 a 59dBm loss, etc.

Page 36: Presentation l`aquila new

Wireless sensors networks for health care.

Page 37: Presentation l`aquila new

And more….

Page 38: Presentation l`aquila new

Sprinkler

Temp

Sprinkler

Temp

Smoke

Smoke

Temp

Smoke

Future work: optimal deployment

• Which one is the optimum deployment to improve the network lifetime?

Page 39: Presentation l`aquila new

Sprinkler

Temp

Sprinkler

Temp

Smoke

Smoke

Temp

Smoke

Future work: optimal deployment

• Here we avoid obstacles but nodes must act as routes

Page 40: Presentation l`aquila new

Castalia

A simulator for Wireless Sensor Networks and Body Area Networks, Partly enabled by OMNeT++

For Testing Distributed algorithms, Protocols @ realistic node behaviour, especially relating to access the radio.

Main features include : Advanced channel model Advanced radio model Extended sensing modelling provisions Node clock drift MAC and routing protocols available.

Designed for adaptation and expansion

Page 41: Presentation l`aquila new

Hierarchical relations

Page 42: Presentation l`aquila new

Grazie mille!