Transcript 3.0M PowerPoint Presentation

A Al-Anbuky, H Sabat, M I Rawi & S Sivakumar
SeNSe Research Centre
http://SenSe.aut.ac.nz
AUT University, Auckland
Presentation Overview
 Info on the upcoming Co-Located ICT conferences 2010
 SeNSe lab research overview
 Human Comfort and passive house research
 Wildlife Sensor network and network connectivity research
 Data stream mining research
2010 Co-located ICT conferences
31 Oct – 3 Nov Auckland NZ
http://APCC2010.aut.ac.nz
2010 Co-located ICT conferences
31 Oct – 3 Nov Auckland NZ
http://APCC2010.aut.ac.nz
Empowering global connectivity
Today we are confronted by global challenges such as climate
change, resource consumption, environmental stress and population
health. In responding to these challenges engineers are recognising
the increasing importance of communications and connectivity.
Sensor networks provide unprecedented volumes of information
about our environments. Wireless and fixed communications
networks facilitate the sharing of this information. Intelligence and
cognition enable the efficient use and management of our resources.
Meanwhile,
humans
and
devices
demand
increasing
communications connectivity and systems interoperability. Under
the theme of "Empowering global connectivity", APCC 2010
provides a forum for researchers and engineers in the Asia-Pacific
region to present and discuss topics related to advances in
information and communication technologies, while encouraging
collaboration and innovation that may help in saving the planet.
Partnership & Fund Raising
Available opportunities varies from $2.5k to $20k
Sponsors privileges could include
 Seats for membership within organizing committee
(Key sponsors only)
 Free seats for conference Registration (sponsorship
dependent)
 Logos on CFP, Conference web site and proceeding
 Listed as sponsor within the proceedings
The Venue
The Venue
SeNSe Lab -AUT
 Wildlife Cognitive Sensor Network
 Mobile subjects localization
 Connectivity & opportunistic networks
 Wildfire hazard detection
 Hunters friendly fire avoidance
 Data stream mining & network energy efficiency
 Object Centric Ambient Intelligence
 Human comfort & passive home ambient intelligence
 Thermal mapping & food property dynamic tracking
 pH sensor network & red meat tenderization
 Vehicular Communication
 Train localization & railway signalling system
 Microwave Sensing
 Timber property mapping
 Distributed Signature Analysis
 Power System fault detection
MohD Izani Rawi
SeNSe Lab
AUT University
Overview
 Passive House System Overview
 Architecture Overview
 Passive House System Manager
 Thermal Comfort
 Human Centric Thermal Comfort Concept
 Thermal Comfort Operation
 Thermal Comfort Simulation
 Thermal Comfort Result
 Discussion & Further Work
Passive House System Overview
Architecture Overview
Sensors / Actuators
(Location, appliances,
environmental)
Going home
Mobile Device (ID)
•Going home
•Mobile Device Notify Home
•Personalise home environment
•Learn occupant behaviour
•Adaptation & personalisation
Human Centric Activity – Automation, personalisation, adaptation
Passive House System Overview
Heating / Cooling
 Passive House System Manager
Actuator
Window Position
Control
Thermal
Visual
Air
Shading Position
Illuminance Level
PH Manager
Occupant
Preferences
Energy
Usage
Heating / Cooling
Hot Water
Appliances
Ventilation
Human Comfort
PMV
Light
Thermal
Comfort
Visual
Comfort
Ta, MRT, RH, Vel
Clo, Met
Illuminance Level
Shading Level
AQ
Indoor Air
Comfort
CO2 Concentration
Noise
Acoustical
Comfort
Sound Level
Spatial
Comfort
Passive House System Overview
Thermal Comfort
M: metabolism
W: external work, equal to zero for most activity
Icl: thermal resistance of clothing
fcl: ratio of body’s surface area when fully clothed to body’s
surface area when nude
PMV Value
ta: air temperature
+3
tr: mean radiant temperature
+2
Va: air velocity
+1
Pa: partial water vapour pressure
0
hc: convectional heat transfer coefficient
-1
tcl: surface temperature of clothing
-2
-3
Meaning
Hot
Warm
Slight Warm
Neutral
Slight Cool
Cool
Cold
Human Centric Thermal
Comfort Concept
 Thermal Comfort Operation
 Single Node PMV Calculations
 Tested on Sun SPOT wireless platform
 SeNSe lab air temperature & PMV
Human Centric Thermal Comfort
Concept
 Thermal Comfort Simulation
 PMV of a Given Living Space
 Inverse Distance Weighted (IDW) interpolation
technique
Human Centric Thermal Comfort
Concept
 Thermal Comfort Results
Nodes
Thermal Comfort
Parameters
N1
N2
DBT
24
23
MRT
27
25
RH
50
Vel
Met
Clo
N
3
2
1
N4
N0
22
22.50
2
0
21
23.41
57
6
0
54
55.94
0.1
0.1
0
.
2
0.2
0.14
-
-
-
PMV at N0 =
1
1
-0.16
Sivakumar Sivaramakrishnan
SeNSe Lab
AUT University
Connectivity Issues in Wildlife
Monitoring
 Short Range Nodes
 Network is Adhoc
 Network Holes (region of no connectivity)
 Movement results in Temporary Connectivity
 Node Discovery
Varying Node Density
 Animals have different habitat
 This determines the grouping of the nodes
Varying node Density Depending on Animal Habitat
Due to connectivity holes data transmission is opportunistic.
Adaptive Opportunistic Connectivity
Opportunistic Networking
• Data Hand-off Mechanism
• Adaptive Node Discovery
• Doppler Shift to Detect Direction
•
A
B
E
C
D
Hand-Off under Random
motion of the animal
F
Energy Dissipation for
Connectivity with and without
Hand-off
Preliminary Results
Due to Predictive Sampling:
Fig. shows adaptive sampling
saves on energy as the number
of unsuccessful searches are
less
Due to Hand-off:
Fig. shows the energy
consumption due to Handoff scheme is less than
without hand-off
Hakilo Sabit
SeNSe Lab
AUT University
Sensors data streams
 A data stream can be roughly
thought as an ordered sequence of
items, where the input arrives more
or less continuously as time
progresses.
 Examples of data streams include
computer network traffic, phone
conversation, Web searches, Sensor
data and etc.
 Data streams are characterised by
continuous flow of data with infinite
length.
Data stream processing
 Sensors deployed for
monitoring application (ex.
traffic flow monitoring,
environmental monitoring,
patient health monitoring)
produce data with such (data
stream) characteristics.
 Data steams generate large
quantity of real-time/near
real-time data (structured
records).
 The stream processing has to
be done in real-time or near
real-time and in bounded
storage.
WSN stream mining
 WSN are know for their limited resources (storage, processing and energy).
 High resolution sensor data streams contain useful information
 excellent environment for data mining
 Fuzzy logic based distributed stream clustering algorithm (SUBFCM)
 designed and optimised for WSN environment
The SUBFCM algorithm
 SUBFCM compute local clusters at
designated GH nodes and only transmit
the local representatives- Reduced data
bits to transmit means energy saving,
besides bandwidth efficiency.
 Based on single scan of data items to
extract the representative patterns & no
intermediate data stored - memory
Data processing
scalable.
centre
• SUBFCM compute the complete cluster at
a central location based on the local
representatives
•Stream modelling results will generate a
control signal for the local nodes to adjust
their parameters
Residents
Internet
Fire
Department
Local
Industry
Sink
Group
head
Sensor
Preliminary Results
90
-3
1.4
Data reduction
fcm, multi-hop fcm and subfcm algorithms
x 10
85
hot spot 1
1.2
Relative Humidity [%]
80
Energy consumption
0.8
-->fcm
0.6
75
hot spot 2
70
65
multi-hop-->
0.4
->q=70
hot spot 3
subfcm-->
60
0.2
0
0
20
40
60
Total Distance [m]
80
100
120
Cluster accuracy vs central algorithm
55
24
25
26
27
Temperature [oC]
28
29
30
Cluster accuracy vs fcm algorithm
Temp Erorr
Average Temp Error
Temp Error
Average Temp Error
RH Erorr
Average RH Error
RH Error
Average RH Error
25
10
20
8
15
6
Error
Error
Total Energy [Joules]
1
10
4
2
5
0
0
1
3
5
7
9
11
13
15
17
No. of runs
19
21
23
25
27
29
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
No. of runs
Happy to Talk
http://SeNSe.aut.ac.nz