distributed intelligent systems – w07: an introduction to ......wireless sensor networks: – are...
TRANSCRIPT
Distributed Intelligent Systems – W07:An Introduction to Wireless
Sensor Networks from a Distributed Intelligent Systems Perspective
1
Outline
• Wireless Sensor Networks (WSN) as a special class of DIS
• Motivating applications• Technology• Tools used in this course
– Mica-z– 802.15.4 radio for e-puck robots– Webots extensions
• Closing the loop with multi-robot systems– Collective decisions as a benchmark– Multi-level modeling for WSN
2
A Special Class of Distributed Intelligent
Systems: Wireless Sensor Networks
3
WSN and DISWireless sensor networks:
– are spatially distributed systems– exploit wireless networking as main inter-node interaction
channel– typically consist of static, resource-constrained nodes– energy saving is a crucial driver for the design of WSN– have nodes which can sense, act (typically no physical
movement), compute and communicate in an unattended mode
Are WSN a special class of Distributed Intelligent Systems?
4
WSN and DISThe potential is there but currently we observe:- Limited embedded intelligence/adaptation:
- sensing data are typically only collected for a particular application and rarely used to control node actions: WSN are typically data agnostic!
- no emphasis on local real-time perception-to-action loop- activity pattern (sensing, computing, networking) are typically
a priori scheduled- static nodes face lower unpredictability than mobile ones
- Limited control distributedness- the fact that WSN are spatially distributed does not necessarily
mean distributed control: the existence of a sink allows for centralized control which in turn often promote energy saving 5
WSN vs. Networked Multi-Robot Systems
Networking is common, sensor nodes = mobile robots without self-locomotion capabilities or mobile robots = sensor nodes with self-locomotion capabilities. So minimal difference? Not really …
• Mobility changes completely the picture of the problem: more unpredictability, noise, … .
• Self-locomotion even more so: real-time control loop at the node level + energy budget breakdown radically different
• Typically different objective functions and performance evaluation metrics
6
Motivating Applications
7
Motivation
• Micro-sensors, on-board processing, and wireless interfaces all feasible at very small scale– can monitor
phenomena “up close”
• Will enable spatially and temporally dense environmental monitoring
• Will enable precise, real-time alarm triggering
• Embedded networked sensing will reveal previously unobservable phenomena
Seismic Structure response
Contaminant Transport
Marine Microorganisms
Ecosystems, Biocomplexity
Adapted from D. Estrin, UCLA8
Pioneering deployments –The WISARD Project
(Flikkema, NAU, 2001 -)
– Microclimate measuring in the Redwood forest
– Impact of fine-scale ecological disturbances on diversity
– Micro-measurement of energy, water, carbon fluxes
9
Pioneering deployments –Great Duck Island
• originally a 9 month deployment (2002)• 32 nodes: light, temp, humidity, barometer• in 2003, added ~200 nodes with various
sensors; still active– www.greatduckisland.net
10
Application 1 - Permasense
• What is measured:– rock temperature– rock resistivity– crack width– earth pressure– water pressure
Pictures: courtesy of Permasense11
Application 1 - Permasense
• Why:“[…] gathering of
environmental data that helps to understand the processes that connect climate change and rock fall in permafrost areas”
Pictures: courtesy of Permasense12
Application 2 - GITEWS
• What is measured:– seismic events– water pressure
Pictures: courtesy of Deutsches GeoForschungsZentrum (GFZ)
German Indonesian Tsunami Early Warning System
13
Application 2 - GITEWS
• Why:To detect seismic events which could cause a Tsunami. Detect a Tsunami and predict its propagation.
Pictures: courtesy of Deutsches GeoForschungsZentrum (GFZ) 14
Application 3 - Sensorscope
• What is measured:– temperature– humidity– precipitation– wind speed/direction– solar radiation– soil moisture
Pictures: courtesy of SwissExperiment 15
Application 3 - Sensorscope
Pictures: courtesy of SwissExperiment
• Why:Capture meteorological events with high spatial density.
16
Application 4: Ecosystem MonitoringScience• Understand response of wild populations (plants and animals) to habitats
over time.• Develop in situ observation of species and ecosystem dynamics.
Techniques• Data acquisition of physical and chemical properties, at various
spatial and temporal scales, appropriate to the ecosystem, species and habitat.
• Automatic identification of organisms(current techniques involve close-range human observation).
• Measurements over long period of time,taken in-situ.
• Harsh environments with extremes in temperature, moisture, obstructions, ...
Source: D. Estrin, UCLA17
WSN Requirements for Habitat/Ecophysiology Applications
• Diverse sensor sizes (1-10 cm), spatial sampling intervals (1 cm - 100 m), and temporal sampling intervals (1 ms -days), depending on habitats and organisms.
• Naive approach → Too many sensors →Too many data.– In-network, distributed information processing
• Wireless communication due to climate, terrain, thick vegetation.
• Self-Organization to achieve reliable, long-lived, operation in dynamic, resource-limited, harsh environment.
• Mobility for deploying scarce resources (e.g., high resolution sensors).
Source: D. Estrin, UCLA18
Enabling Technologies and Challenges
19
Enabling Technologies
Embedded Networked
Sensing& Actuation
Control system w/small form factoruntethered nodes
Exploitcollaborativesensing, action
Tightly coupled to physical world
Embed numerous distributed devices to monitor and interact with physical world
Network devices to coordinate and perform higher-level tasks
Exploit spatially and temporally dense, in situ, sensing and actuation
Source: D. Estrin, UCLA20
Sensor Node Energy Roadmap
2000 2002 2004
10,000
1,000
100
10
1
.1
Aver
age
Pow
er (m
W)
• Deployed (5W)
• PAC/C Baseline (.5W)
• (50 mW)
(1mW)
Rehosting to Low Power COTS(10x)
-System-On-Chip-Adv Power ManagementAlgorithms (50x)
Source: ISI & DARPA PAC/C Program
21
Communication/Computation Technology Projection
Assume: 10kbit/sec. Radio, 10 m range.
Large cost of communications relative to computation continues
1999 (Bluetooth
Technology)2004
(150nJ/bit) (5nJ/bit)1.5mW* 50uW
~ 190 MOPS(5pJ/OP)
Computation
Communication
Source: ISI & DARPA PAC/C Program
22
Free Space Path Loss
• Signal power decay in air:
• Proportional to the square of the distance d• Proportional to the square of the frequency f
– high frequency = high loss– low frequency = low bandwidth
23
Generalization: Friis Laws
• Basic Friis law (open environment) Pr = received powerPt = transmitted powerGt = gain transmitting antennaGr= gain receiving antennaλ = signal wavelength R = distance emitter-receiver
• Modified Friis law (cluttered, urban environment)
n between 2 and 5!
f = c/λ!
24
Communication Power Budget: Not only Transmission
• In general transceiver power consumption dominated by listening (radio on)
• ChipCon CC2420: 19.7 mA in receiving, 17.4 mAtransmitting @ 0 dBm (2.4 GHz)
• SemTech SX 1211: 3 mA in receiving, 25 mA @ +10 dBm in transmitting (900 MHz)
• Synchronization is key• Low power listening protocols are key for
saving power 25
Hardware and Software Modules used in this
Course
26
MICA mote family
• designed in EECS at UCBerkeley• manufactured/marketed by Crossbow
– several thousand produced– used by several hundred research groups– about CHF 250/piece
• variety of available sensors
27
MICAz
• Atmel ATmega128L– 8 bit microprocessor, ~8MHz– 128kB program memory, 4kB SRAM– 512kB external flash (data logger)
• Chipcon CC2420– Respect 802.15.4 at the physical/MAC layer and
therefore can support Zigbee-compliant stacks • 2 AA batteries
– about 5 days active (15-20 mA)– about 20 years sleeping (15-20 µA)
• TinyOS 28
Sensor board
• MTS 300 CA– Light (Clairex CL94L)– Temp (Panasonic ERT-J1VR103J)– Acoustic (WM-62A Microphone)– Sounder (4 kHz Resonator)
29
TinyOS: description
• Minimal OS designed for Sensor Networks• Event-driven execution• Programming language: nesC (C-like syntax
but supports TinyOS concurrency model)• Widespread usage on motes
– MICA (ATmega128L)– TELOS (TI MSP430)
• Provided simulator: TosSim30
31
The epuck 802.15.4 Radio
• Custom module designed specifically for short range• Software controllable (~5cm-5m)• Radio stack implemented in TinyOS (not fully ZigBee
compliant) • Interoperable with MICAz, etc.
[Cianci et al, SAB-SRW06] 31
802.15.4 / Zigbee
• Emerging standard for low-power wireless monitoring and control– 2.4 GHz ISM band (84 channels), 250 kbps data rate
• Chipcon/Ember CC2420: Single-chip transceiver– 1.8V supply
• 19.7 mA receiving• 17.4 mA transmitting
– Easy to integrate: Open source drivers– O-QPSK modulation; “plays nice”
with 802.11 and Bluetooth32
Comparison to other standards
33
Communication Plug-In for Webots
342008-11-06 : Cianci
• OmNET++ engine• OSI framework• Custom Layers
- 802.15.4 - ZigBee
• Physical communication model:- semi-radial disk with noise- channel intensity fading
34
An Example of Mobile Sensor Network:
The Wireless Connected Swarm
35
The Wireless Connected Swarm• Idea: using the communication channel as a crude
binary sensor (“I can communicate” or “I cannot communicate”)
• Two algorithms – α-algorithm: maintain the number of direct connections
around α (parameter)– β-algorithm: maintain a minimal node connectivity
regulated by β (parameter)• Add the possibility of sensing an environmental cue
(e.g., light) to modulate the individual nodes’ β parameter and generate heterogeneity in the swarm and therefore targeted movement (indirect distributed taxis) 36
The α-algorithm
37
In simulation: • radial disk communication• Proximity sensors for robot
avoidance• Unbounded arena
A: connected robots, different headingB: unconnected robotsC: 180º turns for reacquiring connection D: new random heading
B+C+D A
The α-Algorithm: Microscopic Simulator
38[Nembrini et al, SAB 2002]Note: fragility of the algorithm in maintaining connectivity
The α-Algorithm: Webots and Real Robots
39[Pereira et al, IROS 2013]
The β-algorithm
40β = 1 β = 4
• Connection recovery (coherence maneuver) as before but based on a different communication-based perceptual input
• If a robot (A) lose the connection with another specific node (B), it looks of how many of its neighboring nodes (C and D) still have this specific node (B) as neighbor
• If this number is less or equal than β than it start a coherence maneuver; once recovered random heading
The β-algorithm: Microscopic Simulator
41[Nembrini et al, SAB 2002]
Note: red robots perceive light and raise their β to infinite
Collective Decisions
42
Collective Decisions
• A general benchmark for testing distributed intelligent algorithms
• Feasible without mobility relying exclusively on networking
43
Modeling
Individual behaviors and local interactions
Global structuresand collective
decisions
• Local rules and appopriate amplification and/or coordination mechanisms can lead to collective decisions
• Modeling to understand the underlying mechanisms and generate ideas for artificial systems
Ideas forartificialsystems
Understanding Collective Decisions
44
Choice occurs randomly
(Deneubourg et al., 1990)
Example 1: Selecting a Path (W1)
45
Example 2: Selecting a Food Source (W1)
46
[Halloy et al., Science, Nov. 2007]
• A simple decision-making scenario: 1 arena, 2 shelters
• Shelters of the same and different darkness
• Groups of pure cockroaches (16), mixed robot+cockroaches (12+4)
• Infiltration using chemical camouflage and statistical behavioral model
• More in week 13
Example 3: Selecting a Shelter• Leurre: European project focusing on mixed insect-robot
societies (http://leurre.ulb.ac.be)
47
Example 4: Selecting a Direction
Converging on the direction of rotation (clockwise or anticlockwise):• 11 Alice I robots• local com, infrared based• Idea: G. Theraulaz (and A. Martinoli); implementation: G. Caprari, W. Agassounon
48
[Cianci et al, SAB-SRW 06]
Set-up and Collective Decision Algorithm
49
Some Results
[Cianci et al, SAB-SRW 06] 50
Alternative Scenario: Networking Sensor Nodes and Robots
[Cianci et al, SAB-SRW 06] 51
522008-11-06 : Cianci
Example 5:Assessing Acoustic Events
• Non-trivial event medium– Unpredictable in both
space & time– Generality
• Applicable to other similar media
– Highly localized• Facilitates experimentation
– No or weak assumptions about the underlying acoustic field
[Cianci et al., ICRA 2008] 52
532008-11-06 : Cianci
The Physical Set-up
• Realization of the general case– 1.5 x 1.5m tabletop arena
• Multiple elements– Nodes (Robots)– Radio– Sound– Independent event source
[potentially mobile]
53
542008-11-06 : Cianci
Submicroscopic Model (Webots)• Realistic simulation in
Webots• Calibrated modules
internal to nodes– e-puck
(wheels, distance sensors,…)
– Sound propagation (Image-Source)
– Radio communication (OmNET++; semi-radial disk with noise, channel intensity fading, OSI layers)
54
Measurement Confidence in Acoustic Event Detection
• In some situations, event detection may trigger a costly process – (i.e. human intervention, fire
brigade, etc…)• A simple consensus mechanism
may help limit false positives– Require k nodes to agree on
detection before reporting– Here, k=2 shown
• Test against “desirable” & “undesirable” event sources of different relative intensities– Decoy = {50%, 75%, 95%}
* Target intensity
Detected by more nodes = louder
– Measuring intensity with a binary sensor
55
56
Hierarchical Suite of Models• Microscopic
• Non-spatial (see also Week 8)• Discrete-spatial• Continuous-spatial
• Submicroscopic (called module-based in ICRA 2008)
A area of interestN number of nodesD(N) distribution of nodesE events in the environmentPdet(r,Ie) probability of detectionPcom(r,It) probability of communication
56
57
Performance Metric
false negatives false positives measurements messages
⋅⋅
−+
⋅⋅
−+
−+
=
msstotfp
fp
totE FTN
PLFTN
SEE
EEEM 1
/1
),max(1),,,( det δγβαδγβα
Edet : the number of events reportedEtot : the total number presentedEfp : the number of false positives reportedN : the number of nodes in the networkS : total number of measurements (whole network)T : length of experiment (time)Fs : sampling frequencyLs : samples per measurementP : total number of messages (whole network)Fm : maximum message transmission rate 57
58
Validation Results• All four modeling levels
presented agree quite closely with the results from the real system– Avg & Std over 20 runs
shown for each:• 100 events (real & module-
based)• 1,000 events (continuous &
discrete spatial)• 10,000 events (discrete
non-spatial)– Additional experimentation
using the models should therefore remain applicable to the target system
( )
−+=
totfp
fp
tot
detE EE
EEEM
,max1
21
210,0,
21,
21
[Cianci et al., ICRA 2008] 58
59
Comparison of Execution TimesModel Speed Factor
Non-spatial Microscopic (Matlab) 90.81xDiscrete Spatial Microscopic
(Matlab)23.27x
Continuous Spatial Microscopic (Matlab)
17.07x
Submicroscopic (C) 1.36xPhysical System 1.0x
Potential 10x for Matlab -> C implementation59
Conclusion
60
Take Home Messages• WSNs represent a very promising technology for a
number of applications• Commonalities and synergies between distributed,
networked multi-robot systems and WSNs are appearing but their potential need still to be further investigated and formalized
• Collective decisions represent interesting benchmarks for testing distributed intelligent algorithms on WSN (static and mobile)
• A first multi-level modeling attempt for WSN has been carried out using a framework similar to that used for swarm robotic systems, an effort potentially allowing for formal investigation of commonalities and synergies of hybrid robotic/static WSNs 61
Additional Literature – Week 7• Permasense http://www.permasense.ch• GITWES – the German Indonesian Tsunami Early Warning System
http://www.gitews.de ftp://ftp.cordis.europa.eu/pub/fp7/ict/docs/sustainable-growth/workshops/workshop-20070531-jwachter_en.pdf
• Sensorscope http://www.sensorscope.ch/• Mobicom 02 tutorial:
http://nesl.ee.ucla.edu/tutorials/mobicom02/• Course list:
http://www-net.cs.umass.edu/cs791_sensornets/additional_resources.htm• TinyOS:
http://www.tinyos.net/• Smart Dust Project
http://robotics.eecs.berkeley.edu/~pister/SmartDust/• UCLA Center for Embedded Networking Center
http://www.cens.ucla.edu/• Intel research Lab at Berkeley
http://www.intel-research.net/berkeley/• NCCR-MICS at EPFL and other Swiss institutions
http://www.mics.org62