Transcript Ubicomp

Pervasive & Ubiquitous
Computing
Hao Chu (朱浩華)
Lecture #2
2/23/2004
1
Administrative Announcements
• Course homepage is up.
– http://www.csie.ntu.edu.tw/~hchu/ubicomp_course/ubi
comp_course.htm
– Reading list, project overview, project ideas, …
• Course WIKI page is up.
• “How to use WIKI?” by James
2
Vision and Challenges
• M. Weiser, “The Computer for the 21th Century",
Scientific American, September 1991.
• M. Weiser, “Some Computer Science Issues in
Ubiquitous Computing”, Communications of the ACM,
36(7):75-85, July 1993.
• M. Weiser, J. S. Brown, "The Coming Age of Calm
Technology“, 1996.
• M. Satyanarayanan. "Fundamental Challenges in Mobile
Computing", Fifteenth ACM Symposium on Principles of
Distributed Computing, May 1996.
• M. Satyanarayanan. “Pervasive Computing: Vision and
Challenges”, IEEE Personal Communications, August,
2001.
3
M. Weiser. “The Computer for the
21th Century", Scientific
American, September 1991.
4
Two Key Points
• Ubiquitous computing is about computing so
well integrated with our physical environment
such that people fail to take notice of them.
– Computing being everywhere, yet fading into the
background
– Computing becoming disappearing and invisible
• Location and scale are important issues.
– Adapt their behavior intelligentlly without complex AI
– Scale suitable to different tasks
5
Examples of Disappearing
Technologies
•
•
•
•
Writing
Electric Motors
Micro-controllers in CEs
They are everywhere (embedding & hidden into
physical objects), yet invisible.
• When a technology matures, they disappear!
– Maturity = cheap, small, widely applicable, good
usability, …
6
Why Good Technology Is Invisible?
• “Invisible” stays out of the way of task
– Like a good pencil stays out of the way of the writing
– Like a good car stays out of the way of the driving
• Bad technology draws attention to itself:
– Like a broken, or skipping, or dull pencil
– Like a car that needs frequent tune-ups
• Computers are mostly not invisible
– They dominate interaction with them.
• Ubicomp is about making computers invisible.
7
Ubicomp vs. Virtual Reality
• Should we live in virtual computing world? Or
should computing come out and live in our
physical world?
• VR is about simulating physical world & putting
people inside virtual computing world.
– Limited applications & activities
• Ubicomp is about bringing computing to people’s
physical world.
– Integrating with everyday objects and activities
8
Ubicomp vs. Multimedia
• Multimedia grabs user attention for
entertainment purpose.
• Ubicomp reduces user distraction, allowing
people to focus on tasks.
9
Intelligence
• Computing embedded and enhancing physical
objects
•
physical objects
A
c h
ie v e
in
t e llig
e n
c e
t h
r o
u
g
h
in
t e r c o
n
n
e c t io
n
o
f
– Wall talking to heating or projector
• Achieve intelligence through location awareness
(without AI)
– Automated call forwarding, lighting control
10
PARC Ubicomp Work (1991)
• Focus on devices that
transmit & display information.
• Device scale targets different
tasks.
– Consider three sizes: tabs, pads,
boards.
11
Tabs
• Inch-scale Ubicomp devices
– Post-It notes
• Carried around by a person
• Hundreds in a room
–
–
–
–
–
–
Credit cards, ID cards
Remote controllers
Badges
Tags / Labels (RFID)
Locating system (tags as library catalogs)
Animate static physical objects (active calendar, active
map)
12
Pads
• Foot-scale Ubicomp
devices
– A sheet of paper / tablet PC
• Tens in a room
– Like scrap papers that can
be grabbed and used
anywhere, no unique ID.
• Like windows in Apple
Macintosh, but can spread
them out on a real desk.
13
Boards
• Yard-scale Ubicomp device
• One in a room
• White board with e-chalk
– Shared white board with remote participants
• Video screen
• Electronic Bookcases
14
Hardware Challenges (1991)
•
•
•
•
•
•
High resolution flat panel display (cheap)
High speed processor
High capability storage
High bandwidth wireless Network
Lower power consumption
How well do today’s HW technologies meet
these challenges?
15
Software Challenges (1991)
• Dynamic configuration of HW/SW in ubicomp
environments (device mobility)
• Application migration across heterogeneous
ubicomp environments (user mobility)
• Transparent linking of wired and wireless
networks
– Tiny range wireless, long range wireless, high speed
wired
• How well do today’s SW technologies meet
these challenges?
16
Sal Scenario
• Proactively brew coffee
• Electronic Trails of
neighbor coming and
going (Privacy issue)
• Automatic recording pen
• Email locating garbage
door opener
• Window tells weather
• Share location info with
Joe (Privacy Issue)
• Share tabs/pads with Joe
(Miniaturization)
• Gesture to project
blinking tab to projector.
• Memory augmentation on
meeting with Mary.
17
Privacy Issue
• Hundreds and thousands of invisible computers
sensing and watching people
– A bit scaring?
18
Misc
• Ubicomp as a Savor to Healthy Social
Interaction?
– People don’t have to hole up in windowless offices
before computer screens all days.
• Ubicomp as relaxing as “talking a walk in
the woods”
19
Discussion
•
•
•
•
Tradeoff between intelligence and privacy
Tradeoff between invisibility and privacy
What are killer ubicomp applications?
Think about interesting scenarios for interconnecting
home appliances
• AI & ubicomp?
• Should computing stay out of the physical world?
– Human dependency on computing
– Environmental impacts
– Social impacts
20
M. Weiser. “Some computer
science issues in ubiquitous
computing.”Communications of
the ACM, 36(7):75-85, July 1993.
21
Key Point
• Based on their PARC experiments with tabs,
pads, and boards, this paper tries to define
some ubicomp challenges and where ubicomp is
going.
22
Ubicomp as Experimental CS
• Construct working prototype
• Evaluate working prototype in everyday use.
• Importance of “working prototype”
23
Hardware Prototype Issues
• Power consumption: impossible to change
batteries to many ubicomp devices frequently.
• Balance of HW/SW feature: display, network,
processing, memory, storage capability,
multitasking, QoS, etc.
• Ease of expansion & modification (integration vs.
modular design)
24
Hardware Issues
• Low-power computing
• High-speed wireless bandwidth (802.11g 54
Mbps)
• Pens for very large display
25
Network Issues
• Wireless Media Access (802.11, Bluetooth,
Cellular Networks)
• Quality of Services (RSVP, etc)
• Ubicomp devices changing network attachment
(Mobile IP)
26
Interaction Substrate Issues
• Interaction substrates are UI software for pens
and screens.
– Handprinting recognition
– Voice recognition
– Display migration (follow-me display)
27
Application Issues
• “Applications are of course the whole point of
ubiquitous computing”.
• Locating people (active badges)
– Automated call forwarding
– Tracking down people for meeting
– Watching general activity in a building (feel in touch
with surrounding environments)
• Shared drawing in virtual meeting
– Scalability to 5000 peoples (multicast for bandwidth
efficiency)
28
Location Privacy
• Centralized location database as one possible
solution, but not scalable, single point of attack,
one break-in reveals all.
• Move toward more distributed approach.
29
Computational Method Issue
• Due to unpredictable network to ubicomp
devices, (file) caching can be used to improve
performance.
30
Discussion
• Open ….
31
M. Weiser, J. S. Brown. "The
Coming Age of Calm
Technology." 1996.
32
Key Points
• The computing trends is moving toward many
(hundreds of) computers sharing each of us.
Since computers are everywhere, they better
stay out of our ways. Weiser called it Calm
technology.
• A proposed solution (Calm technology) is to
place most information in periphery of our
attention, but allow fast & easy moving to center
of our attention.
33
34
Computing Trends
These trends are results of computing getting
smaller, faster, and cheapers.
• Mainframe Era: many people sharing a
computer.
• PC Era: one computer per person
• Internet Era: (interconnecting PCs)
• Ubicomp Era:
– many computers (everyday objects) sharing one
person
– interconnecting everyday objects
35
Ubicomp Era
• Cheap “Thin servers” in every household
appliances.
• Average home appliances (Microwaves, TV,
DVD players, etc.) have embedded processors,
but they are not Ubicomp devices, because
– They are not networked.
– They are not connected to the Internet.
36
Calm Technology
• Calm and uncalm technology differs how to
engage our attention.
– Divide our attention into two parts: periphery and
center.
• Periphery is informing without overburdening
– E.g., driving a car, center = roads and radio, periphery
= engine noise
• Calm technology can move easily & quickly
between periphery and center.
37
Examples of Calm Technology
• Inner Office Windows
– Extend periphery to what are going on the hallway
– E.g., notice lunch gathering, meeting, but not
distracting to work
• In comparison to open cubicles with low
partitions
– Force too much to the center
– E.g., noises in the hallway can become distracting to
work
38
Dangling String
• Represent bits flowing
over a wire through
motion and sound.
• The output is
(beautifully?) integrated
with our physical
environment.
39
Discussion
• Is Ubicomp defined by the devices or user
experience?
• Are inner office window and dangling string
really calm and relaxing?
• Does Calm give more or less information?
• This is related to ambient display research. UI
People know more about this research area?
40
M. Satyanarayanan.
"Fundamental Challenges in
Mobile Computing", Fifteenth
ACM Symposium on Principles of
Distributed Computing, May 1996.
41
Key Points
• What are fundamental challenges in mobile
computing (What is so special about mobile
computing)?
–
–
–
–
Resource-limited mobile devices
Mobility is hazardous
Unpredictable wireless networks
Finite power source
• Adaptation is key to mobility.
42
Evolution from Distributed Systems
to Mobile Computing
• What are fundamental differences between
Distributed Systems and Mobile Computing?
– Distributed systems = Applications running over
Computers Connected Over Network
– Mobile systems = Applications running over mobile
devices connected over wireless networks
– Mobile devices constraints: resource-limited, finite
power source, easy-to-lose (weak security)
– Wireless network constraints: unpredictability
bandwidth and frequent disconnections
43
Adaptation is Key to Mobility
• Adapt applications to continuously changing
computing environment (due to mobility):
– Power, network bandwidth / availability, …
44
Adaptation Strategies
• Need resource management at the system layer to
allocate resources to different applications.
• Need application level semantics, e.g., frame-rate,
resolution, etc.
• The best approach is to have both application and
system supports.
45
Extended Client Server Model
• Traditional (thin) client (fat) server model in distributed
computing needed adjustments to address
unpredictable network, finite power, and performance.
• Extended client-server model (called smart client model)
places some server functionalities to client.
46
Coda & Odyssey
• Coda supports application-transparent
adaptation.
– It is distributed file system (FS).
– It can cache/hoard some parts of FS on client.
– It can support disconnected operations from FS
cache.
• Odyssey supports application-aware adaptation.
– Adapt application quality/fidelity (e.g., video
framework, resolution) based on dynamic network
condition, power saving, and processing loads.
47
Research Topics for Exploration
• Caching metrics (what to cache/hoard?)
• Caching coherence (Semantic callbacks and
validators)
• Algorithms for resource revocation
• Agility (靈敏) vs. stability (Analysis of adaptation)
• (Global) Resource estimation from local
observations
48
M. Satyanarayanan. “Pervasive
Computing: Vision and
Challenges”, IEEE Personal
Communications, August, 2001.
49
Key Points
• What are fundamental challenges in
pervasive/ubicomp computing (What is so
special about pervasive computing)?
–
–
–
–
Effective use of smart spaces
Invisibility
Localized Scalability
Masking Uneven Conditioning
• We will hear a distinguished talk from author
directly.
– http://www.cs.uiuc.edu/news/dls/distlectpst.html
50
Discussion
• What exactly is proactivity? How does it differ
from adaptation?
• Is Ubicomp still about old technical issues ..
– Integrating different technologies, devices and
services.
– Resource discovery protocol
– Optimization (different parameters such as power)
– Adaptation (with environments)
51