Proactive computing David L' Tennenhouse

1
Proactive computingDavid L. Tennenhouse

• 2003. 06. 14
• Inho Park
• (happyeno_at_iat.chonnam.ac.kr)

2
Contents

• Introduction
• What Should We Do Differently?
• Why now?
• Lets Get Physical
• Aggregating Nodes into Systems
• Lets Get Real
• Lets Get Out
• Lets Reinvent Computer Science
• Conclusion

3
Introduction

• For the past 40 years, most of the IT research
  community has focused on interactive computing(
  J.C.R. Licklider )
• The number of networked interactive computers
  will surpass the number of people on the planet
• Beyond interactive computing

4
What Should We Do Differently?(1)

• First steps toward a new agenda must include
• A fundamental reexamination of the boundary
  between the physical and virtual worlds
• Movement from human-centered to human-supervised(
  or even unsupervised ) computing
• Three loci for new research activities
• Getting physical
• Getting real
• Getting out

5
What Should We Do Differently?(2)

• Getting physical
• Proactive systems use sensors and actuators to
  both monitor and shape their physical
  surroundings
• Research into getting physical explores the
  pervasive coupling of networked systems to their
  environments

6
What Should We Do Differently?(3)

• Getting real
• Proactive computers will routinely respond to
  external stimuli at faster-than-human speeds
• Research in this area must bridge the gap between
  control theory and computer science
• Getting out
• Shrinking time constants and sheer numbers demand
  research into proactive modes of operation in
  which humans are above the loop

7
What Should We Do Differently?(4)

• Two simple reasons why we should divert some of
  our intellectual resources to proactive computing
• The vast majority of new computers will be
  proactive
• These nodes will be the principal sources and
  sinks of information

8
What Should We Do Differently?(5)

• As shown in Figures 1 and 2, the vast majority of
  these devices will be embedded in other objects
• They will be in direct contact with their
  environments rather than with human beings
• Able to monitor and shape the physical phenomena
  that surround them

9
Why now?(1)

• Although Broadening of the Internet will continue
  a an impressive rate, The number of nodes can be
  increased an additional 50-fold by reaching down
  into all of the embedded devices at each
  geographic location
• The lack of network connectivity has stranded the
  data obtained by embedded processors
• Various efforts are beginning to unlock the
  information derived by sensors and remote access
  to their actuators

10
Why now?(2)

• Isnt this the same as ubiquitous computing?
• Computation would become so ubiquitous, Mark
  Weiser
• Proactive computing represents an agenda for the
  exploration of the upper-right quadrant of
  ubiquitous computing
• Manual vs. Autonomous()
• Office vs. Field()

11
Lets Get Physical(1)

• Herbert Simon identified the importance of
  bridging the physical and virtual domains
• Has been ignored by the mainstream artificial
  intelligence and computer science communities
• Research has been human-centered May have blinded
  us to many opportunities beyond interactive
  computing

12
Lets Get Physical(2)

• Following paragraphs address some of the
  node-level challenges and opportunities
  associated with getting physical
• Sensors and actuators
• Inexpensive network connectivity
• Sample-friendly microarchitectures
• Operating system implications

13
Lets Get Physical(3)

• Sensors and actuators
• One of the distinguishing aspects of proactive
  node
• MEMS( Micro Electro Mechanical System )
• Small and inexpensive sensor
• May lead to the first breakthroughs in actuator
  technologies

14
Lets Get Physical(4)

• Inexpensive network connectivity
• Challenge is the integration of network
  connectivity that is inexpensive by present
  standards and tailored to operate in environments

15
Lets Get Physical(5)

• Sample-friendly microarchitectures
• Proactive computing will leverage new
  architectures that streamline the processing of
  the information acquired at individual node
• Diverse forms of sampled information
• The Berkeley Intelligent RAM

16
Lets Get Physical(6)

• Operating system implications
• Sample-friendly processor architectures will
  enable the bulk, if not all, of the application
  software to be temporally decoupled from the
  physical world
• This decoupling implies the need for operating
  system mechanisms that virtualize the incoming
  and outgoing sample streams and make them
  available to user-mode application software

17
Aggregating Nodes into Systems(1)

• Leveraging large numbers of computers will demand
  new systems and networking technologies that
  aggregate and share their capabilities
• Traditional sensor networks are dedicated to a
  single application and organized hierarchically

18
Aggregating Nodes into Systems(2)

• However, proactive sensor networks might be based
  on very different concepts, including
• Sensor multiplexing
• Not specific users and application but
  network-based service( authorized users)
• Scaling to large numbers of potential users is
  likely to depend on intelligent multicasting
  technology
• Inverse and peer tasking
• Sensors master-slave environment
• Mobile code

19
Aggregating Nodes into Systems(3)

• Querying and fusion of real-time observations
• Sensors of many different types
• Capabilities, configuration, and location of
  sensors will be dynamic
• Querying
• Decomposition and Fusion
• Sample provenance
• For security, legal, and Scientific purposes, We
  need not sampled information but original
  information
• Working backward through any fusion
• Processing mechanisms to the initial point of
  acquisition of the information

20
Lets Get Real

• Proactive environments operate at
  faster-than-human speeds( major change )
• Ex) automobile antilock braking system
• Software-enabled control
• ( software-friendly approach)
• Network-enabled control( packet-based network )
• Online measurement and tuning

21
Lets Get Out(1)

• The sheer numbers of networked computers will
  preclude human cognizance at the level of
  individual nodes
• Getting humans out of the loop suggests the need
  for work in three addition research spaces user
  interfaces, software, reuse

22
Lets Get Out(2)

• User vs. supervisory interfaces
• Faster, how are things different?( proactive
  computing )
• Software creation
• Will be necessary to reduce human involvement in
  the software-creation process
• A few good abstractions
• Reuse software, much simpler and more fundamental
  is long

23
Lets Reinvent Computer science(1)

• Cross-cutting opportunities ripe for
  investigation include
• Active technologies supporting code mobility
• Leveraging large number
• Number of users and nodes
• Amount of current information
• Number of historical information
• Scale of the problem and solution spaces being
  explore

24
Lets Reinvent Computer science(1)

• Reinventing engineering
• Safety, social, and ethical issues ex)
  authority
• Curriculum development academic program

25
Conclusion

• Over the past 40 years, computer science has
  addressed only about 2 of the worlds computing
  requirements
• Its time to get physical, get real, and get out
  and build proactive systems