Printable Version of this PageHome PageRecent ChangesSearchSign In

Ubicomp readings

As We May Think, Vannevar Bush
BibTex: memex45

This was a visionary paper written in 1945 near the end of WWII. In it, Bush describes the coming (for him) and evolving (for us) digital age. In the article, he floats ideas that come close to the digital camera, digital storage, speech recognition, credit cards, etc. However, the main contribution of the paper is his idea of a "memex". This was a mechanical device that comes close (in concept) to a combination of a computer, bookmarking, and internet databases. The memex was designed to help a person deal with the massive amounts of information accumulated over a lifetime by providing easy, associative (similar to the human mind) access. The "essential feature" of the memex was the "process of tying two items together". In contrast the current (1945) search algorithm was indexing.

The Computer for the 21st Century, Mark Weiser
BibTex: weiser91

Seminal paper. Weiser gives broad overview and definition ubicomp. It also notes a major area of problems/research for ubicomp: context awareness: "If a comptuer knows merely what room it is in, it can adapt its behavior in significant ways..." It presents some early Xerox PARC work in developing and using a ubiquitous computing environment (tabs, pads, and boards). It also gives a vision of the future world through a narrative about "Sal" and use of ubiquitous computing.

Some Computer Science Issues in Ubiquitous Computing, Mark Weiser
Bibtex: weiser93

This paper details Xerox PARC's initial foray into Ubicomp. It describes PARC's system of Tabs, Pads, and Boards in gory, hardware detail. He then goes on to generalize what he saw as some of the unique problems ubicomp would generate in the larger world of computer science. These included hardware issues (low power requirements, pens, and wireless networks), network protocols (wireless cellular protocols and routing protocols), interaction substrates (x-windowing systems, menuing systems), applications (the two examples he gives are location finders (similar to ActiveBadge system) and drawing programs), privacy, and computational methods (optimal cache sharing).

The Coming Age of Calm Technology, Mark Weiser and John Seely Brown
BibTex: calm97

Major points:

  • It defines UbiComp in a historical context:
    1. Mainframe (one computer, many users)
    2. PCs (one computer, one user)
    (Internet and Distributed computing transitioning to...)
    3. UbiComp (many computers, one user) or (many computers, many users)

  • The authors note that ubicomp must become commonplace and unremarkable so that we'll forget it's impact on our daily lives. To do this, they suggest that ubicomp technology must become "calm". They
    define calm technology as tech. which moves from the periphery of our attention to the center and back again while engaging both areas with no disruption. They consider this movement calming for two reasons:
    1. Information is offloaded into the periphery of our senses/attention. This allows the periphery info to be "informing without overburdening".
    2. Bringing periphery information to the center is an act of control over that information.

  • Three signs of calm technology
    1. Easy movement between center and periphery
    2. Increases "peripheral reach" - more details into periphery
    3. "Locatedness" - more connection to the world
  • Examples of calm technology:
    1. Inner office windows (enhance peripheral reach and locatedness - allows for easy movement between periphery and center)
    2. Internet multicast (increases peripheral reach by involving another sense (sight) in interaction)
    3. Dangling String (increases peripheral reach vs. invisible bits

A Survey and Taxonomy of Location Systems for Ubiquitous Computing, Jeffrey Hightower and Gaetano Borriello
BibTex: hightowerlocation

This paper decided underwhelmed me. It presents a generalized taxonomy of location systems and then a survey of the field circa 2001. A loose outline of their taxonomy is:
I.) Location Sensing Techniques
A.) Triangulation
1.) Lateration
a.) Direct
b.) Time-of-flight
c.) Attenuation
2.) Angulation
B.) Scene Analysis
C.) Proximity
1.) Detecting physical contact
2.) Monitoring wireless cellular access points
3.) Observing automatic ID systems
II.) Location System Properties
A.) Physical Position and Symbolic Location
B.) Absolute vs. Relative Position
C.) Localized Location Computation
D.) Accuracy and Precision
E.) Scaling
F.) Recognition
G.) Cost
H.) Limitations

The paper goes on to survey a number of location systems and gives several useful tables and graphs comparing systems.

Privacy as part of taxonomy? Receiver identified as specific ID or not?

Phidgets: Easy Developmennt of Physical Interfaces through Physical Widgets,Saul Greenberg and Chester Fitchett
BibTex: greenbergphidgets

This paper extends the widget metaphor for GUIs to input device. Greenberg introduces "phidgets" which are physical widgets. These physical widgets are designed to be prepackaged electrical components which computer scientists usually do not have the capability to build. They enable rapid prototyping of physical interfaces as widgets do for guis. Without getting into too much detail, they not only packages the physical hardware, but also packaged the communication software and made it easy to use via APIs. They also wrote "drop in" widgets (ActiveX and later Khai and Jason Brotherton extended it to Java) which plug into GUI software (VB, etc) to enable easy programming of these devices.

There could be debate about whether this is an engineering contribution vs. research contribution. While there are a limited number of devices, this paper serves as an ideal proof of concept. Gregory: "Makes (electrically) easy things easy. Context toolkit makes hard things easy."

see also iRoom and iStuff out of Stanford

The Context Toolkit: Aiding the Development of Context-Enabled Applications, Daniel Salber, Anind Dey, and Gregory Abowd
BibTex: contexttoolkit

This paper introduces the idea of a context toolkit designed to make incorporating context in applications easier. It details two widgets "IdentityPresence" and "Activity" and their implementations. It also shows a couple of potential applications using the widgets.

The hardware sensor (mic, rfid tags, etc) are refered to as "Generators", "Interpreters" abstract that raw data into something useful, and "Aggregators" or servers collect the information into larger sets.

This system eventually failed because it is hard to add sensors due to tight coupling between Aggregator and Interpretter. Scaling issues. Given arbitrary set of sensors, code must be added for given config to aggregator and interpretter.

How to evaluate toolkits is a larger meta-problem.

Charting Past, Present, and Future Research in Uniquitous Computing, Gregory Abowd and Elizabeth Mynatt
BibTex: abowdmillenium

This paper details three major research areas for ubicomp in the millenium:
1) Natural Interface computing - often combines multiple "data types" (ex. speech and writing in classroom settings). Errors are a large problem w/ three parts: error reduction, error discovery and error recovery.
2) Context aware computing - context defined as who, what, when, where, why. "Context fusion" involving more than one aspect of context (necessary because of errors and sensor limitations).
3) Capture and Access - capturing snippets of environment and accessing them later. Often want non-sequential access. Privacy concerns.

Using the Experience Sampling Method to Evaluate Ubicomp Applications, Sunny Consolvo and Miriam Walker
BibTex: consolvoESM

This paper shows to two researchers adapted a commong technique in psychology, ESM, for use in evaluating ubicomp applications. The Personal Server team at Intel Research Seattle needed to know what devices were present in the environment that people might want to annex. The eval. team implemented an ESM on a PDA to judge. The paper provides poor details on the exact intent of the study, but provide a useful rubric for determining specifics when implementing an ESM. This paper would be appropriate for people who needed to learn ESM and already have a firm idea of what questions they need to ask their subjects. Also worth noting that they gave their subjects financial incentive to respond to the alerts.

Useful would be a study design showing how to apply their methodology from the ground up. Study explanation was very poor and vague. Hard to tell if this was embarrassment or just oversight.

A Diary Study of Information Capture in Working Life, Barry Brown, Abigail Sellen, and Kenton O'Hara
BibTex: browndiary

This is a slight twist on a diary study. The team needed to evaluate the market for a new device, the "CapShare", a hand-held scanner. They divided people into two groups, "Paper" and "Multimedia" and gave them digital cameras to snap pictures of things they wanted to "record". They told the paper group to only record paper/documents items while they told the other group to record anything/everything. They then interviewed the subjects using the pictures to jog people's memories. They discovered that 35% of the time, people in the paper group wanted to capture something other than letter size documents, meaning their device needed to accomodate post-its and flip charts size things. They also developed an interesting taxonomy which stemmed from what people were trying to capture. In a nutshell, those were:
  1. Capture to discuss
  2. Capture to distribute
  3. Capture to post in a common information space
  4. Capture to archive
  5. Capture to collect and collate
  6. Capture to read and reflect
  7. Capture for task management
  8. Capture to refer to
  9. Capture to re-use
  10. Capture to a living document

They applied this taxonomy to their device to help inform software development and features of the device.

Disappearing Hardware, Roy Want, Trevor Pering, Gaetano Borriello, and Keith Farkas
BibTex: wantdisappearing

Article concentrates on the hardware issues of ubicomp. Makes the case that hardware (he sort of also includes software) should "disappear" from the user and allow users to concentrate on what they want to do rather than the tools needed to do it.

Four most notable improvements in last 10 years:
  1. Wireless networking
  2. Processing capability
  3. Storage capacity
  4. High-quality displays

Trends that will drive ubicomp:
  1. Personal systems (Personal Server, wearables)
  2. Infrastructure systems (Berkley's smart dust/motes)
  3. Proactive interaction (anticipate and react to real world issues w/o users command)
  4. Robotics (this seems kinda thrown in here, and I don't buy it)

Hard problems/Physical limitations:
  1. Size and weight
  2. Energy
  3. User interfaces

Future challenges/research areas:
  1. Hardware issues - how much hardware necessary for future (not now)
  2. Disappearing software
  3. Interaction design
  4. Applications

Interestings stats: PCs make up only 2% of processors sold

Beyond prototypes: Challenges in deploying ubiquitous systems, Nigel Davies and Hans-Werner Gellersen

Makes the case that the whole of a ubicomp application is far greater than each of it's parts (privacy, infrastructure, sensor networks, cost, etc)

Research challenges:
  1. component interac tions - open systems to extend beyond original intended uses
  2. adaptation and contextual sensitivity - large field means many cues/triggers to respond to
  3. appropriate management - system administration over many domains
  4. component association - accurately determine users task & intentions, and develop associations between components of infrastructure to assist
  5. viable economic models - nuff said
  6. UI integration - coordination among many apps with coherent interface and switching between
  7. social, legal, and technical sol'ns to privacy and security

The human experience, Gregory D. Abowd, Elizabeth D. Mynatt and Tom Rodden.
BibTex: abowdhuman

Idea that machines should be about the humans behind them.
Physical interaction issues:
  1. "implicit input"
  2. multiscale/distributed output
  3. integration of physical and virtual worlds

Important themes:
  1. Context aware computing
  2. Automated capture and access
  3. continuous interaction or "everyday computing" - "focus on activities as opposed to tasks"

Design and eval issues:
  1. models of interaction - MHP doesn't work for ubicomp areas (good summary of activity theory, situated action, and dicog)
  2. better understanding of places ubicomp used - ethnographic approach, probes
  3. assessment of use after deployment - how is big question/research area
  4. new measures - new representations of human activity

Stuck in the Middle: The Challenges of User-Centered Design and Evaluation of Infrastructure, Keith Edwards, Victoria Belotti, Anind Dey, and Mark Newman
BibTex: edwardsmiddle

Paper looks at (ubicomp) infrastructures and the struggle between design and evaluation of the infrastructure itself. They detail 8 lessons learned from two case studies: Placeless Documents and the Context Toolkit.
  1. Prioritize core infrastructure features
  2. Build lightweight prototypes that express core objectives of infrastructure
  3. Don't ignore usability and usefulness in test applications
  4. Initial prototypes should be lightweight
  5. Be clear about what your test-application prototypes tell you about the underlying infrastructure
  6. Don't confuse design and testing of infrastructure with providing infrastructure for experimental application developers
  7. Define a limited scope for test applications and permissible uses of infrastructure
  8. No point in faking components and data if you want/need to test user experience benefits

Developing Consumer-Friendly Pervasive Retail Systems, Panos Kourouthanassis and George Roussos

Paper talks about MyGrocer, a ubicomp application for use in grocery stores. Based on RFID for products. Looked at what shoppers would want given a display on their cart that tracked their purchases. Things such as store layout maps, running total of items in cart, expedited checkout, shopping list based on past purchases, etc.

Technology is nothing spectacular, but user studies were good. Started with focus groups and then a limited implementation (2 aisles and about 100 products).

Classroom 2000: An experiment with the instrumentation of a living educational environment, G. Abowd
BibTex: abowdclassroom2k

Details the first few years of Classroom 2000/eclass. Talks about the hand in hand design of system and evaluation of system. Timeline structure.

Initial prototype split into Preproduction, Live Capture, Postproduction, and Access. Helped quantify what were most important features and what should be focused on full time. Transitioned into "living laboratory" which was wider deployment and eval. Found users tended to take less notes but found no differences in attendance or exam results.

Lessons learned:
  1. important to prototype ubicomp apps quickly
  2. structure of system (even if evolving) matters
  3. cost is not a limiting factor
  4. capture is meaningless w/o access
  5. create an open environment which leverages stakeholders into developers
  6. can't build everything in a research environment
  7. understand difference between demonstration prototype and evaluation prototype
  8. valuable to simulate automation with manual effort, but only for awhile

User Study Techniques in the Design and Evaluation of a Ubicomp Environment, Sunny Consolvo, Larry Arnstein, B. Robert Franza

Discusses problems of evaluating ubicomp apps. Shows how different techniques can be used for the design and evaluation sections.
For design, they used Intensive Interviewing (lab setting, open ended interview) and Contextual Field Research (CFR) (following people in environment). II first then leading to CFR. In the eval. they use Lag Sequential Analysis (code things in relation to other events) instead of usability testing.

Used in context of LabScape and a biology lab.

PlantCare: An investigation in practical ubiquitous computing systems, Anthony LaMarca, Waylon Brunette, David Koizumi, Matthew Lease, Stefan B Sigurdsson, Kevin Sikorski, Dieter Fox and Gaetano Borriello
BibTex: lamarcaplant

Automated system for taking care of plants; robot navigates and waters them. Scenario for building system, but really investigates true "out of the box", background appliances. Integrated wireless sensors and power issues resolved in novel ways. True integration of sensors, hardware, and software. Conclusion: integration is crucial. Autonomous operation.

Issues in Personalizing Shared Ubiquitous Devices.Trevor, J., Hilbert, D.M. and Schilit, B.N

Two seperate pieces to paper. Evaluation framework and personalized devices implmentation.

Makes the case for "comparative evaluation" which says that ubicomp devices have to be evaluated against other things; the ethnography approach isn't enough. Thus, when they decided to implement an approach to personalizing devices, they went with two types of devices (integrated and remote).

They wanted to make group devices (presentation computer, brainstorming display, and printer/copier) personalizable. Added terminals or built in capabilities to computers. Also augmented systems to allow connections and control from remote, mobile devices thru a wireless network. Fair bit about architecture of system. Looked at Usability, Utility, Availability, Trust, and Privacy.

Tools for Studying Behavior and Technology in Natural Settings., Intille, et al.
BibTex: intilletools

Tools for studying ubicomp in environment over long periods of time: environmental sensors, Context-aware experience sampling, and image-based experience sampling.

Sensors deployed in environement. "tape to" environement for fast install/removal. Then use ipaq w/ sensor detecting to trigger "context aware" sampling. Built open-source software for
experience sampling beyond commerically available stuff. context-aware sampling tries to overcome self-report biases. image-based sampling invoves "sampling" the user in a highly instrumented environment by snapping pictues and then interviewing or using a diary to trigger subject's recall later.

Privacy by Design - Principles of Privacy-Aware Ubiquitous Systems, Marc Langheinrich

A best practices sort of document for ubiquitous computing.

Starts with history of privacy and the 5 types of privacy: behavioral or media, territorial, communication, bodily, and information privacy.

US Privacy Act of 74 defined 7 principles:
  1. openness and transparency
  2. individual participation
  3. collection limitation
  4. data quality
  5. use limitation
  6. reasonable security
  7. accountability

Principles and Guidelines (specifically for ubicomp)
  1. Notice
  2. Choice and Consent
  3. Anonymity and Pseudonymity
  4. Proximity and Locality
  5. Adequate Security
  6. Access and Recourse

The cloak of invisibility: Challenges and Applications, Franco Zambonelli and Marco Mamei.
BibTex: zambonellicloak

"shows" in a pencil-and-paper kinda way toward invisibility cloak. Out there, but kinda fun paper.

At Home with Ubiquitous Computing: Seven Challenges, Keith Edwards and Beki Grinter
BibTex: edwardshome

Looks specifically at challenges to ubicomp in home.
  1. "Accidentally" smart home - home contains mix of ubicomp technologies not built from ground up.
  2. Impromptu interoperability - components interact either not at all or spontaneously
  3. No systems admin
  4. Designing for domestic use
  5. social implications of aware home technology - labor saving or merely shift burden of work, good for children, etc.
  6. Reliability
  7. Inference in the presense of ambiguity

  1. G. Forman, J. Zahorjan The Challenges of Mobile Computing," IEEE Computer, vol. 27, no. 4, April 1994, pp. 38-47.
  2. Lamming, M. and Flynn, M. "Forget-me-not": Intimate Computing in Support of Human Memory. Proceedings of FRIEND21: International Symposium on Next Generation Human Interface, pp. 150-158. February 1994.
  3. M. Satyanarayanan. "Fundamental Challenges in Mobile Computing", Fifteenth ACM Symposium on Principles of Distributed Computing, May 1996.
  4. M. Satyanarayanan, "Pervasive Computing: Vision and Challenges.", IEEE Personal Communications, August 2001.
  5. Deborah Estrin, David Culler, Kris Pister and Gaurav Sukhatme. Connecting the physical world with pervasive networks. IEEE Pervasive Computing, Volume 1, Number 1, pp. 59-69, January-March 2002.
  6. Tim Kindberg and Armando Fox. System software for ubiquitous computing. IEEE Pervasive Computing, Volume 1, Number 1, pp. 70-81, January-March 2002.
  7. Hani Naguib and George Coulouris (2001) Location information management. In the Proceedings of Ubicomp 2001, Atlanta, GA.
  8. Lederer, S., Hong, J., Dey, A.K., and Landay, J. Personal Privacy through Understanding and Action: Five Pitfalls for Designers. Submitted to CHI 2004. September 2003. [pdf]

Last modified 28 March 2005 at 10:56 am by Valerie Henderson