Classical analysis of when and how to apply basic engineering design techniques.
Describes a distributed system, where decisions must be reached by voting. However, in compromised or damaged systems, some of the nodes in the system may conspire to vote against the systems best interests. A great deal of research has been done to:
Two Phase commit is used in distributed database systems to establish distribued checkpoints. One of the nodes (any node can assume this role) is designated the leader, and instigates a global checkpointing operation. The steps are:
A seminal concept piece that predicted the world wide web, hypertext and ubiquitous computing (e.g. in Section 1, he describes an untethered researcher, with a mobile data management tool).
Seminal Intro to some of the technical issues in Ubiquitous Computing.
Introduction to what is wearable computing
Provides a More Recent Coverage, and Satyanarayanan's got an interesting vision.
It is a bit early to know what impact this will have, since it is so current. However it seemed interesting on a first glance. I especially liked the "nagging because Mom's calls didn't get through feature".
This paper has an information retreival bias, but still interesting. However, I have not yet found the correct citation for it (was it published?).
This paper is an extension of a very interesting work at MobiDE in 1999 and ISADS 1999 by GOTO et al., which focused on visually impaired passengers. GOTO and KAMBAYASHI address mobility concerns and extends previous work on interface design and location dependent queries.
I think the idea of recording observations is a key application area for ubicomp, in this paper, they discuss the deployment of a prototype based in the HP CoolTown architecture in a Museum to assist visitors in browsing and reviewing the exhibits.
Suggests that power not memory bandwidth is now limiting processor design. Suggests going for parallel low power devices with low clock rates rather than high clock rate sequential solutions.
I really wish I could find a paper that covers the issues presented in this talk.
Models memory power consumption and presents a model for putting RAMBus memory to sleep (on a timer). Processor speed and architectural tradeoffs (e.g. out of order execution and clock speed) are also considered.
An exciting new development in electronic paper, uses charge to reposition ink and allow for high speed, high definition and brightly colored thin, flexible video devices. Note that a subscription is required to download this paper, I think University at Albany has a site license, I downloaded the paper from my lab, but could not download at home (over road runner).
Describes how nematic LCD's work and how a bistable device can hold its state when powered down. What this means is that only the pixels being updated need power.
An overview of a small high performance mobile device architecture.
Seminal paper on discussing handling disconnection in environments where there are many small and seldom shared files.
The Millywatt group at Duke has been doing some interesting investigation of power management,
A study of how power can be traded off for quality of results, for a nice intro to Multifidelity approaches see:
Multi-Fidelity Algorithms for Interactive Mobile Applications, by M. Satyanarayanan and Dushyanth Narayanan. in Wireless Networks, 7:6, 2001. pp. 601-607.
Describes a fast lookup technique which imposes a logical ring topology and uses pointer jumping combined with heirarchical decomposition to provide fast access. Distributed hashing is used to load balance during addresses computation.
Covers challenges in P2P systems.
Bloom filters are widely used for hint generation in distributed systems (e.g. resolving which server has a file given the file name). Recent work by Saar Cohen and Yossi Matias (in SigMod 2003) on Spectral Bloom filters focuses an efficient (but nontrivial) method for using the minimum number of bits in the counters.
Need the reference or a better citations
An overview of many important areas in streaming data management, but not strictly focused on ubiquitous computing.
Discusses issues associated with scheduling in complex queries, a very nice paper.
Gives a nice overview of several threat models.
Ad Hoc Networks by definition lack fixded infrastructure. They can be secured by careful secret sharing and distributed key management to prevent a small number of nodes from conspiring to accesss protected data.
A paper describing methods for precisely specifying spatial policies in mobile systems (e.g. whenever Alice enters a room, turn on her favorite music). It seems that these policies are more general than security, there may be a followup paper at the same site which may be worth seeing when it becomes available.