Chairs: Jadwiga Indulska (University of Queensland, Australia), Daqing Zhang (Institute TELECOM & Management, SudParis, France)
After almost a decade of research, pervasive computing is gradually coming of age. The research community has addressed a large scope of the scientific problems related to context-awareness, adaptability, communication paradigms, sensor/device management, human-computer interactions, software engineering and security/privacy of pervasive systems and has also created models and prototypes. Does this mean that the vision of pervasive computing is about to be realised and that we will shortly see implementations/deployments of pervasive computing systems in various aspects of our life and work?
The panelists will outline a vision for pervasive/intelligent computing and discuss whether the research progress to date in pervasive computing will allow us to achieve this vision. The questions the panelists will try to answer include: What do we expect from pervasive/intelligent computing? in general in various domains (work, home, health, education, telecommunication, etc) Have we achieved this vision and, if not, what is required to achieve it?
Chairs: Hein Meling (University of Stavanger, Norway) and Kevin Kwiat (Air Force Research Laboratory, Information Grid Division, Cyber Defense Branch, USA)
The ubiquitous nature of computers has made them a victim of their own success. They are everywhere, but their creators demand that they be simple and low cost. Let me clarify the latter phrase as it too has a contradictory nature. "Simple" is not what necessarily sells. For example, a cell-phone manufacturer conducted a study of their two phone designs, one with N features and another with N + M features were prototyped. The more complex one (the one with M more features) was vastly preferred, in spite of its greater cost, by buyers over the simpler one. Ironically, the purchasers of the more complex, more costly phone did not even use the additional M features; instead they seemed to pay for the additional features because they initially thought that they might use them or did not want to be without them in case they ever needed them (which never arose). In light that the market demands designs that are not necessarily simple, we should re-cast "simple" to mean understandable to a user and well-understood by the designer. The greater a designer's understanding, however, the greater the likelihood that the product will be of reasonable cost (so production can meet what the market will support) and whose subsequent performance is safe. Simplicity, however, will be only in appearance: through information hiding and machine abstractions complexity is seemingly conquered. This is to be a daunting task particularly because complexity is the breeding-ground for faults and vulnerabilities that can penetrate the surrounding aura of simplicity to reveal that the underlying complexity was indeed untamed. Our job then becomes one of furnishing system designers with the tools and insights to make their designs for ubiquitous computing safe.
For instance, a group within the Air Force Research Laboratory (AFRL), has been working on voting protocols to support fault tolerance - including faults that may be induced by attackers. The intention is to treat voting as a black box that, with the proper interfaces, can be plugged everywhere into a system. Designed for energy-efficiency, customization, ubiquity of voting can be realized by insulating the information receivers from the semantic knowledge of the information being voted upon. This makes the voting protocols pluggable and therefore amenable for middleware realizations .