The 31st IPP Symposium
Middleware Issues in Pervasive Computing
Jim Waldo, Sun Microsystems
Usual examples of pervasive computing envision a world in which we talk to or gesture at the various controllers that surround us in smart spaces to allow us seamless access to computing environments no matter where we go. I will argue that this vision emphasizes the computing aspect of pervasive computing but is almost entirely wrong-headed, both because of the limited functionality we could expect of such "smart spaces" and because such visions cloud the set of real problems that need to be addressed if we are to enable true pervasive computing.
Such computing really has to do with making the computer disappear into everyday objects and rests on the ability of these embedded computing devices to interact with each other, not with people. The real problems that middleware will need to solve include such things as separating the identity of a computing entity from its location, designing around resource limitations, and insuring that the failure models are well understood by the users. Understanding the immediacy of the problem and the nature of the resulting systems is a needed first step, but others will require techniques such as mobile objects, moving computation to the edge of the network, and high degrees of redundancy in overall system design.
