Vision, Issues, and Architecture for Nomadic Computing

Rajive Bagrodia, Wesley W. Chu, Leonard Kleinrock, and Gerald Popek

One-line summary: "Nomadicity may be defined as the system support needed to provide a rich set of computing and communication capabilities and services to nomads as they move from lace to place in a transparent, integrated and convenient form."

Overview/Main Points

Nomadic computing: given the surge of mobility (portable computers, wireless connectivity/communications infrastructure), support for nomadicity must be developed. Nomadicity is defined as the system support needed to "provide a rich set of computing and communication capabilities and services to nomads ... in a transparent, integrated and convenient form." The claim is that nomadicity is largely a middleware problem.
Architecture, Protocols, and Reference Model: These must be defined. They must handle heterogenous operation, unpredictability and variability in all parts of the system, scale and degradation, QoS, ad-hoc access, cooperation, and adaptability. (Did I get all the catch-phrases?)
Transparent Virtual Networking: This is essentially disconnected operation, but at all levels of the system (file system, RPC, databases, management). Optimistic replication and precaching is the method suggested to achieve this.
Intelligent Databases: Computer, analyze everything in the universe and tell me the answer, then tell me the question. The authors envisage an intelligent resource discovery facility that can handle all sorts of heterogeniety in resource types, locations, accesibility, through the use of AI-flavoured "knowledge bases". Adaptive agents will provide approximate matching and query relaxation - the process of matching a query when the user has no way of having enough information to intelligently form the query.
Performance Evaluation: Nomadic systems are large and complex; modeling and evaluating such systems will become a difficult task. A hybrid approach to simulation is suggested, in which real components of the system are interspersed with simulators to achieve the overall behaviour of the system at minimal cost (in time or effort).

Relevance

Nomadicity is a real issue, and is a difficult issue. This paper does a good job of introducing the concepts and problems we will face in the future, and as such is a good contribution to the field of computer science.

Flaws

The authors attempted to show vision through suggesting reference models and potential solutions to some of the problems, but IMHO their visions were overly speculative, and as such, were more science-fiction than science.
Why did all of this database discussion creep into the paper? It is to some degree irrelevant and obtrusive, given the paper's visionary flavour.

Back to index