In this paper we propose a simple powerful method to increase communication performance in large networks such as the Internet. Our approach is based on the usage multiple disjunct paths for the transmitting data from one node to another one. The method exploits the dense connectivity of the Internet by making use of the different routes engaged in communications originating from different nodes to one target nodes. We show that depending on the bandwidth offered and the amount of data to be transfered, a linear speedup can be achieved.

Current tools available for high performance computing require that all the computing nodes used in a parallel execution be known in advance: the execution environment must know where the different "chunks" of programs will be executed, and each computer involved in the execution must be properly configured. In this paper, we describe how the ) environment may be used to dynamically locate available computers to perform such computations and how these computers are dynamically configured.

We summarize the interaction between the theory behind intensional programming, as seen in Lucid; and intensional versioning, as seen in Lemur, Intensional HTML, ISE, VMAKE, etc. These two concepts can be seen to be duals of each other, and they rely on dual notions of store, the warehouse for caching values, and the catalog for providing definitions. Catalogs contain intensional objects, which are openable boxes labelled by Lucid contexts. Finally, Lucid contexts are now first-class values. 1

Abstract. The last few years, a tendency arose to integrate various programmable devices through ever expanding computer networks. One particular domain in which this evolution stood out clearly is home automation. The EIB 1 standard defines a home automation solution that consists of a network of cooperating components with little computational power. Bridging the EIB network with a computer network allows software to interact with these EIB components. However, past attempts to link EIB components with computer networks fell short in dynamism, automatism or user friendliness. In this paper we present a distributed software framework that overcomes these issues, and is capable of automatically generating a software model based on the configuration of the components on the EIB fieldbus. A framework that can be used in this perspective is Jini. Its excellent capabilities for dynamic reconfiguration and its proven deployment in domestic and office environments make it an appropriate candidate for supporting home automation systems. 1

Abstract. The large number of services and the huge amount of information available in today’s large scale distributed systems such as the Internet, make efficient search mechanisms an unconditional issue. Many search methods and engines in such systems have been proposed and are in use. However, most of them are based on some centralized catalogue. Only recently, distributed search methods have become available in the context of Peer-to-Peer networks (e.g. Gnutellalike systems). This paper proposes a distributed search mechanism based on multiple concurrent search message chains. The method has been evaluated with a stochastic simulation model and with a prototype implementation. The results obtained from both experimentations correlate well, thus showing the viability of the approach. 1

Resource localization. Current tools available for high performance computing require that all the computing nodes used in a parallel execution be known in advance: the execution environment must know where the different “chunks ” of programs will be executed, and each computer involved in the execution must be properly configured. In this paper, we describe how the Web Operating System (WOS TM) environment may be used to dynamically locate available computers to perform such computations and how these computers are dynamically configured. The WOS TM (Kropf 1999) is a virtual operating system which is suitable for supporting and managing