Development of a Java-based distributed platform for the implementation of computational intelligence techniques

This paper reports three phases of development of a. Java-based distributed system for the implementation of computatiod intefigenfe techniques to solve pmctid inherently parallel. Their structures are thereby well-suited to some of the parallel computer structures such as cluster systems. Further, it is more cost-effective to build a parallel computer than to improve performance of a single engineering problem. The first design of the Java-hased system is P m Worker. It is a Library of AppK=tion Pmgram Interface (APIS) to facilitate parallel computation. N~ ~ a new system, para Worker 2-4 at impmving esse of pmb. and performanee. developers across the industry now agree that clock speed ad & feat,, = of improved d y n d C object is no longer the key measure of processor performance. reallocation. adautive consistenw orotocols. and location Rather, the aim of improving the overall throughput of a computer. physical limitations and increasingly sophisticated chips design are two major challenges in the development of high performane processors 111. Chip transpmncy. However, neither i f bese systems addressed the fundamental issues of Java Virtual Machine (JVM) in distributed environment. The third phase of development of the Java-based system is a new form of distributed JVM i