In the August 2009 column, the use of software agents was discussed in the context of an application in optimum 3G network planning. This issue's contribution continues the theme of discussing contemporary concepts in IT within computational electromagnetics: in this case, implementing a distributed antennaoptimization platform using Java mobile agents and Particle-Swarm Optimization (PSO).Additionally, we have received correspondence regarding the June 2009 column, and this appears following this month's contribution.As always, we thank all the authors for their contributions.
AbstractVarious techniques and platforms have been introduced for the effective use of distributed computational resources when using specialized computational electromagnetics (CEM) simulation software for designing antennas. These techniques usually impose significant entry barriers to researchers, in the form of a steep learning curve for specialized application programming interfaces (APls), or limited access to resources such as clusters or computational grids. The current paper discusses a sophisticated approach for alleviating these barriers, with the fusion of the programming paradigms of objectoriented programming and mobile agents. A case study concerning the development of a distributed Particle Swarm Optimization (PSO) platform -an increasingly popular algorithm within the antenna-research community -is thoroughly analyzed and presented as an example of data-level parallelism. Various aspects of computational efficiency in terms of scaling have been examined, in order to estimate the merits of the proposed approach. The aim of the analysis is to stress the advantages of the above-mentioned techniques for the effective loose coupling of CEM programs with distributed resources, demonstrating ease of application, improved scalability, and future expandability.