Vol. 25 · No. 4

This paper focuses on one of the applications involved into the CrossGrid project, the STEM-II air pollution model used to simulate the environment of As Pontes Power Plant in A Coru na (Spain). The CrossGrid project offers us a Grid environment oriented towards computation- and data-intensive applications that need interaction with an external user. The air pollution model needs the interaction of an expert in order to make decisions about modifications in the industrial process to fulfill the European standard on emissions and air quality. The benefits of using different CrossGrid components for running the application on a Grid infrastructure are shown in this paper, and some preliminary results on the CrossGrid testbed are displayed.

The rapid advance in information and communication technology (ICT) has given new impetus to shift the information services paradigm from platform centric to network centric computing. Commercial activity is blooming on the internet. The user or customer has become dynamic and has changing tastes and pattern in the demand of service, and desires the service at any time or place. The service provider therefore has to tailor the service provision format to suit the dynamic nature of users of information services. A business cannot afford to ignore the rapid and evolving nature of its customers. However, the current state of the wide area network services is finding it difficult to respond to constantly changing and heterogeneous demands of modern business, being centralized in nature, with the service provided through a single URL. It is imperative to update the pattern of information service provision and utilization. Faded information field architecture (FIF), reported recently, holds the potential to address these issues, being a demand-oriented architecture. Although research into various aspects of FIF has been reported, we suggest algorithms to characterize the behavior of mobile agents to seek the required information at a given node in the FIF architecture. Simulations were carried out to show the effect of various parameters on the performance of the FIF system.

In the wireless world human perspectives play an important role in understanding and developing 3G and beyond mobile and wireless applications. For this we need to develop some reference models in order to best describe the requirements in developing and deploying future generation multimedia mobile and wireless applications. In this paper it first comes up with an introduction of a reference model which forms the basis in Wireless Strategic Initiative (WSI) research project. A user-focused model is then developed which describes the users' wants and needs in mobile and wireless applications. The CyPhone system is then presented and it is followed by both static and dynamic reference modelling, which are drawn from CyPhone mobile navigation scenarios. This paper contributes to the understanding of the generic reference models and static and dynamic reference modelling in the wireless world. This paper ends up with a conclusion and future work is outlined.

Niching techniques for evolutionary algorithms are used in order to locate basins of attraction of the local minima of multi-modal fitness functions. Co-evolutionary techniques are aimed at overcoming limited adaptive capabilities of evolutionary algorithms resulting from the loss of useful population diversity. In this paper the idea of niching co-evolutionary multi-agent system (NCoEMAS) is introduced. In such a system the species formation phenomena occurs within one of the preexisting species as a result of co-evolutionary interactions. The results of experiments with Rastrigin and Schwefel multi-modal test functions aimed at the comparison of NCoEMAS to other niching techniques are presented. Also, the influence of resource sharing mechanism's parameters on the quality of speciation processes in NCoEMAS are investigated.

We present a new in-place selection algorithm that finds the k rm th smallest element in an array of n elements in linear time, using only ε.n element moves. Here ε>0 denotes an arbitrarily small, but fixed, real constant. As a consequence, partitioning the array in-place into segments of elements with ranks smaller than, equal to, and larger than k can be performed with (1+ε).n element moves. Minimizing the sum of comparisons and moves, we get a selection algorithm using C(n)<10.236 n comparisons and M(n)<0.644 n moves. The algorithm can be further optimized, by tuning up for the given cost ratio between a single move and a single comparison. As an example, we present an algorithm with C(n)+10.M(n)leq 13.634n.