Editorial for the special issue on "Adaptivity in Heterogeneous Environments"

The increasing complexity of systems and applications in heterogeneous environments have brought to the fore the issue of seamless access to the resources that the Internet and other foundational systems are making available. Resources can be generated or mediated by users, applications, networks and distributed systems. They identify different contexts of interaction where heterogeneity may occur. The dynamic and non-deterministic nature of these environments requires the adoption of decentralised models, policies and mechanisms in order to address issues of complexity, interoperability and integration. In this context the ability of systems to reconcile conflicting requirements and negotiate appropriate state or behaviour to suit environmental conditions delineates a wide spectrum of adaptivity. From basic configurability to autonomic behaviour adaptivity is inherently a systemic issue. In hypermedia systems the personalisation of content through the generation and interposition of implicit and explicit user profiles is underpinned by adaptive principles. The mediation of scalability, the marshalling of resources and the scheduling of tasks in Grid Computing, for example, are areas where heterogeneity is resolved. Appropriate modes of aggregation such as Web Services composition are often supported by adaptive mechanisms and optimised by schemes such as load balancing. Reflective architectures at middleware level have been proposed as effective means for extending adaptively the span of systems and applications. In wireless networks handover operations are increasingly dependent on adaptive methods, while the emergence of self-awareness in networks is enhancing their capacity for self-management. The demand for dependable systems and especially the provision of appropriate levels of quality of service (QoS) are strong incentives for the application of transparent adaptive techniques. Adaptivity is finding its full expression in new domains such as smart environments that pervasive and ubiquitous computing have ushered in. Context and self-awareness in particular are reinforcing the adaptive capacity of systems as they unfold across environmental boundaries. The aim of this special issue is to offer a perspective on adaptivity in heterogeneous environments by presenting a selection of papers by contributors to the International Workshop on Adaptive Systems in Heterogeneous Environments (ASHEs). The workshop was held in March 2009 in Fukuoka, Japan, in conjunction with the International Conference on Complex, Intelligent and Software Intensive Systems (CISIS). The scope of the special issue was widened by including invited submissions from researchers with a track record in adaptive systems. Extended versions of the papers were considered and subjected to a rigorous review process. The review has resulted in the selection of eight papers from the original ten submissions. In the following summary of the papers the thread of adaptivity that binds the different themes is manifest. Although the introduction and adoption of adaptive principles and techniques in the different research strands is motivated by specific concerns the main driving factors are mediation and scheduling. These aspects can also be expressed in terms of QoS provision. One of the most common manifestations of adaptivity is in the reconciliation of conflicting requirements. In Lin et al. the main concern is the creation of a framework that will enable a large group of Web Services to overcome their differences and reach a consensus. They propose a scheme for clustering Web Services into multi-groups according to their fuzzy QoS dispositions and their preferences over these attributes. The clustering and analysis process can be moderated dynamically according to feedback from the internal learning mechanism and the web service consumers. An extensive case study is presented as a demonstration of the effectiveness and efficiency of the proposed model. The heterogeneity of resources is another area, which demands adaptive solutions. Glasner et al. address the issue of efficient scheduling and its inherent complexity in Grid computing. They propose a system, which operates over multiple phases, adjusting its behaviour according to the jobs submitted by users and predicting adaptively the run-time of submitted jobs. The accuracy of the system is refined by clustering jobs according to their characteristics and by applying appropriate prediction techniques. The simulation results indicate that this adaptive adjustment can lead to accurate predictions. Caminero et al. offer a wider perspective on the discovery, access and aggregation of geographically distributed resources in Grid Computing. The proposed approach to the efficient coordination of tasks in different administrative domains follows a P2P mode of interaction. Properties of the domains are used to determine where a request for resources should