Emerging and future large-scale pervasive systems open new perspectives that would have appeared infeasible a few years ago In particular, they are expected to perform complex tasks in a transparent way at a global scale. The great heterogeneity of the systems involved, their geographic pervasiveness, external or internal mobility, and administrative independence make a centralized approach infeasible. Moreover, the huge scale of these systems, the limited capabilities of the individual devices, and the dynamic on-the-?y changes of position and status of the system components, pose dramatic computational and algorithmic challenges. Even basic tasks, such as computing the average of a collection of measurements, maybe challenging problems when they have to be the collective effort of a myriad of geographically distributed, independent devices, having only limited computational power and storage capacity, and missing a global view of the network and a centralized administration.
Large-scale networks of artefacts may have an emergent behavior, which makes it very dif?cult to predict their future behavior The emergent global system is much more complex and powerful than each of their components. The goal of network modeling and analysis is to provide the ability to understand network behavior. With this, we can make decisions that are more informed at future junctures.
The previously described settings introduce a set of new characteristics to consider: highly decentralized systems, possibility of changing environments, resilience to failure of portions of the system (temporarily or permanently), dif?culty or impossibility of energy acquisition. Therefore, energy consumption has become one of the most important parameters in these networks, which affects several basic functions, of which the most relevant are: communication and transmission of information, lifespan of the artefacts, mobility, ability of replication. Taking into consideration these parameters, we need to come up with new complexity concepts for the existing and future pervasive networks.
The objective of this workpackage is to de?ne a set of concepts and tools that allow us to formulate a new, perhaps partial, theory for adaptively organized societies of artefacts. We will take concrete steps towards answering the following fundamental, open questions in the ?eld: “Is there a single, unifying, abstract model for such large systems of tiny artefacts explaining their emergent behavior?” Such a model does not need to be build in a clean-slate fashion, and could be based on the different existing parallel and distributed models.
Our methodological approaches focus on three different aspects of pervasive networks and concentrate our foundational work on the key issues that we want to understand. We aim to know the parameters and performance measures that are relevant to the correct and smooth desired network behavior. Accordingly, we have split the work into three tasks:
- Dynamicity models and algorithmic consequences
- Computation models and algorithmic consequences
- Cooperation models and algorithmic consequences