The liberalization of the electricity market and the expansion of new forms of electricity production and consumption are paving the way for new smart digital services. These new services will certainly be relying on a new generation of smart meter (SM) that will offer, among other things, prediction of consumption and production at both household and microgrid levels.These predictions can be obtained either through a generic model trained on data collected from all SMs, or through specific models developed for each SM based on its individual data. The benefit of the generic model is that it guarantees an optimal solution. However, its implementation is not possible for security reasons. The use of specific models requires managing a large number of SMs, which poses a significant challenge.This paper presents a Federated Learning (FL) approach, a decentralized privacy-preserving paradigm that achieves com-parable performance to the generic model, considering both consumption and production scenarios.The dataset, gathered from 1153 SMs over a period of 18 months, is provided by a Swiss Distribution System Operators (DSO). Although, data are individually collected per device, the generic model is trained to holistically process this data.Our experimental results demonstrate that our FL based Long Short-Term Memory (LSTM) model performs as well as the generic model and outperforms the specific models, while preserving data privacy and security.

This paper proposes Tuple Map Reduce, a new foundational model extending Map Reduce with the notion of tuples. Tuple Map Reduce allows to bridge the gap between the low-level constructs provided by Map Reduce and higher-level needs required by programmers, such as compound records, sorting or joins. This paper presents as well Pangool, an open-source framework implementing Tuple Map Reduce. Pangool eases the design and implementation of applications based on Map Reduce and increases their flexibility, still maintaining Hadoop's performance.

This article summarizes five relevant methods for developing self-organizing multiagent systems. It identifies their most relevant aspects and provides a description of each one under the form of method fragments expressed using SPEM (Software and System Process Engineering Metamodel). The use of a “metamodel” to describe fragments facilitates the comparison of the methods and their respective fragments. These fragments can be combined and be part of a more general ad hoc methodology, created according to the needs of the designer. Self-organizing traffic lights controllers and self-organizing displays are chosen as case studies to illustrate the methods and to underline which fragments are important for self-organizing systems. Finally, we illustrate how to augment PASSI2, an agent-based methodology which does not consider self-organization aspects, with some of the identified fragments for self-organization.

This article defines and analyzes a collection of algorithms for persistent storage of data at specific geographical zones exploiting the memory of mobile devices located in these areas. Contrarily to other approaches for data dissemination, our approach uses a viral programming model. Data performs an active role in the storage process. It acts as a virus or a mobile agent, finding its own storage and relocating when necessary. We consider geographical areas of any shape and size. Simulation results show that our algorithms are scalable and converge quickly, even though none of them outperform the others for all performance metrics considered.

This paper addresses a vision of future manufacturing systems, which are highly agile, user friendly, and increasingly based on autonomous components. Evolvable assembly systems (EASs) provide a solution for agile assembly, including a concept for reconfigurability with modularity at the mechanical as well as at the control level. It takes the multilateral relations among product, processes, and systems into account and allows the systems to evolve together with the requirements. Self-organizing assembly systems (SOASs) are a further development of EAS, allowing them to play an active role in layout design and production. This paper focuses on the self-organization mechanisms for the design of SOAS, as well as the system architecture, including agents and self-knowledge. Agentified modules participate in their own arrangement in the system layout and monitor themselves during production. Policies and metadata for self-management during production are described, and performance metrics for agility scenarios are indicated.

Agile manufacturing requires high responsiveness at all levels of a company, but is especially challenging on the shop-floor level. Evolvable Assembly Systems (EAS) are a solution: agentified modules can be seamlessly integrated into existing systems, or removed at any instant. EAS offer a more flexible solution to automation production, but many system design and integration tasks are still done manually. Our goal is to make EAS increasingly self-managing: 1) to easily and quickly produce a new or re-configured assembly system each time a new product order arrives or each time a failure or weakness arises in the current assembly system and 2) to maintain production also under degraded conditions. This article describes an architecture for self-managing evolvable assembly systems. It involves on-the-fly self-assembly of robotic modules, dynamic coordination of tasks and self-adaptation to production conditions, mainly self-healing and self-optimisation. The architecture exploits self-description of modules, monitored modules behaviour and dynamic policies.

This paper describes a decentralised car traffic control simulation with re-routing and propagation of messages among traffic nodes (roads intersections and traffic lights). The values of the parameters governing the simulations are identified through the use of a genetic algorithm. This paper reports as well on results obtained regarding the convergence of the genetic algorithm towards fittest solutions.

introduction Introduction Share on Editor: Giovanna Di Marzo Serugendo View Profile Authors Info & Claims ACM Transactions on Autonomous and Adaptive SystemsVolume 1Issue 1September 2006 pp 1–3https://doi.org/10.1145/1152934.1152935Published:01 September 2006 1citation739DownloadsMetricsTotal Citations1Total Downloads739Last 12 Months6Last 6 weeks1 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

The rapid expansion of Java programs into the software market is often not supported by a proper development methodology. We present a formal development methodology, well suited for Java dependable distributed applications. It is based on the stepwise refinement of model oriented formal specifications, and enables validation of the obtained system wrt the client's requirements. Three refinement steps have been identified in the case of fault tolerant distributed applications: first, starting from informal requirements, an initial formal specification is derived. It does not depend on implementation constraints and provides a centralized solution; second, dependability and distribution constraints are integrated; third, the Java implementation is realised. The CO-OPN/2 language is used to express specifications formally; and the dependability and distribution design as based on the Coordinated Atomic action concept. The methodology and the three refinement steps are presented through a very simple fault tolerant distributed Java application.