ABSTRACT Software agents and multi‐agent systems are critical components in the development of distributed autonomous systems, as they exhibit a wide range of “intelligence,” from reactive behaviors to advanced reasoning and planning. Integrating these capabilities into multi‐agent systems enables distributed artificial intelligence, a programming paradigm in which autonomous entities work together efficiently across a network to achieve shared goals. Another important aspect we consider in this work is represented by the disruptive impact of 5G technology, which is moving computation to the edge of the network. As a result of this shift, embedded devices with microcontrollers—which are renowned for their low power consumption and limited computing power—have proliferated. This paradigm fits well an environment made of small cooperating entities scattered across the network, but, at the same time, it is difficult to provide agent platforms for embedded devices. Indeed, the most recent state‐of‐the‐art multi‐agent systems platforms, which are primarily Java‐based, raise issues with microcontroller speed and memory overhead. To achieve this goal, we designed and implemented an ecosystem, entirely written in C++, to create multi‐agent systems in an actual Internet of Things setting. Our ecosystem is composed by Democle , a logic/declarative agent programming framework, and Hermes , a unified wireless network interface, both specifically created for embedded systems. In this paper, we describe our ecosystem's functionality, architecture, and benefits of use. In addition, we provide a case study and a few experiments to prove how our platform works and its efficiency in terms of overhead.

This paper introduces an intelligent monitoring and anti-theft system designed for marine environments. Using Internet of Things technologies, the system utilises onboard devices that function as a finite-state machine, equipped with multiple sensors to gather speed, orientation and movement data. The onboard device is configured by the user with a personal device, which is used to arm and disarm the system via Bluetooth Low Energy. In case of theft, live position data is transmitted across a LoRa Mesh network, enabling long-range communication provided that enough boat nodes are available. The packets are transmitted using an adequate cryptography system based on AES128 keys. The data transmitted is recovered by a dock node, and it can be seen by the dock keeper and simultaneously provided to the users’ application.

This work introduces an advanced system designed to ensure operator safety in dead zones, common in underground and remote areas with limited internet connectivity. Using Internet of Things technologies, the system employs wearable devices with sensors to provide real-time data for precise location tracking, movement analysis, and monitoring of high-risk areas within these zones. The system enables real-time vital sign monitoring through wearable technology and integrates machine learning for fall detection, health monitoring, and distance estimation between operators and anchors in tunnels. These wearable devices operate independently in connectivity-challenged environments, enhancing system coverage and troubleshooting efficiency. Data collected is securely stored internally and accessible via an external IP network for remote monitoring, configuration management, and oversight of operator location and health status.

As the demand for computational power continues to grow, traditional data processing approaches face limitations, primarily related to physical constraints in materials and processing units. In order to deal with such limitations, the computing industry has turned to multi-core and Symmetric multiprocessing technology, harnessing the combined processing capabilities of multiple processors to spread workloads to improve performance. One of the key strategies in providing computational power for parallel processing is represented by the "multi-tenant approach" of Cloud Computing, where multiple tenants can share a huge amount of computational power for the execution of services. Private Cloud is a slightly different approach which ensures more security and control. At the same time, it can provide service continuity and reliability at the system level rather than relying on individual components. In this paper, we discuss the key aspects of cloud computing w.r.t. the existing different service models, particularly focusing on the private Cloud model. To this end, we discuss established technologies for Cloud Computing, particularly the well-known OpenStack. We discuss the key aspects related to the deployment of OpenStack using PuppetStack, which is a software solution we employed for our research. PuppetStack can support the deployment of high-availability private cloud infrastructure.

The Internet of Things is rapidly evolving, thanks also to technological innovations, in particular to new technologies of radio communication with low energy consumption. Nevertheless, testing a possible complex infrastructure to verify the management of several devices and their communication could be difficult to implement and especially wasteful. LWN Simulator is a LoRaWAN network simulator, able to simulate device nodes and gateway and it is equipped with a web interface with which to interact with the simulation in real-time. The simulator lets the user create and interact with Virtual Gateways, changing their status, geographical position and other network information or adding a Real Gateway already present in a real network. Also, the simulator allows the simulation of all device classes, supporting the Adaptive Data Rate, MAC commands and the personalisation of the payloads. Finally, the entire simulation environment, devices and gateways, runs on a real LoRaWAN infrastructure, such as Chirpstack or The Things Stack. This feature allows a real network design that can be implemented using real devices with the same settings and configurations.

This paper presents a software architecture to let users interact with their smart home devices, through the commonly used social network channels. The software, called FABULOS, is the result of the combination of components used to interact with devices and social networks, and a rule-based artificial intelligence, which implements the base logic for the automation. The core of the software architecture is the translation service, which has the objective of extracting the meaning of the sentences provided by text and voice messages by users in natural language, transforming them into proper device commands. In order to achieve this, the proposed solution relies on an approach which has the capability to interpret and extract the meaning of the intentions plus the entities involved, associated to the sentences sent by the users. A description of a realistic case-study, which shows an example of how the proposed software behaves in a real interaction with an user, is also included in the paper.

This paper presents the software architecture of a robotic museum guide application called Cuma. It is intended to run upon the Pepper robotic platform and has the objective of guiding visitors of a museum accompanying them in the tour, explaining museum works, and interacting with them in order to gather feedback. Cuma has been partially implemented and preliminarily tested. The results reported in the paper, highlight that even if Pepper, from the structural point of view, seems particularly suited for this kind of application, the provided software platform presents some important limitations thus requiring the integration of external tools and algorithms.

This paper proposes a trust model for a competitive federation of Cloud computing systems. The reference scenario is a large-scale environment in which different Cloud providers compete in offering their pay-per-use XaaS (Anything as a Service) resources to the various users. In this context, to obtain the assignment of a user request, providers may state to possess more resources than the amount really available: the result could be the provisioning of a services with a lower QoS than expected. A trust model is thus mandatory to help the requester to properly choose the adequate provider on the basis of (i) the reliability shown in the past in providing services, and (ii) the reputation the provider has to other users. In this sense, the paper presents an extension to the RRAF model, which has been suitably adapted to a large-scale environment by means of the introduction of a multi-dimensional approach which fits the various reliability issues, typical of Cloud computing systems.

In an agent-based cloud scenario, when two agents, provided with different ontologies exchange messages to negotiate a Service Level Agreement for a cloud service, they could fail to understand both the correct meaning of their messages and the offered service quality, due to the differences present in their ontologies. In all these cases, a semantic negotiation is required to reciprocally making understandable the messages exchanged about the cloud service characteristics. In this work we propose a protocol for supporting the SLA negotiation process, involving semantic issues, which do not require the use of a common global agent ontology sharing knowledge in advance. In detail, this protocol considers that each agent is able to partition the other agents based on both their expertise in cloud services and similarity with their own ontologies.

This paper describes GCS (GPU-aware ComplexSim), a simulation framework for complex systems which is capable of executing on GPUs exploiting the CUDA programming model. GCS is based on an architecture similar to a previous work, ComplexSim, which provides simulation functionalities on symmetric multiprocessing (SMP) systems. With the current architecture of GCS, the simulation of the complex system run on GPUs, while tasks related to graph analysis still run on the host, by exploiting the embedded multi-thread engine of ComplexSim. The user code can be provided as a behaviour function, without taking into account issues related to task parallelisation, which are managed by the engine of GCS. In GCS, network data -" as links and mailboxes -" are organised as SoA (Structure of Array) to deal with the constraints and optimisation issues related to GPU architecture and the CUDA programming model. Moreover nodes attributes are defined by the user as in the case of ComplexSim, but GCS automatically organises them into SoA. GCS exhibits a significant improvement, in terms of simulation times, if compared to ComplexSim running on a SMP system.

This paper describes Complex Sim, a C-based simulation framework aiming at providing a simple and powerful support to the study of complex networks and P2P protocols evolution by means of simulations on symmetric multiprocessing (SMP) systems. The lower architectural layer of complex Sim, the Parallel Simulation Kernel, includes a complete support for multi-threading which allows to gain a speedup when running simulations on SMP systems. The Parallel Simulation Kernel allows to define and schedule (parallel) tasks and events with a simple API, without dealing with the management of threads, which are hidden by the simulation kernel itself. The Complex Network Data & Runtime is the architectural upper layer. It allows to define a complex network or P2P system model as a graph of entities with user-defined attributes and runtime behavior. By means of the API of Complex Sim, the user code is able to simulate the evolution of its own model by focusing only on the run-time behavior of the network and the nodes, without necessary dealing with aspects such as graph management, message exchange between nodes, event management, task/thread scheduling.

This paper presents a novel approach for resource finding and job allocation in a computational Grid. The technique proposed, called HYGRA for HYperspace-based Grid Resource Allocation, organises Grid nodes in a peer-to-peer architecture by building an overlay network in which each node is virtually connected to one or more neighbours. HYGRA is strongly based on spatial computing concepts: the available amount of each resource type is used as a geometric coordinate in a virtual Euclidean hyperspace. Therefore each Grid node is represented as a point in the hyperspace, on the basis of the amount of its free resources. A job request is also represented as a point in this hyperspace, according to the resource amounts needed for job execution. Thus, in this model, finding a Grid node suitable for execution of a job request means navigating the said resource hyperspace. Node search goes from neighbour to neighbour, following the overlay links trying to minimize the Euclidean distance between the current node and the target point representing the job, until the node nearest to it is found. The paper describes the algorithms for overlay construction and resource finding and assesses their validity and performances by means of simulation.

Abstract This paper proposes a software architecture based onmobile agentsfor distributed process control applications. A set of agents is employed to handle, in a single manufacturing cell, automatic assignment of control tasks to controllers, monitoring of cell functionalities and dynamic cell reconfiguration. The agents operate in a two‐layered structure: at the highest level, theplanning agentsanalyse the inputs of the system designer and automatically create thefield agents, which operate at the lowest level and embed the control tasks to be executed. Field agents, which are mobile, are able to reach autonomously the controllers of the cell, in order to perform the control activity there. Exploiting the mobility enables a field agent to change its running device when the variation of the design parameters or a system fault requires a new task distribution. Aload‐balancingalgorithm is introduced, with the objective of assigning each field agent to a controller of the manufacturing cell in order to fairly distribute the computation load. The algorithm uses abranch‐and‐boundtechnique to explore all possible solutions and applies two heuristics to throw away non‐feasible solutions and select the best branch to analyse. The algorithm is designed to run on‐line in order to allow a fast task redistribution when a fault condition occurs in the process control environment. Copyright © 2008 John Wiley & Sons, Ltd.

Abstract This paper presents a peer‐to‐peer (P2P) approach for the management, in a computational Grid, of those resources that are featured by numerical quantity and thus characterized by a coefficient of utilization, such as percentage of CPU time, disk space, memory space, etc. The proposed approach exploits spatial computing concepts and models a Grid by means of a flat P2P architecture consisting of nodes connected by an overlay network; such a network topology, together with the quantity of resource available in each node, forms a three‐dimensional surface, where valleys correspond to nodes with a large quantity of available resource. In this scenario, this paper proposes an algorithm for resource discovery that is based on navigating such a surface, in search of the deepest valley (global minimum, that is, the best node). The algorithm, which aims at fairly distributing among nodes the quantity of leased resource, is based on some heuristics that mimic the laws of kinematics. Experimental results show the effectiveness of the algorithm. Copyright © 2006 John Wiley & Sons, Ltd.

First International Workshop on Emerging Technologies for Next-generation GRID (

Abstract Agent Communication Contexts (ACCs) are virtual environments where agents may live and interact. In this way, as in a human society, interactions may be subject to conventions and laws depending on the context where they occur. For this to be possible, an ACC should embed the communication laws relevant to the intended class of agent applications and enforce them, as interactions among agents take place. Although context is a communication aspect relevant for all the agents of an application, its modelling should be, in principle, an orthogonal concern with respect to the design of the activities of each agent. Consistently with this view, this work advocates the separate development of, respectively, agent behaviour, and the interaction aspects constituting the context. The latter is first abstractly specified as a set of communication laws, then automatically implemented by a tool that generates the necessary ACC management and checking code from the specification. The appropriate portions of this code should be activated whenever an interaction between agents takes place, so as to ensure that (i) the constraints specified by the laws are respected by the interaction, and (ii) the actions some of the laws require are carried out before the interaction actually occurs. Moreover, this work proposes an infrastructure whereby ACC code is triggered at runtime, whenever agents interact with each other. No source code modification or recompilation is required for this. All is seamlessly accomplished by means of computational reflection, which transparently changes the meaning of the communication primitives normally used by agent programmers. Copyright © 2005 John Wiley & Sons, Ltd.

The real-time specification for Java (RTSJ) introduces a new memory model featuring some programming constraints that impede the "as-is" use of many well known design patterns. In this context, this paper describes and evaluates two design patterns, developed by the authors for distributed real-time Java software, that are able to overcome the limitations imposed by RTSJ. The first pattern, RTJ-leader-follower; is a RTSJ-compliant version of the well-known leader/follower pattern. The second pattern, called scoped tunnels, provides a new communication mechanism for RTSJ threads executing in different and incompatible memory areas, thus making possible the realization of the standard pattern half-sync/half-async for efficient network I/O handling. The paper presents both a qualitative and quantitative evaluation of these patterns, showing, above all, that they are able to provide a safe execution environment

Assistant agents are software systems that help users during activities by carrying out some task as a reaction to events in their environment. This paper proposes a software architecture that allows assistant agents to extend applications by autonomously giving users suggestions and activating useful application functionalities. The connection between applications and assistant agents is realised by means of computational reflection, which allows applications to evolve essentially without changes to their source code. The proposed approach is general in that it does not depend on a specific application nor platform. Its application is demonstrated by two examples of reflective assistant agents for a Web browser supporting data presentation and e-shopping respectively.

Finding the position of a mobile agent in a wide distributed system still represents an open research issue. The paper proposes a naming scheme and a location protocol of general validity for mobile agents able to effectively meet all the typical requirements of mobile agent environments and, thus, easy to integrate into different platforms. The paper identifies the main characteristics which an agent naming scheme and a location protocol of general validity should have, and suggests some properties and parameters to be taken into account to evaluate the effectiveness of naming schemes and location protocols. Then, we propose a "human readable" agent naming scheme based on the distributed environment outlined in MASIF, and a suitable location finding protocol called the Search-By-Path-Chase. Both of them are compared with some of the solutions already provided, using the properties and the parameters suggested. The performances are finally evaluated by means of a set of measurements.

The problem of finding the position of a mobile agent in a wide distributed system still represents an open research area. In fact, although today several mobile agent platforms have been developed, each of them implements an ad hoc naming and location policy, suitably studied to address the specific requirements deriving from the design choices made. What is really needed is a naming and location scheme of general validity, able to effectively meet all the typical requirements of mobile agent environments and thus easy to integrate into different platforms. A first contribution of the paper is the identification of the main requirements which an agent naming scheme of general validity should meet. Then, some properties are suggested to be taken into account, in order to make a qualitative naming scheme classification. The second contribution of the paper is the proposal of a "human readable" agent naming scheme, based on the distributed environment outlined in MASIF, the standardization OMG proposal for mobile agent systems. In addition, two location protocols suitable for the proposed naming scheme are presented, which are particularly appropriate for Internet/intranet environments.