The project is divided into work packages according to these objectives:


WP1. Abstract Modelling of Adaptive Smart Areas

Task 1.1 State of the Issue to be addressed

D 1.1: State of the art about Adaptive Smart Areas

Task 1.2 Requirements and analysis of the proposed case studies

D 1.2: Model of use case scenarios

D 1.3: Model of components and relationships

Task 1.3 Abstract Model for COSASS

D 1.4: COSASS Abstract Model

Task 1.4 Analysis of KPIs

D 1.5: List of KPIs for Adaptive Smart Areas

WP2. Data Acquisition Infrastructure

Task 2.1. Study and implementation of low-energy IIoT infrastructures

D 2.1: Technical report of IIoT infrastructures

D2.2: IIoT infrastructure designed

Task 2.2. Edge Computing technologies for information collecting and processing at the network edge

D 2.3: Machine learning models to be run at the edge of the network designed

Task 2.3. Software Defined Virtual Edge-IoT Networks for dynamic reconfiguration of network resources

D 2.4: Report of network devices in configuration costs of IIoT networks

D 2.5: Design of SDN and Deep Reinforcement Learning algorithms for dynamic reconfiguration of network

WP3. Edge decision making

Task 3.1. ETL layers and Smart data processing close to RT

D 3.1: Design of ETL and Smart Data technologies

Task 3.2. Low level decision making on the Edge

D 3.2: Design of AI Algorithms

Task 3.3. Federated Learning and MA-DRL mechanisms for distributed training of models

D 3.3: Multi-agent deep reinforcement learning mechanisms designed

D 3.4: Identification of external systems and sources

WP4. Cloud decision making

Task 4.1 Multi-agent coordination for limited real-time resource sharing

D 4.1: A set of protocols for sharing time-restricted limited resources.

D 4.2: Analysis of the time-boundaries of the set of protocols defined.

Task 4.2 Digital Twins and Federated Digital Twins (DT-FDT) module

D 4.3: DT-FDT design specification and DT-FDT module implementation

Task 4.3. Federated-like learning

D 4.4: A set of Federated-like Learning algorithms for the cloud level.

D 4.5: A set of Federated-like Learning algorithms to interact with the edge level of the system

WP5. AI-enabled continuum from Cloud to Edge

Task 5.1 Formalisation of the cloud-edge context

D 5.1: Report with specification of formal model for cloud-edge context and selection/instantiation methodology

Task 5.2 Cloud-edge context monitoring and event management

D 5.2: Report with detection of relevant cloud-edge context change

Task 5.3 Adaptation of cloud-edge continuum to contingencies

D 5.3: Report with adaptation methods for edge-cloud continuum contingencies

Task 5.4 Design and development of cloud-edge self-adaptable decision making processes

D 5.4: Report with cloud-edge self-adaptable decision making methods

WP6. Trustworthy AI Framework

Task 6.1 Legal and ethical requirements

D 6.1: Report with ethical and legal requirements for trustworthy AI solutions in Adaptive Smart Areas

Task 6.2 Legal compliance

D 6.2: Report with legal compliance study of proposed solutions

Task 6.3 Decision explanation and argumentation

D 6.3: Report with explainability needs

D 6.4: Framework for explanation generation of cooperative decision making solutions

WP7. Use cases (Viticulture)

Task 7.1. Adaptive Smart Area: Continental Vineyard

D 7.1: Functional and technical requirements report

D 7.2: Design of the edge/IoT model and algorithms specification

D 7.3: Final outcomes

Task 7.2. Adaptive Smart Area: Mediterranean Vineyard

D 7.4: Microclimate characterization report

D 7.5: Design of the Federated-like algorithm

D 7.6: Final outcomes

Task 7.3 Agriculture field monitoring and vehicle coordination

D 7.7: Simulation specification and evaluation report