I. INTRODUCTION

Demand response (DR) programs have emerged as one of the most important ways for energy grid operators to reduce energy shortages or excesses and, as a result, enhance system dependability. 1 More crucially, with the increasing prevalence of Renewable Energy Sources (RES),2 Energy Storage Systems (ESS [1]), and even Electric Vehicles (EV3), extra flexible assets are easily accessible to provide new avenues for profitability, maximize current ones, and reduce overall risks. The economic benefits of this new era are not restricted to essential energy players such as grid operators and merchants. End-customers, or individual units, and the different types of Distributed Energy Resources (DERs) placed on their premises are, in fact, playing an increasingly important role, which may be linked to the development of new business models centered on aggregation and virtual power plants (VPP4). Aside from the aforementioned economic benefits, coordinating such a diverse landscape of DERs enables exploiting the underutilized flexibility accessible at lower scales. Indeed, residential and tertiary customers have been recognized as substantial sources of flexibility [2, 3], made even more so by the arrival of prosumers, or consumers who also create energy. For example, in 2016, the EU member states produced more than 33GW
of home solar photovoltaics (PV), 53% of which was transferred to the grid [3]. This is likely to rise as a result of the EU’s strong green energy policy, which aim to attain a 32% proportion of RES by 2030. 5 However, because to the lack of a scalable Information and Communications Technology (ICT) infrastructure capable of managing the sheer bulk of small to medium-scale clients, disaster recovery (DR) programs are currently primarily supplied to big industrial customers. In this context, DR proposals are typically designed to work in a closed world in which new data sources are not expected to appear, and thus they do not consider the necessity of integrating new data sources that rely on different formats, models, or protocols to exchange and consume data with them. Numerous DR data models have been suggested, but only a handful are ontologies that allow for the provision of a semantic interoperable layer for data sharing [4]. Similarly, proposals rely on non-semantic models [5-12], while there has recently been a movement toward building proposals using ontologies [13-18]. However, the majority of these ideas lack natural means for integrating additional data sources, necessitating a significant data harmonisation effort to incorporate new data sources. Furthermore, the DR plans rely on a diverse set of protocols, some of which need infrastructures to openly give their data (HTTP) or broadcast their information in low-security environments (MQTT); in fact, security is a component that is frequently overlooked or ignored in most proposals. The CIM middleware is introduced in this article to handle the practical issues that arise in real- world DR systems. The CIM’s major purpose is to create a private and secure peer-to-peer cloud network enabling disaster recovery systems and data infrastructures to transparently interchange and consume data, despite the fact that the systems and infrastructures use various formats or models. To that purpose, the CIM employs semantic interoperability modules, which enable bidirectional data translation mechanisms to convert data represented in disparate forms or models into a semantically compatible version based on RDF
[19] and an ontology [20], and vice versa. The CIM tool was created in the framework of the European DELTA project, in which semantic interoperable data adheres to the DELTA ontology [21], although it may be used with any ontology. Furthermore, anytime a data payload is transferred, the CIM performs semantic validation of data based on W3C SHACL shapes [22], verifying its validity and coherence with the ontology. The CIM is an Open Source tool6 that implements its semantic interoperability layer [23] using well-established IoT techniques that have been adapted to the DR and the decentralized cloud context. However, the CIM may be utilized in a variety of application domains; it is not limited to ontologies relating to the energy sector. The CIM has been utilized as middleware in the context of DELTA for communications across DR systems and data infrastructure using an edge-cloud architecture. The CIM, on the other hand, goes beyond encouraging the decentralization of DR systems and the integration of distributed data sources by providing them with a distributed, scalable, secure, and end-to-end privacy- preserving peer-to-peer (P2P) network that can be hosted across multiple cloud providers and ensures service liveness even in the face of failures. The CIM enables the unification of discrete real-world corporate DR systems without needing them to disclose any data through public channels. A careful experiment was put out to demonstrate the features of the CIM. First, the accuracy of the CIMs’ semantic compatibility has been established. The capacity of the CIMs to respond to concurrent requests while exchanging ordinary payloads or payloads that must be translated in order to be semantically compatible. Finally, the ability of CIMs to exchange larger payloads has been proven.
The remainder of this article is organized as follows: Section 2 analyzes current DR proposals in the literature, Section 3 elicits practical challenges that DR proposals must address, Section 4 reports the CIM tool, providing an insight into its architecture and functionalities, Section 5 explains the experiments performed to validate the CIM, as well as the results obtained, and Section 6 summarizes our findings and conclusions.