Homomorphic encryption for privacy-preserving data aggregation in data spaces.

Homomorphic encryption for privacy-preserving data aggregation in data spaces - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Data Brief . 2026 May 21;66:112882. doi: 10.1016/j.dib.2026.112882 Homomorphic encryption for privacy-preserving data aggregation in data spaces Gorka Calvo a , Anhelina Kovach a, ⁎ , Jorge Lanza b , Leticia Montalvillo a , Aitor Urbieta a Author information Article notes Copyright and License information a IKERLAN Technology Research Centre, Basque Research and Technology Alliance (BRTA), Mondragón 20500, Spain b Network Planning and Mobile Communications Laboratory, Universidad de Cantabria, Santander 38005, Spain ⁎ Corresponding author. akovach@ikerlan.es Received 2025 Sep 30; Revised 2026 Mar 11; Accepted 2026 May 18; Collection date 2026 Jun. © 2026 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). PMC Copyright notice PMCID: PMC13234483  PMID: 42254573 Abstract Privacy preservation in data spaces is a critical requirement for enabling trusted, cross-organizational data sharing aligned with European data governance principles. To address this gap, we present a technically enforced privacy-enhancing solution based on homomorphic encryption (HE) to enable privacy-preserving data aggregation. We built upon a Consumer-Aggregator-Provider (CAP) data computation architecture and implement a dedicated privacy module leveraging Eclipse Dataspace Components (EDC) and the Microsoft SEAL library to add HE capabilities to data spaces. Our solution is compliant with the Data Space Protocol (DSP) and is integrated within data spaces adhering to the International Data Spaces Association (IDSA) reference architecture. To demonstrate its practical applicability, we present a use case for energy consumption data aggregation by geographical zones and an analysis of the solution’s performance and scalability. Our results demonstrate the feasibility and effectiveness of HE for privacy-preserving data aggregation in data spaces while identifying current limitations. Keywords: Data space, Homomorphic encryption, Privacy-enhancing technologies, Energy data space, Microsoft SEAL, Eclipse Dataspace Components 1. Introduction The digital economy increasingly recognizes data as a critical asset, fueling innovation and new business models [ 1 , 2 ]. Data spaces have emerged as collaborative ecosystems enabling secure data sharing across organizations under common policies and technical frameworks [ 3 ]. They empower stakeholders to maximize the value derived from data analysis and generate insights that would otherwise remain hidden in isolated data silos [ 4 , 5 ]. This approach is driving advances in health [ 6 ], manufacturing [ 7 ], and energy [ 8 , 9 ], a vision that aligns with the European Union (EU)'s European Strategy for Data [ 10 ], which envisions Common European Data Spaces across key sectors while ensuring European values are respected. However, realizing the full potential of data spaces requires addressing fundamental challenges in privacy preservation [ 11 ] and data sovereignty [ 12 , 13 ]. Particularly when sensitive data must be shared and processed across organizational boundaries while maintaining confidentiality and compliance with regulations such as the General Data Protection Regulation (GDPR) [ 14 ]. In traditional data-sharing models, data providers lose control over their assets once they are shared with third parties, compromising their data sovereignty and thus discouraging participation [ 15 , 16 ]. This challenge is particularly significant in the energy sector, where organizations must balance the benefits of data collaboration against privacy risks [ 9 ]. Energy data aggregation combining individual consumption patterns, grid operational metrics, and infrastructure data can inadvertently expose sensitive information about households, businesses, or critical infrastructure [ 17 ]. These privacy concerns create barriers to participation in data-sharing initiatives, ultimately hindering the sector's digital transformation and its contribution to energy transition goals. To address these privacy challenges, this work proposes a paradigm shift towards technologically enforced privacy through Privacy-Enhancing Technologies (PETs) [ 18 ], particularly Homomorphic Encryption (HE) [ 19 , 20 ]. Unlike traditional contractual trust models [ 21 , 22 ], HE enables collaborative data analysis without exposing plaintext information to any party, ensuring sensitive data remains under its owner's control throughout the processing pipeline [ 11 , 23 ]. This approach transforms data sharing from a trust-based model to a cryptographically guaranteed privacy-preserving framework. In this work, we present the design and implementation of an HE-based privacy module tailored for secure data aggregation within data space architectures. The solution maintains compatibility with existing data space frameworks, such as the International Data Spaces (IDS) Reference Architecture Model (RAM) [ 1 ] and Gaia-X [ 24 ] through adherence to the Data Space Protocol (DSP) [ 25 ], ensuring its integration with reference data space connector implementations like the Eclipse Dataspace Components (EDC) [ 26 ]. The main contributions of this work are listed as follows: ▪ The definition of the Consumer-Aggregator-Provider (CAP) computational architecture, specifying roles and interaction patterns to enable privacy-preserving data sharing. ▪ The design and implementation of a HE-based privacy module supporting multiple HE schemes and ensuring compatibility with the DSP and data space connectors. ▪ A proof-of-concept deployment, demonstrating practical feasibility through integration with data space connectors, showcased in an energy consumption data aggregation use case. The work carried out, known as PETSpaces, is included as part of the Data Space Support Centre (DSSC) Toolbox, 1 demonstrating both its practical validation and its relevance for the data space community. The remainder of this paper is organized as follows. Section 2 introduces fundamental concepts of European data spaces, PETs, and HE, followed by a review of related work. Section 3 presents the CAP architecture design and privacy module specification. Section 4 details the implementation and integration of the privacy module within data spaces. Section 5 presents an energy consumption data aggregation use case. Section 6 analyses the privacy module, while Section 7 discusses limitations and Section 8 concludes the paper. 2. Background This section introduces the foundational concepts underlying our proposed solution: European data spaces as frameworks for secure data collaboration, PETs for trusted sharing, and HE as our core cryptographic solution. We conclude with a review of related work. 2.1. European data spaces The EU is developing common European data spaces as frameworks that integrate data infrastructures with robust governance mechanisms to facilitate data sharing [ 10 ]. The objective is to establish ``EU-wide common, interoperable data spaces in strategic sectors that overcome existing legal and technical barriers to data sharing'' . This initiative aims to unleash the potential of data-driven innovation, fostering a single market for data that enhances Europe's global competitiveness and reinforces its data sovereignty. Some key features of these spaces are: (1) openness for participation, (2) secure and private infrastructures, (3) fair access rules, and (4) adherence to EU values, particularly personal data protection (e.g., GDPR [ 14 ]). This framework is supported by regulations such as the Data Governance Act (DGA) [ 27 ] and the Data Act [ 28 ], which aim to increase data availability and trust in data sharing. The significant potential of data spaces has led to various initiatives promoting their development and adoption. Prominent examples include the International Data Spaces Association (IDSA) 2 [ 1 ], Gaia-X 3 [ 24 ], and FIWARE 4 [ 29 ]. Fig. 1 illustrates the reference architecture aligned with the IDS RAM, highlighting participating roles, their interactions and core processes. Fig. 1. Open in a new tab Data space reference architecture, participating roles and processes. Based on the IDS-RAM [ 1 ]. In support of these initiatives, implementations of data space connectors such as the EDC [ 26 ] have been developed to provide reference technology for data space integration. These implementations typically adopt and extend the DSP [ 25 ], which is maintained and supported by the IDSA to ensure interoperability and compliance with established data space standards. 2.2. Privacy-Enhancing technologies (PETs) PETs are techniques and tools designed to protect sensitive information by minimizing data exposure, maximizing security, and enabling individuals to retain greater control over their personal data [ 11 ]. Their fundamental purpose is to facilitate the collection, analysis, and sharing of data while ensuring confidentiality and privacy, thereby reducing the risk of unauthorized disclosure or misuse [ 30 ]. While originally developed to safeguard personal information, PETs have broadened their scope to protect a wide range of sensitive data types, including intellectual property and trade secrets [ 31 ]. PETs achieve these goals by ensuring that sensitive data is never shared or processed in a directly readable form, often employing techniques such as encryption or anonymization. The integration of PETs within data spaces creates a synergistic relationship where PETs extend the capabilities and use cases of data sharing by enhancing privacy guarantees, while data spaces offer a scalable and interoperable environment to deploy PET solutions effectively [ 18 ]. 2.3. Homomorphic encryption (HE) Among various PETs, HE stands out as a leading cryptographic method for preserving data integrity and confidentiality [ 32 ]. HE enables computations to be performed directly on encrypted data without requiring decryption. Remarkably, the results of these encrypted computations, once decrypted, are identical to those obtained if performed on the original unencrypted data, while the output remains encrypted. Formally, this property is expressed as follows [ 11 ]. E ( m 1 ) ▵ E ( m 2 ) ≡ E ( m 1 ▵ m 2 ) ∀ m 1 , m 2 ∈ M where applying an operation ▵ on ciphertexts is equivalent to performing the same operation on plaintexts followed by encryption, allowing for potential minor errors introduced by some HE schemes. This capability addresses a critical vulnerability in traditional data processing, where data must be decrypted and thus exposed to risk during use or transit [ 33 ]. HE enables secure collaboration by allowing sensitive or proprietary data to be shared and analyzed without exposing the raw information [ 34 ]. In sectors such as energy or healthcare, HE can support secure internal data analysis to identify trends and anomalies without exposing confidential customer information and facilitates outsourcing data analytics to cloud providers while maintaining strict data confidentiality. There are three main types of HE [ 19 ]: (1) Partially Homomorphic Encryption (PHE) allows only one type of operation (addition or multiplication) an unlimited number of times, and is the simplest form, (2) Somewhat Homomorphic Encryption (SHE) extends this by allowing both additions and multiplications on encrypted data but only a limited number of times, and (3) Fully Homomorphic Encryption (FHE) is the most powerful, enabling both operations an unlimited number of times on encrypted data, making it the most versatile option [ 34 ]. An essential aspect of FHE is the noise budget , the margin of error introduced into ciphertexts during encryption and computation. Each operation increases this noise, and once the budget is exhausted, the ciphertext becomes corrupted. While PHE and SHE are inherently limited by noise growth, FHE employs techniques to refresh ciphertexts and restore the budget. However, since this process is computationally expensive, careful noise management is required. Although FHE currently faces challenges like high computational overhead [ 35 ] and significant resource demands [ 36 ], ongoing research is steadily improving its efficiency and practical applicability [ 20 ]. Considering the need for broad, long-term use cases in secure European data spaces, FHE is the most suitable choice, as it provides the flexibility required to support diverse and evolving data-driven applications while maintaining strict data confidentiality. 2.4. Related work A significant body of research has established the integration of PETs as a fundamental approach for safeguarding privacy in data spaces [ 11 , 18 ]. Consequently, various PET-based methods have been investigated to address privacy challenges. The most prominent approaches in the literature include: ▪ Federated Learning (FL) enables multiple participants to collaboratively train a shared machine learning model without centralizing or exposing raw data. This approach is particularly beneficial in data spaces to foster collective intelligence while preserving data sovereignty [ 37 ]. For instance, Mitrovska et al. [ 38 ] demonstrate a practical FL framework leveraging the EDC connector. ▪ Differential Privacy (DP) introduces calibrated noise into data or query outputs, providing formal mathematical guarantees that limit the disclosure risk of individual records. While DP can be employed within broader anonymization processes, it is itself a PET that enables privacy-preserving data analysis and publication in data spaces without requiring full trust among participants [ 18 , 39 ]. ▪ Secure Multi-Party Computation (SMPC) involves cryptographic protocols that enable joint computation on private inputs without revealing them, essential for untrusted collaborative computations in data spaces. In this context, Siska et al. [ 40 ] analyse SMPC and data spaces integration challenges, including identity, access control, and standardization. ▪ Zero-Knowledge Proof (ZKP) is a cryptographic protocol that allows one party to prove possession of certain information or the validity of a statement to another party without revealing the underlying data itself. In data spaces, ZKPs can enable verifiable data exchanges and access control while preserving confidentiality [ 41 , 42 ]. ▪ Trusted Execution Environment (TEE) provides isolated enclave where computations can be performed on sensitive data, ensuring that neither the host system nor external parties can access the data during processing. TEEs offer integrity and confidentiality guarantees, making them suitable for privacy-preserving computation in multi-party data space scenarios [ 43 ]. While each of these approaches offers valuable privacy guarantees, their suitability for privacy‑preserving data aggregation depends critically on the computational task and interaction model. FL is designed for collaborative model training and is not designed to support direct numerical aggregation of raw business values such as meter readings. DP provides formal privacy guarantees by injecting calibrated noise into query results [ 44 ], but the privacy budget is gradually depleted with repeated queries, and the resulting accuracy degradation [ 45 ] is incompatible with precision‑sensitive use cases like energy metering, where exact aggregation outcomes are required for billing. SMPC yields strong cryptographic confidentiality without a trusted intermediary and can compute exact aggregates, but it relies on interactive protocols among multiple parties and assumes that at least a threshold of participants does not collude. In practice, this increases protocol complexity, communication overhead, and operational coordination requirements, especially when many small data providers participate in a data space. ZKPs, in turn, allow parties to prove the correctness of computations or compliance with constraints without revealing underlying inputs, but they do not by themselves enable computation on hidden data and therefore act as a complementary building block rather than a standalone solution for numerical aggregation. TEEs provide hardware‑enforced confidentiality for computations inside enclaves, yet their guarantees ultimately depend on the trustworthiness of the hardware vendor and the absence of exploitable side channels [ 46 ]. HE, by contrast, enables an aggregator to operate directly on ciphertexts and derive aggregation results that are equivalent to those that would have been obtained on the underlying plaintexts. In this context, HE provides a cryptographic, rather than policy-based or hardware-based, guarantee of confidentiality. This combination of properties makes HE particularly well-suited for the data aggregation use case addressed in this paper. Table 1 summarizes these considerations by contrasting these PET approaches in the context of data-space-based data aggregation. Table 1. Comparison of PETs for privacy-preserving data aggregation in data spaces. Criteria HE FL DP SMPC ZKP TEE Primary purpose Computation on encrypted data Collaborative model training Privacy gain through noise injection on data Decentralized joint computation Verifiable proofs of statement Hardware-isolated computation Applicable to data aggregation Yes No Yes (with noise) Yes (with interaction) No Yes (with hardware trust) Aggregation accuracy Exact ⁎ N/A Approximate (noise) Exact N/A Exact Primary privacy-utility trade-off Computational overhead N/A: Limited to model training tasks Noise degrades aggregation accuracy Complexity grows with number of parties and circuit depth N/A: Does not enable computation Hardware dependency Open in a new tab ⁎ Additive and many lattice‑based schemes (e.g., BFV/BGV) are indeed exact for integer arithmetic; approximate schemes like CKKS introduce a controlled rounding error [ 47 ]. The application of HE in data spaces has been investigated as a powerful tool for performing computations on encrypted data [ 20 ]. Some studies demonstrate the theoretical benefits of applying HE in data spaces [ 11 , 48 ], but there is a lack of those providing concrete implementations within functional data space environments. The closest effort is that of Michikata et al. [ 49 ], who propose an HE-based solution and allude to its potential for deployment within data spaces, yet do not elaborate on the practical steps for this integration, leaving the implementation details ambiguous. Despite the recognized potential of HE in data spaces, to the best of our knowledge, no prior work has presented a detailed implementation of a data space that utilizes HE for privacy-preserving data aggregation in the manner proposed in this paper. Our work addresses this gap by delivering a functionally implemented solution aligned with the IDS RAM [ 1 ] and IDS Information Model [ 50 ], and compatible with the DSP [ 25 ]. Consequently, our privacy module can be seamlessly integrated within data space infrastructures based on reference connectors such as the EDC [ 26 ]. 3. Solution design This section presents the design of our privacy-preserving solution for energy data spaces. We first introduce the CAP computational architecture, then we specify the functionalities of the privacy module and detail the operational workflow that integrates HE capabilities into the DSP. 3.1. Consumer-Aggregator-Provider (CAP) architecture The CAP architecture structures data aggregation in our data space. Consumers, aggregators, and providers collaborate to perform computations on homomorphically encrypted data, preserving privacy throughout. Fig. 2 illustrates the roles and interactions within this architecture. ▪ Consumer : This is the final data consumer. Their primary role is to generate the encryption keys (public key K P U B and private key K P R I V ) (1) and to find an aggregator to transmit the K P U B . At the end of process (4), they just have access to the aggregated computation of the providers' data, which they can decrypt with K P R I V . ▪ Aggregator : In our model, the aggregator is a technical role within the data space, operated by a trusted and accredited entity. It is responsible for receiving encrypted inputs and performing homomorphic aggregations. It is not, by itself, assumed to be a data intermediation services provider in the sense of the DGA [ 27 ], although such a service provider could operate one or more aggregators in practice. It needs to find data providers to distribute the public encryption key K P U B (2) so that they can encrypt their own data. Once it receives the data, it performs the computation with the providers' encrypted data and sends it to the consumer (3). Since the aggregator only knows the K P U B provided by the consumer, it unable to decrypt the data from the providers. ▪ Provider: This is the entity that possesses data of interest to the consumer. It encrypts its data with K P U B (3) and sends it to the aggregator. Fig. 2. Open in a new tab Scheme detailing the CAP architecture composed of a consumer, an aggregator, and N providers. The interactions among these components are numbered sequentially from 1 to 4. 3.2. Privacy module While the CAP architecture defines a conceptual model for secure computation, its practical implementation within a data space requires additional functionality, due to the ecosystem lacking native support for HE operations. To address this, we present the privacy module, which adds HE functions to data spaces, allowing the CAP architecture to be effectively demonstrated and tested. Each stakeholder (consumer, aggregator, provider) runs their own instance of the privacy module. This distributed deployment ensures that sensitive operations such as key generation, encryption, and decryption are executed securely and under the exclusive control of each participant. The privacy module, instantiated at each site, provides the following cryptographic operations: ▪ Cryptographic key generation. A consumer’s privacy module generates the asymmetric key pair needed for HE: a public key ( K P U B ) for encryption and a private key ( K P R I V ) for decryption. The private key never leaves the consumer’s environment, maintaining end-to-end security. ▪ Data encryption and decryption. Providers use their local privacy module to encrypt data with K P U B , while the consumer uses their module to decrypt the aggregated result with K P R I V . ▪ Homomorphic computation. An aggregator’s privacy module performs mathematical operations (such as summation) directly on encrypted data, producing a ciphertext result that, once decrypted by the consumer, reveals the correct computation on the original plaintext. Leveraging these functions, the aggregator computes on encrypted data. Which, in our case, consists of a summation of all providers input data. This process can be implemented following two different strategies: waiting for all encrypted values at once and performing one large-scale summation or adding two encrypted values at a time each time they are received. We adopt the second approach as it allows partial sums to be computed as data arrives and avoids the large payload sizes that result from the aggregation all encrypted data at once. 3.3. Integration within data space architecture Building on the CAP architecture and the privacy module, this section describes how these components are integrated within the data space infrastructure. As shown in Fig. 3 , the privacy module is deployed by every participant, enabling privacy-preserving data computation without affecting core data space processes. Participants interact with both the privacy module and the data space connector through a User Interface (UI), facilitating user engagement with the system. Fig. 3. Open in a new tab Data space architecture showing the interaction between three types of participants: consumer, aggregator, and provider. Every participant deploys a dedicated connector for data space access. Fig. 4 illustrates the coordinated sequence of actions we define among consumers, aggregators, and providers. Each participant leverages their privacy module to perform key generation, data encryption, encrypted computation, and decryption. This workflow guarantees that: (1) data remains encrypted during transfer, (2) the consumer can decrypt the aggregated result, (3) privacy of individual provider data is retained, and (4) the aggregator processes only encrypted data without access to plaintext. Fig. 4. Open in a new tab Operational flow depicting the application of the CAP architecture within a data space, illustrating the coordinated steps from 0 to 5 among consumer, aggregator, and provider participants. Before applying the CAP architecture within the data space, preliminary steps regarding asset upload and contract negotiation must be completed. These steps are designated as step 0 in the operational flow diagram shown in Fig. 4 : ▪ Providers publish data offers with associated usage policies using the Open Digital Rights Language (ODRL) specification, 5 including a field specifying that the data must be shared encrypted using HE (see Fig. 5 ). ▪ The aggregator first negotiates with providers for their offered data and then publishes an offer presenting the aggregated data from all participating providers. ▪ The consumer identifies relevant offers from an aggregator and initiates contract negotiations. Fig. 5. Open in a new tab Example of a data space policy definition in ORDL that specifies the use of HE, under the BFV scheme. The remaining operational steps (1–5), correspond to the CAP architecture shown in Fig. 3 with step 5 extending beyond that figure. These are summarized as follows: ▪ Step 1: The consumer generates a pair of encryption keys through the privacy module, transmitting the public key K P U B to the aggregator, while the private key K P R I V remains local. ▪ Step 2: The aggregator distributes K P U B to participating providers. ▪ Step 3: Upon receiving K P U B , providers encrypt their data via the privacy module and send the encrypted data to the aggregator. ▪ Step 4: The aggregator performs the agreed computation on encrypted data and sends the encrypted result to the consumer, remaining unable to decrypt provider inputs. ▪ Step 5: The consumer decrypts the aggregated result using K P R I V , thus obtaining the computation outcome without accessing individual provider data. 4. Implementation This section details the implementation of our proposed privacy module, which code is available in our repository. 6 It also describes the deployment of the data space infrastructure, and explains how both are integrated to enable secure, privacy-preserving data exchange and computation. 4.1. Data space A data space facilitates interactions between various participants. While traditionally these are data providers and consumers, our solution also incorporates an aggregator. Each of these participants requires a connector to act as their gateway to the data space [ 1 ]. For the implementation of our data space infrastructure, we selected the EDC, 7 an open-source solution widely recognized in the literature as one of the most prominent open-source data space connectors available today [ 51 ]. The EDC connector provides an extensible implementation of the DSP, providing processes as catalog management, contract negotiation and data transfer. Among various EDC-based implementations, we deploy the Minimum Viable Dataspace (MVD). 8 As seen in Fig. 6 , the MVD includes: the EDC connector that acts as the gateway for participants; a Data Dashboard providing an UI for managing policies, contracts, and data transfers; a Data Storage repository; an Identity Provider for connector identity management; and a Catalog Service that enables providers to publish offerings and consumers to discover them. Fig. 6. Open in a new tab General architecture of the MVD, showing interactions between EDC connectors and main components. The processes among these are marked (1 to 4). 4.2. Privacy module This module is developed in C++, using Microsoft Simple Encrypted Arithmetic Library (SEAL) 9 to implement HE computation. The choice of using SEAL is motivated by its extensive documentation [ 36 ] and the numerous academic studies that validate its performance, often showing superior efficiency in comparison with other HE libraries [ [52] , [53] , [54] ]. SEAL supports both the Brakerski-Fan-Vercauteren (BFV) [ 55 ] (for integer operations) and the Cheon, Kim, Kim, and Song (CKKS) [ 47 ] (for floating-point operations) homomorphic schemes. The privacy module exposes a RESTful Application Programming Interface (API) through the Pistache 10 framework to enable its integration with the data space. The functionality of the module is accessible through a set of REST endpoints that operate with JSON payloads and include a sealContext block specifying cryptographic parameters (see Fig. 7 ). Table 2 provides a summary of these endpoints. Fig. 7. Open in a new tab HE configuration parameters: schemeType indicates the HE scheme to use, polyModulusDegree specified the polynomial modulus degree, and coeffModulus denotes the coefficient modulus to apply. Table 2. Description of the endpoints implemented in the HE-based privacy module, detailing their functions, required inputs, and produced outputs. Endpoint Description Input Output /homomorphic/generate-keys Generates a pair of HE keys for the given parameters sealContext Public Key Secret Key /homomorphic/encrypt Encrypts given data with the given parameters sealContext Public Key Plaintext (Data) Ciphertext /homomorphic/decrypt Decrypts the given data with the given parameters Secret Key Ciphertext Plaintext /homomorphic/add ⁎ Adds 2 ciphertexts sealContext Ciphertext1 Ciphertext2 Ciphertext(1 + 2) Open in a new tab ⁎ It can be any computation function. A key aspect of the module's implementation is the management of encryption parameters, which are specified within the JSON requests as shown in Fig. 7 . Among these, the polynomial modulus degree and coefficient modulus selection are particularly critical as they define the initial noise budget for the HE schemes. As previously mentioned in Section 2 , the noise budget is fundamental in HE. If accumulated noise surpasses the threshold set by these parameters, the ciphertext becomes corrupted, preventing accurate decryption of the original data. Therefore, careful tuning of the polynomial modulus degree and coefficient modulus is required to maintain sufficient noise budget, ensuring all required operations can be performed securely. 4.3. Integration within data space The privacy module is integrated within the MVD infrastructure, exposing its HE functionalities via RESTful APIs consumed by MVD components to coordinate encryption, decryption, and computation tasks. To enhance user interaction and workflow management, the MVD’s UI was extended with new functionalities specifically supporting HE capabilities. The added ``PETSpaces'' tab, illustrated in Fig. 8a , acts as a centralized dashboard where participants can manage these contracts, perform cryptographic tasks, and monitor progress. Each participant may assume any CAP role, allowing flexible workflows where participants can act as multiple roles simultaneously. During the contract negotiation and computation phases, the UI continuously reflects the current state of the operation workflow, providing real-time feedback on operation status such as key generation, data encryption, encrypted data transfer, and result decryption. Fig. 8. Open in a new tab Extended UI, supporting each phase of the defined workflow, from the initial data offering to the final decryption on the consumer side. The provider’s interface (see Fig. 8b ) allows the encryption of data using the public key created by the consumer. Meanwhile, the consumer’s interface presents the transfer history, enabling retrieval and decryption of the aggregated encrypted results received from the aggregator (see Fig. 8c ). 5. Use case The proposed CAP architecture and privacy module are designed to be domain-agnostic and applicable across a range of sectors requiring HE computation . This approach can support privacy-preserving data analytics in healthcare for patient data aggregation, financial services for confidential transaction analysis, and smart manufacturing for industrial Internet of Things (IoT) data processing. Specifically, to illustrate practical applicability, we present a detailed use case within the domain of energy consumption data aggregation , where privacy concerns are especially pronounced. Energy consumption data collected at a granular level can reveal sensitive information about individual households, including behavioural patterns, occupancy, and even health-related activities [ 56 , 57 ]. The unauthorized disclosure or misuse of such data poses significant privacy risks, such as profiling and targeted surveillance [ 58 ]. Consequently, robust privacy-enhancing mechanisms are paramount to enabling beneficial data analysis while safeguarding individual privacy [ 59 ]. In our use case, individual households act as data providers, generating detailed electricity consumption data. Aggregators collect data within defined geographic clusters, such as neighbourhoods, where they aggregate information while enforcing privacy policies by applying HE techniques. Consumers, which can be energy companies or municipal authorities, receive aggregated data to analyse consumption patterns, forecast demand, and make informed decisions, all while ensuring that individual data remains confidential. This expanded architecture is depicted in Fig. 9 , illustrating the multi-aggregator data flow from providers to consumers. Fig. 9. Open in a new tab Expanded CAP architecture illustrating a realistic implementation scenario for the energy data aggregation. Aggregators (1 to W) are associated with a group of individual data providers. To validate our use case, we deploy three EDC connectors, each assigned one of the CAP architecture roles (provider, aggregator, consumer). The connectors and privacy modules are deployed within Docker containers to simulate a realistic distributed environment, with each participant running its own instance of the privacy module (as described in Fig. 3 ). We simulate a simplified aggregation scenario where the provider sends multiple encrypted assets, each tied to different assets and contracts, effectively emulating multiple providers. For data, we use a publicly available electricity consumption dataset, 11 adapted to extract instantaneous hourly consumption values. A set of videos demonstrating the complete process is available in our repository. 12 While the energy consumption scenario provides a concrete validation of the proposed architecture, the same CAP model is directly transferable to other sectors where privacy-preserving data aggregation is required. In healthcare, individual hospitals or clinics could act as providers contributing encrypted patient data such as disease incidence rates, treatment outcomes, or resource utilization metrics to an aggregator, enabling regional or national health authorities (consumers) to derive population-level insights without accessing individual patient records, thereby supporting compliance with strict medical data regulations. In financial services, banks or insurance companies could contribute encrypted transaction volumes or risk indicators, allowing sector-wide aggregation for regulatory reporting, fraud trend detection, or systemic risk assessment, where the confidentiality of each institution's data is paramount. In smart manufacturing, industrial IoT devices generating production metrics, defect rates, or energy consumption at factory level could be aggregated across supply chain partners, enabling collaborative optimization while protecting proprietary operational data. In each of these scenarios, HE provides the same fundamental advantage demonstrated in our energy use case: accurate computation over private data without requiring any party to expose plaintext information. 6. Analysis We perform our analysis to evaluate how the inclusion of the privacy module affects a standard data space, specifically measuring the latency it introduces into the workflow. To this end, two tests were performed: (1) measuring the average response time of the privacy module’s endpoints, and (2) analyzing how aggregation time increases as the number of providers grows. The experimental setup was based on a virtualized environment designed to simulate the deployment of the proposed CAP architecture. The execution environment was a virtual machine configured in Oracle VirtualBox, running Ubuntu 24.04 LTS with 4 CPU cores and 9800 MB of RAM. This virtual machine was hosted on a Dell Latitude 5520 laptop equipped with an Intel Core vPro i7 processor and 16 GB of RAM. The privacy module was deployed inside a Docker container. For evaluation, we used the publicly available electricity consumption dataset mentioned in Section 5. We employ the encryption configuration parameters shown in Fig. 7 , which, according to the Microsoft SEAL documentation, are sufficient to perform the required operations with a CKKS HE scheme without incurring excessive computational overhead. As shown in Fig. 10 , the results obtained are adequate for the aforementioned scenario, however they should be considered for each use case. The measured times for key generation, encryption, and decryption are relatively short and less impactful compared to the overall latency experienced during the workflow with multiple providers. In scenarios where a user interacts with the privacy module through the UI, HE operation times between 0.08 and 0.12 s are negligible compared to the longer delays typically experienced as the user completes each step in the workflow. The main source of delay arises during the addition operations, especially as the number of providers increases, leading to a higher volume of requests to reach the final sum. This scalability issue, prominently influenced by the accumulation of addition times, is further analysed in the following paragraph. Fig. 10. Open in a new tab Average time taken for key generation, encryption and decryption operations, illustrating the computational overhead of the HE module. Fig. 11 illustrates the results obtained from our tests with providers ranging from 2 to 100. We found that the computation time is linearly correlated with the number of participants, which is a significant performance cost. This drawback, however, is balanced by a crucial privacy advantage: more providers mean less influence from any single participant on the final aggregated result. Fig. 11. Open in a new tab Cumulative addition time vs. number of providers, showing how the aggregator’s computational time increases as more providers participate in the addition process. 7. Discussion Our proposed solution aligns with privacy-by-design principles [ 60 ], effectively minimizing data exposure through continuous encryption within the data space workflow. A key benefit of this design is the CAP architecture, which balances the extraction of aggregated insights with individual privacy by ensuring that decryption keys remain exclusively with the consumer, preventing aggregators from accessing plaintext data. Building on this architecture, the dedicated privacy module enhances modularity and flexibility, facilitating seamless integration with DSP-compatible data space connectors. Within this framework, the aggregator is designated as a trusted and accredited entity responsible for homomorphic computations on encrypted data. While in our solution HE ensures that the aggregator cannot decrypt individual data, it does not inherently prevent a malicious or compromised aggregator from manipulating the encrypted data (e.g., omitting data, performing incorrect operations), introducing a single point of trust regarding computational integrity. This shifts the focus to the verification of the aggregator’s actions. Notably, a technical challenge in employing HE is the management of the noise budget (see Section 2.3 ), which increases with each operation on encrypted data. If this noise surpasses a limit, the data cannot be decrypted correctly. Different operations have varying noise requirements. In particular, homomorphic multiplications are significantly more costly in terms of noise consumption than additions. Consequently, operations relying primarily on additions, such as summation or averaging, require considerably less noise budget than those involving multiplications, such as computing a standard deviation, which requires squaring operations on ciphertexts. More complex operations thus demand larger noise budgets, which in turn increase both computational cost and ciphertext size. Achieving an optimal balance between these factors is essential to maintain both functionality and performance in practical deployments. A promising approach to mitigate these limitations is the combined use of complementary PETs. The trusted aggregator concern could be addressed by integrating ZKPs, which would allow the aggregator to provide verifiable proof that the computation was performed correctly and on the complete set of provider inputs, without revealing the underlying data. This combination would preserve the non-interactive advantages of HE while adding a cryptographic layer of computational integrity verification, at the cost of additional proof generation and verification overhead. Similarly, the reliance on a single aggregator could be reduced by incorporating SMPC to distribute the aggregation task across multiple parties, eliminating the single point of trust but introducing increased communication complexity and the requirement for synchronous coordination among aggregators. Regarding the noise budget limitation, hardware acceleration through TEEs could complement HE by offloading noise-intensive operations to secure enclaves, although this would reintroduce hardware-level trust assumptions that HE alone avoids. Each of these combinations entails a trade-off between the privacy and integrity guarantees gained and the additional computational, communication, or trust costs introduced, and identifying the optimal configuration remains use-case dependent. 8. Conclusions We develop an HE-based privacy solution integrated within DSP-compatible data spaces to technically enforce data privacy and sovereignty in data aggregation scenarios . The paper introduces its design, implementation, and validation. The work delivers three fundamental contributions: the design of the CAP data computing architecture, the development of a privacy module based on HE, and the definition of an operational flow to enable its integration into a data space. Together, these elements represent an important advancement in the evolution of secure data spaces. By effectively addressing the critical challenge of data privacy, this solution demonstrates the feasibility of secure sharing of sensitive information, laying a strong foundation for future larger-scale implementations. Looking forward, the combined use of complementary PETs, particularly the integration of ZKP for verifiable computation, SMPC for decentralized aggregation, and TEE-based hardware acceleration represents the most promising direction for addressing the current limitations of our solution. Further research is needed to evaluate the practical feasibility and performance impact of these combinations within operational data space environments. Ethics Statement The authors have read and followed the ethical requirements for publication in Data in Brief and confirmed that the current work does not involve human subjects, animal experiments, or any data collected from social media platforms. CRediT Author Statement Gorka Calvo : Conceptualization, Software, Validation, Writing – review and editing; Anhelina Kovach : Conceptualization, Methodology, Writing – review and editing; Jorge Lanza : Supervision, Reviewing, Writing - review and editing; Leticia Montalvillo : Supervision, Validation, Funding Acquisition, Reviewing; Aitor Urbieta : Supervision, Validation, Reviewing, Funding Acquisition. Data Availability All the artifacts of the presented paper are available under request. Please contact the corresponding author of the manuscript. Acknowledgements This work has been supported by the project CRITIC (disruptive teChnologies for the pRotectIon, evaluation and secure operaTIon of Connected industrial devices, Grant Agreement No. PLEC2024-011222) funded by AEI/10.13039/501100011033, ERDF (European Regional Development Fund) and EU (European Union). Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. 5 https://www.w3.org/ns/odrl/2/ . 1 https://toolbox.dssc.eu/?tool=pet-spaces&tool-details=pet-spaces . 2 https://internationaldataspaces.org/ . 3 https://gaia-x.eu . 4 https://github.com/fiware . 6 https://github.com/Schr0ding/HE_Privacy_Module . 7 https://github.com/eclipse-edc/Connector . 8 https://github.com/eclipse-edc/MinimumViableDataspace . 9 https://github.com/microsoft/SEAL . 10 https://github.com/pistacheio/pistache . 11 http://data.open-power-system-data.org/household_data/2020-04-15/ . 12 https://github.com/Schr0ding/HE_Privacy_Module/tree/main/demo . References 1. B. Otto, S. Steinbuss, A. Teuscher, and S. Bader, “International data spaces reference architecture model (IDS RAM) v4.0,” 2022. 2. Gessler J., Cencic M.R., Metzner C., Wieker H., Lindow K., Schulz W.H. Business models and organizational roles of data spaces: a framework for value creation in data ecosystems. Data Br. 2025;61 doi: 10.1016/j.dib.2025.111795. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Data Spaces Support Centre (DSSC), “DSSC blueprint v2.0,” 2024. 4. Drazen J.M., Morrissey S., Malina D., Hamel M.B., Campion E.W. The importance — and the complexities — of data sharing. N. Engl. J. Med. 2016;375(12):1182–1183. doi: 10.1056/NEJMe1611027. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Otto B. In: Designing Data Spaces: The Ecosystem Approach to Competitive Advantage. Otto Boris W.S., ten Hompel, editors. Springer International Publishing; Cham: 2022. The evolution of data spaces; pp. 3–15. M. M. [ DOI ] [ Google Scholar ] 6. Hussein R., Scherdel L., Nicolet F., Martin-Sanchez F. Towards the European health data space (EHDS) ecosystem: a survey research on future health data scenarios. Int. J. Med. Inf. 2023;170 doi: 10.1016/j.ijmedinf.2022.104949. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Usländer Thomas A., Teuscher . In: Designing Data Spaces : The Ecosystem Approach to Competitive Advantage. M, Otto Boris and ten Hompel W.S., editors. Springer International Publishing; Cham: 2022. Industrial Data spaces; pp. 313–327. [ DOI ] [ Google Scholar ] 8. Janev V., Vidal M.E., Endris K., Pujic D. Managing knowledge in energy data spaces. Companion Proceedings of the Web Conference2021; New York, NY, USA; Association for Computing Machinery; 2021. pp. 7–15. ’21. [ DOI ] [ Google Scholar ] 9. Copei S., Rülicke L. Best practices to overcome challenges and barriers during the implementation of a data space for the energy domain: an experience report. Data Br. 2025;61 doi: 10.1016/j.dib.2025.111838. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. European Commission A European Strategy for Data. Communication from the Commission to the European Parliament, the Council, the European economic and social committee and the committee of the regions, COM(2020) 66 final. 2020. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52020DC0066 [Online]. Available: 11. Auñón J.M., et al. Evaluation and utilisation of privacy enhancing technologies—a data spaces perspective. Data Br. 2024;55 doi: 10.1016/j.dib.2024.110560. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Kovach A., Montalvillo L., Lanza J., Sotres P., Urbieta A. Understanding data spaces: a systematic mapping study of foundations, technical building blocks, and sectoral adoption. Comput. Sci. Rev. 2026;59 doi: 10.1016/j.cosrev.2025.100819. [ DOI ] [ Google Scholar ] 13. Hellmeier M., Pampus J., Qarawlus H., Howar F. Implementing data sovereignty: requirements & challenges from practice. Proceedings of the 18th International Conference on Availability, Reliability and Security; New York, NY, USA; Association for Computing Machinery; 2023. ARES ’23. [ DOI ] [ Google Scholar ] 14. European Parliament and C. of the European Union, “Regulation (EU) 2016/679 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data (General Data Protection Regulation),” 2016. 15. Fruhwirth M., Pammer-Schindler V., Thalmann S. Knowledge leaks in data-driven business models? Exploring different types of knowledge risks and protection measures. Schmalenbach J. Bus. Res. 2024;76(3):357–396. doi: 10.1007/s41471-024-00189-z. [ DOI ] [ Google Scholar ] 16. Alneyadi S., Sithirasenan E., Muthukkumarasamy V. A survey on data leakage prevention systems. J. Netw. Comput. Appl. 2016;62:137–152. doi: 10.1016/j.jnca.2016.01.008. [ DOI ] [ Google Scholar ] 17. Lubarsky B. Re-identification of ‘anonymized data,’. Georget. Law Technol. Rev. 2017;1:202. [ Google Scholar ] 18. Big Data Value Association (BDVA) Leveraging the benefits of combining data spaces and privacy enhancing technologies. Mar. 2024. https://bdva.eu/wp-content/uploads/2024/03/BDVA-CoE-DSC-Whitepaper-Leveraging-the-Benefits-of-Combining-Data-Spaces-and-PETs.pdf 19. Acar A., Aksu H., Uluagac A.S., Conti M. A survey on homomorphic encryption schemes: theory and implementation. ACM. Comput. Surv. 2019;51(no. 4):1–35. doi: 10.1145/3214303. [ DOI ] [ Google Scholar ] 20. Hariss K., Yaacoub J.P.A., Noura H.N. Homomorphic cryptography: challenges and perspectives. Comput. Sci. Rev. 2026;59 doi: 10.1016/j.cosrev.2025.100815. [ DOI ] [ Google Scholar ] 21. Gaia-X European Association for Data and C. AISBL, “Gaia-X Trust Framework - 22.10 release,” 2022. 22. S. Steinbuss and al., “Usage control in the International Data Spaces, version 3.0,” 2021. doi: 10.5281/zenodo.5675884 . 23. Rivest R.L., Adleman L., Dertouzos M.L. On data banks and privacy homomorphisms. Found. Secure Comput. 1978;4(11):169–180. [ Google Scholar ] 24. Gaia-X European Association for Data and C. AISBL, “Gaia-X architecture document - 22.10 release,” 2022. 25. International Data Space Association (IDSA) 2024. “Dataspace Protocol v.2024-1,”. [ Google Scholar ] 26. Eclipse Foundation, “Eclipse dataspace components (EDC),” 2023. 27. European Parliament and C. of the European Union, “Regulation (EU) 2022/868 on european data governance and amending regulation (EU) 2018/1724 (Data governance Act),” 2022. 28. European Parliament and C. of the European Union, “Regulation (EU) 2023/2854 on harmonised rules on fair access to and use of data and amending Regulation (EU) 2017/2394 and directive (EU) 2020/1828 (Data Act),” 2023. 29. Munoz-Arcentales A., López-Pernas S., Pozo A., Alonso Á., Salvachúa J., Huecas G. Data usage and access control in industrial Data spaces: implementation using FIWARE. Sustainability2020. 2020;12(no. 9):3885. doi: 10.3390/SU12093885. [ DOI ] [ Google Scholar ] 30. Tavani H.T., Moor J.H. Privacy protection, control of information, and privacy-enhancing technologies. SIGCAS Comput. Soc. 2001;31(1):6–11. doi: 10.1145/572277.572278. [ DOI ] [ Google Scholar ] 31. Coopamootoo K.P.L. Usage patterns of privacy-enhancing technologies. Proceedings of the 2020 ACM SIGSAC Conference on Computer and Communications Security; New York, NY, USA; Association for Computing Machinery; 2020. pp. 1371–1390. CCS ’20. [ DOI ] [ Google Scholar ] 32. Alloghani M., et al. A systematic review on the status and progress of homomorphic encryption technologies. J. Inf. Secur. Appl. 2019;48 doi: 10.1016/j.jisa.2019.102362. [ DOI ] [ Google Scholar ] 33. Peralta G., Cid-Fuentes R.G., Bilbao J., Crespo P.M. Homomorphic encryption and network coding in IoT architectures: advantages and future challenges. Electronics. 2019;8(8):827. doi: 10.3390/electronics8080827. [ DOI ] [ Google Scholar ] 34. Marcolla C., Sucasas V., Manzano M., Bassoli R., Fitzek F.H.P., Aaraj N. Survey on fully homomorphic encryption, theory, and applications. Proc. IEEE. 2022;110(10):1572–1609. doi: 10.1109/JPROC.2022.3205665. [ DOI ] [ Google Scholar ] 35. Basilakis J., Javadi B. A General framework for privacy-preserving computation on cloud environments. Proceedings of the 12th IEEE/ACM International Conference on Utility and Cloud Computing; New York, NY, USA; Association for Computing Machinery; 2019. pp. 123–132. UCC’19. [ DOI ] [ Google Scholar ] 36. Valera-Rodriguez F.-J., Manzanares-Lopez P., Cano M.-D. Empirical study of fully homomorphic encryption using Microsoft SEAL. Appl. Sci. 2024;14(no. 10):4047. doi: 10.3390/app14104047. [ DOI ] [ Google Scholar ] 37. Farahani B., Monsefi A.K. Smart and collaborative industrial IoT: a federated learning and data space approach. Digit. Commun. Netw. 2023;9(2):436–447. doi: 10.1016/j.dcan.2023.01.022. [ DOI ] [ Google Scholar ] 38. Mitrovska A., Rosas M.F., Safari P., Shariati B., Fischer J.K., Freund R. 2024 IEEE International Conference on Big Data (BigData) 2024. Federated Learning governance using Eclipse Dataspace components connectors; pp. 7946–7954. [ DOI ] [ Google Scholar ] 39. J, H M., E C., J B., Dalmolen Simon B.A., van Hillegersberg . In: Bridging Digital Sourcing, Platforms, and Ecosystems. Schreieck Maximilian K.O., Oshri, editors. Springer Nature Switzerland; Cham: 2026. Common European data spaces in health care: a privacy-preserving ecosystem for cancer research; pp. 78–98. I. and K. J. I. and K. J. [ DOI ] [ Google Scholar ] 40. Siska V., Lorünser T., Krenn S., Fabianek C. Proceedings of the 14th International Conference on Cloud Computing and Services Science - CLOSER. SciTePress; 2024. Integrating secure multiparty computation into data spaces; pp. 346–357. [ DOI ] [ Google Scholar ] 41. García N.B., Padilla M.H., Gómez A.F.S. Privacy preserving enablers for data space ecosystems. Commun. Comput. Inf. Sci. 2025;2328:73–88. doi: 10.1007/978-3-031-78572-6_5. [ DOI ] [ Google Scholar ] 42. Fotiou N., Xylomenos G., Thomas Y. EuroSec 2024 - Proceedings of the 2024 17th European Workshop on Systems Security, Part of EuroSys 2024. 2024. Data integrity protection for data spaces; pp. 44–50. [ DOI ] [ Google Scholar ] 43. Huang Q., Ouyang X., Xie H., Wu F. Proceedings - 2024 China Automation Congress, CAC2024. 2024. Design of a secure industrial data interaction system based on international data space and optimized KP-ABE algorithm; pp. 6888–6893. [ DOI ] [ Google Scholar ] 44. Dwork C. Differential privacy. Lect. Note. Comput. Sci. (incl. subser. Lect. Notes Artif. Intell. Lect. Notes Bioinform.) 2006;4052 LNCS:1–12. doi: 10.1007/11787006_1. [ DOI ] [ Google Scholar ] 45. Sebert A.G., Checri M., Stan O., Sirdey R., Gouy-Pailler C. 2023 20th Annual International Conference on Privacy, Security and Trust, PST2023. 2023. Combining homomorphic encryption and differential privacy in federated learning. [ DOI ] [ Google Scholar ] 46. Lou X., Zhang T., Jiang J., Zhang Y. A survey of microarchitectural side-channel vulnerabilities, attacks, and defenses in cryptography. ACM. Comput. Surv. 2022;54(no. 6):1–37. doi: 10.1145/3456629. [ DOI ] [ Google Scholar ] 47. Cheon J.H., Kim A., Kim M., Song Y. Homomorphic encryption for arithmetic of approximate numbers. 2016. https://eprint.iacr.org/2016/421.pdf [Online]. Available: 48. Kyriakou K., et al. 2024 IEEE International Conference on Big Data (BigData) 2024. A secure and trustworthy biometric data ecosystem for Cross-border suspect identification; pp. 2762–2771. [ DOI ] [ Google Scholar ] 49. Michikata T., Sasaki Y., Koshizuka N. 2023 IEEE International Conference on Big Data (BigData) 2023. Applying homomoprhic encryption to data spaces Takashi Michikata., pp. 4748–4753. [ DOI ] [ Google Scholar ] 50. S . The Semantic Web - ISWC2020. Springer International Publishing; Cham: 2020. The international Data spaces information model - an ontology for sovereign exchange of digital content; pp. 176–192. [ DOI ] [ Google Scholar ] 51. Dam T., Klausner L.D., Neumaier S., Priebe T. A survey of Dataspace connector implementations. arXiv. 2024 doi: 10.48550/arXiv.2309.11282. [ DOI ] [ Google Scholar ] 52. Faneela J., Ahmad B., Ghaleb S.U., Buchanan W.J. 2025 International Conference on New Trends in Computing Sciences (ICTCS) 2025. Cross-platform benchmarking of the FHE libraries: novel insights into SEAL and OpenFHE; pp. 219–226. [ DOI ] [ Google Scholar ] 53. Fawaz S.M., Belal N., ElRefaey A., Fakhr M.W. A comparative study of homomorphic encryption schemes using Microsoft SEAL. J. Phys. Conf. Ser. 2021;2128(1) doi: 10.1088/1742-6596/2128/1/012021. [ DOI ] [ Google Scholar ] 54. M.-O, C L., Aguilar Melchor Carlos R.T., Kilijian . In: Innovative Security Solutions For Information Technology and Communications. Lanet Jean-Louis C., Toma, editors. Springer International Publishing; Cham: 2019. A comparison of the homomorphic encryption libraries HElib, SEAL and FV-NFLlib; pp. 425–442. [ DOI ] [ Google Scholar ] 55. A. Kim, Y. Polyakov, and V. Zucca, “Revisiting homomorphic encryption schemes for finite fields,” 2022. [Online]. Available: https://eprint.iacr.org/2021/204.pdf 56. Kua J., Hossain M.B., Natgunanathan I., Xiang Y. Privacy preservation in smart meters: current status, challenges and future directions. Sensors. 2023;23(no. 7):3697. doi: 10.3390/s23073697. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Asghar M.R., Dán G., Miorandi D., Chlamtac I. Smart meter Data privacy: a survey. IEEE Commun. Surv. Tutor. 2017;19(4):2820–2835. doi: 10.1109/COMST.2017.2720195. [ DOI ] [ Google Scholar ] 58. Jawurek M., Kerschbaum F., Danezis G. Sok: privacy technologies for smart grids–a survey of options. Microsoft Res, Camb. UK. 2012;1:1–16. [ Google Scholar ] 59. Abdalzaher M.S., Fouda M.M., Ibrahem M.I. Data privacy preservation and security in smart metering systems. Energy. 2022;15(19):7419. doi: 10.3390/en15197419. [ DOI ] [ Google Scholar ] 60. Ahle U., et al. Design principles for Data spaces. Int. Data Spaces Assoc. 2021 doi: 10.5281/zenodo.5105744. [ DOI ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Availability Statement All the artifacts of the presented paper are available under request. Please contact the corresponding author of the manuscript. Articles from Data in Brief are provided here courtesy of Elsevier ACTIONS View on publisher site PDF (1.8 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Privacy-Preserving High-Resolution Image Gradient Computation Based on Fully Homomorphic Encryption#665793
• Privacy-Preserving High-Resolution Image Gradient Computation Based on Fully Homomorphic Encryption#664293
• An integrated privacy preserving data aggregation framework for IoT networks using homomorphic encryption and secure computation.#327220
• Privacy-Preserving IoT Data Aggregation Based on Blockchain and Homomorphic Encryption#400539
• Threshold-adaptive pruning with multi-key homomorphic encryption for communication-efficient secure federated learning.#209632