The research proposes a privacy-preserving predictive analytics paradigm that combines Federated Learning (FL), Distributed Ledger Technology (DLT), and Multi-Factor Authentication (MFA) to strengthen data privacy and security. FL supports decentralized model training, avoiding data centralization risks, and DLT protects the integrity of data through immutable and transparent blockchains. MFA strengthens access control through multi-layered authentication, reducing unauthorized access risks. The suggested model performs better than existing solutions such as Trusted Execution Environments (TEEs) and Attribute-Based Encryption (ABE) with accuracy of 93%, efficiency of 92%, and recall of 94%. The hybrid model ensures secure, scalable, and privacy-focused predictive analytics and can be applied in healthcare, finance, and IoT-based environments. This framework provides a strong solution to improve trust, transparency, and operational effectiveness in predictive modelling in sensitive domains by resolving confidentiality issues without compromising high predictive accuracy. Optimization methods and post-quantum cryptography will be investigated in future work for further security.
Human genetic data, crucial for advancing personalized medicine, requires secure and privacy-preserving management solutions. Traditional approaches face challenges in scalability, security, and decentralized access control. This study proposes a blockchain-based framework leveraging Thirdweb and Ethereum smart contracts to address these issues. The framework integrates decentralized storage via IPFS for cost-efficient off-chain genetic data storage, while on-chain smart contracts manage access control, encryption, and audit trails. Utilizing Solidity for smart contract development, the system ensures role-based permissions, wallet-based authentication, and immutable transaction logging. Genetic data in FASTA format, sourced from NCBI, is encrypted and linked to IPFS hashes stored on the blockchain. The architecture supports dual interfaces—command-line for developers and a Thirdweb dashboard for end-users—enabling secure data upload, access, and monitoring. Testing demonstrated functional efficacy in data integrity, access verification, and audit capabilities. Results highlight the system’s ability to enhance privacy, eliminate intermediaries, and provide transparent data governance. The integration of Thirdweb further decentralizes operations, aligning with Web 3.0 principles. Key contributions include a scalable model for genetic data sharing, a customizable smart contract template, and a user-centric design. Future work should explore advanced encryption, real-world healthcare integration, and performance optimization under high-throughput conditions. This research bridges biotechnology and blockchain, offering a robust foundation for secure genomic data ecosystems.
Rodrigo Gonçalves Bueno, André Luiz De Souza Carneiro, João Paulo Aragão Pereira
The increasing adoption of tokenized assets, Decentralized Finance (DeFi) applications, and the exploration of Central Bank Digital Currencies (CBDCs) necessitate sophisticated security architectures for Regulated Tokenized Multi-asset Networks (RTMNs). This paper addresses the complex interplay between privacy, and composability within these emerging decentralized financial ecosystems. It is argued that conventional security paradigms, predominantly reliant on perimeter defenses, are insufficient for the distributed and interconnected nature of DeFi infrastructures. While Zero Trust models offer relevant principles, their direct application within regulated, high-performance financial networks, particularly those involving CBDCs or complex DeFi protocols, presents significant challenges regarding compliance and efficiency. This paper introduces a novel framework meticulously designed to support diverse RTMN use cases, including retail/wholesale CBDC, tokenized deposit, stablecoin and multi-asset platforms operating within a DeFi context. A foundational element of this framework is the implementation of cryptographically enforced information compartmentalization. This ensures that each architectural component operates with the minimum necessary information required for its specific function, inherently embedding privacy-by-design and preventing unauthorized access to comprehensive network or transactional data. The proposed framework is architected to guarantee critical properties essential for robust distributed and decentralized systems: (1) Atomicity of transactions; (2) Composability and Programmability; (3) Settlement Finality, providing transaction immutability; (4) Enhanced Privacy and Security, leveraging cryptographic techniques; (5) Support for Distribution and Decentralization; (6) Performance, addressing throughput and latency demands; and (7) Continuous monitoring and auditing, enabling regulatory oversight without compromising user data. It is provided a detailed analysis of the framework's application across distinct RTMN implementations, identifying specific technical challenges and opportunities within the context of tokenized systems. Furthermore, the paper presents a qualitative and functional evaluation of the framework's characteristics applied to the use case of tokenizing Federal Government Securities in an RTMN, such as Drex. Fundamentally, the inherent trade-offs between cryptographic privacy guarantees, compartmentalization, and programmability are examined, exploring optimization strategies relevant to demanding DeFi and institutional applications, such as decentralized trade finance and tokenized debt instruments.
Federated Learning (FL) offers an attractive framework for collaboratively training AI models while preserving data privacy. However, it also introduces challenges in verifying the integrity and authenticity of model updates across diverse clients. Zero-Knowledge Proofs (ZKP) provide a promising means to address these issues by verifying computations without revealing underlying data. Yet, global verification using ZKP remains computationally expensive and does not scale well. To overcome these limitations, we propose a novel approach grounded in two key principles: (a) partial verification, targeting carefully selected subsets of data, can effectively mitigate adversarial attacks; and (b) robust data verification is essential, ensuring not only the consistency of model parameters but also the authenticity of the underlying data. We highlight the potential operation of this partial verification system, discuss novel research directions, and outline strategies for a wider integration into FL architectures.
As governmental health information systems become increasingly digitized, the need for collaborative analytics across decentralized regions has intensified.However, privacy concerns, regulatory constraints, and infrastructure disparities have limited the extent to which sensitive health data can be aggregated and analyzed across jurisdictions.This paper explores the integration of blockchain technology with federated learning (FL) to enable privacy-preserving data analytics across distributed governmental health information systems.By combining FL's decentralized model training capabilities with blockchain's immutable, transparent ledger and consensus mechanisms, the proposed framework ensures secure, auditable, and policy-compliant data collaboration without requiring raw data exchange.The framework leverages smart contracts to automate access control, consensus validation, and compliance enforcement among participating health institutions.Each node (representing a governmental health entity) trains models locally and shares only encrypted model parameters, which are validated and recorded on the blockchain.This eliminates the need for centralized authorities and reduces the risk of data leakage or manipulation.A core contribution of this work lies in addressing public-sector constraints such as legacy infrastructure, heterogeneous data standards, and institutional trust gaps through a modular, interoperable design.The system includes support for dynamic node participation, real-time updates, and compatibility with health data standards such as HL7 and FHIR.Use-case simulations across municipal, regional, and national health departments demonstrate improved efficiency in outbreak prediction, chronic disease surveillance, and population-level risk stratification while maintaining strict compliance with data protection regulations.This paper advances a scalable and trustworthy architecture for cross-border health collaboration, offering a blueprint for digital public health infrastructures in the age of data sovereignty and distributed intelligence.
Federated Learning (FL) can accelerate the speed of distributed computing in a meta-universe in a wireless environment. We investigate the integration of machine learning models with irreplaceable tokens (NFT) to enable wireless metaverse users (MUs) to control ownership and participate in the economic value allocation by applying FL (FL-NFT). The MUs are grouped into a decentralized-autonomous organization (DAO) to train the global model. To find a cost-benefit tradeoff, MUs and metaverse service providers (MSPs) need to use Stackelberg games to find better strategies and derive the optimal solution by backward induction. We have designed a novel blockchain-based secure auction mechanism (SAM). Theoretical analysis and simulation results show that SAM can enhance FL-NFT and realize the inherent characteristics of incentive mechanisms.
The current centralized model of Public Key Infrastructure (PKI) relies heavily on trusted Certificate Authorities (CAs) to authenticate digital identities. Still, this system faces significant security risks, including fraudulent certificate issuance and CA compromises. This thesis explores the potential of blockchain technology as a decentralized solution to these issues, proposing a distributed PKI framework that removes the single point of failure inherent in traditional systems. By leveraging blockchain’s immutability, consensus protocols, and transparency, this approach aims to enhance digital identity security and offer a more resilient infrastructure for managing certificates. A novel design is presented, incorporating a twophase Proof of Stake consensus mechanism, an account tree for domain owners’ public keys, and advanced certificate extensions for verifying identities within the blockchain network. The proposed framework not only improves the security of certificate issuance but also ensures tamper-proof logging and decentralized control, reducing the risks associated with CA vulnerabilities. This research lays the foundation for a more robust, scalable, and censorship-resistant PKI, providing an innovative solution to the challenges facing modern digital communications.
The rapid evolution of phishing attacks targeting email, chat, and social media platforms poses a significant threat to digital security, with a reported 667% surge in spear-phishing during the 2020 COVID-19 crisis [1]. Current AI-based detection systems face challenges in dataset diversity, adversarial robustness, computational scalability, model interpretability, and privacy preservation, limiting their efficacy in real-time, multi-platform environments. This paper introduces PhishGuard, an innovative framework for real-time phishing detection, designed to overcome these limitations. PhishGuard integrates lightweight transformer models (e.g., distilled BERT), hybrid detection techniques combining natural language processing (NLP), propagation analysis, and user behavior analysis, and explainable AI (XAI) methods like SHAP and LIME for transparent decision-making. Privacy-preserving techniques, including federated learning and local differential privacy, ensure secure processing of sensitive user data. Evaluated on diverse datasets such as PhiKitA, Enron, and a custom social media corpus, PhishGuard achieves up to 97.5% accuracy, 94% F1-score, and inference times below 5 ms, demonstrating scalability for resource-constrained devices. The framework also incorporates zero-knowledge proofs for verifiable inference, addressing trust and integrity concerns. By tackling cross-domain generalization, adversarial robustness, and real-time performance, PhishGuard offers a scalable, user centric solution for secure digital communications, with applications in finance, healthcare, and social media platforms. Future enhancements include multilingual support and image based phishing detection, paving the way for a comprehensive defense against evolving cyber threats.
Hui Dou, Xuewei Wang, Mian Ahmad Jan, Haiwei Sang · 5 authors
Within the realm of 6G Internet of Vehicles (6G-IoV), Federated Learning (FL) has become a notable machine learning framework, providing a decentralized method to protect data privacy while allowing cooperative model training. Specifically, with 6G technology, FL will benefit from ultra-low latency, high reliability and massive connectivity, enabling real-time model updates and efficient data sharing in the 6G-IoV ecosystem. However, FL faces challenges like the single points of failure and potential privacy leakage from data providers. To tackle the aforementioned challenges, we propose a blockchain-based trustworthy verifiable FL scheme for 6G-IoV, that is, AVBFL. Firstly, we introduce blockchain technology to address the issue of decentralization by storing transactions on-chain. Furthermore, to protect the privacy of local gradients, we utilize the Burmester-Desmedt (BD) multi-party key agreement protocol to negotiate a shared key and encrypt the gradients with the AES encryption algorithm. We also sign transactions using the ECDSA signature algorithm. Additionally, we design a time-sensitive Proof of Stake (TPoS) consensus mechanism based on Newton’s cooling law to boost participants’ enthusiasm for training and select the miner with the highest stake to mine the block. Finally, experiments have demonstrated the effectiveness of AVBFL. In the presence of malicious nodes, the average accuracy rate is increased by 71.8% compared to the VFL scheme and by 8.6% compared to the VBFL scheme.
In the domain of authentication, information leakage which can lead to identity theft represents a significant challenge in the field of cybersecurity. This challenge is particularly relevant in the context of 5 G tactical bubbles, where secure and efficient authentication mechanisms are critical to gain access to sensitive information and communication services. The concept of Zero-Knowledge Proofs, in particular non-interactive proofs, has gained attention in recent years as robust cryptographic methods for privacy-preserving protocols. Zero-knowledge proofs enable users to prove possession of specific knowledge to verifiers without revealing the knowledge itself in a single interaction round. Despite its growing popularity, Zero-Knowledge Proofs have not yet been fully explored within 5 G tactical bubbles. In this paper, we perform a comparative analysis between traditional authentication mechanisms and Zero-Knowledge Proofs-enabled authentications. To this end, we evaluate the feasibility in terms of time and computational complexity and determine whether these advanced authentication protocols can ensure enhanced privacy and security in 5 G tactical bubbles.
Smart city infrastructures are resulting in a concern regarding proper security and fine- grained access control for the access of the sensitive data generated by the interconnected devices and services. Having rich yet heterogeneous environments is a common scenario for organizations nowadays, and conventional access control mechanisms often do not suffice to provide support for flexible, scalable and decentralized authorization. Here, we develop a new access control method that leverages Cipher text-Policy Attribute-Based Encryption (CP-ABE) and Ethereum smart contracts to realize decentralized and fine-grained data access in smart city applications. Users are allowed to access and decrypt information only if they own a suitable set of attribute. They leverage smart contracts on the Ethereum blockchain to perform attribute verification, issue keys, and log access actions, which removes the need to trust a centralized authority. We also describe the architecture of the system, the design of smart contracts, and an implementation prototype. The experimental results can validate the computational efficiency, the gas consumption and the policy expressiveness of our approach, showing that it is a promising solution for secure and transparent data access control model in smart cities.
Jiahui Huang, Teng Huang, Changyu Dong, Sisi Duan · 5 authors
The expansion of blockchain applications, particularly on platforms like Ethereum, brings escalating security challenges as account anonymity provides breeding grounds for criminals to commit crimes and cause significant economic losses. As the mainstream architecture of de-anonymization technology, graph neural networks (GNNs) provide empirical tools for law enforcement agencies to investigate illegal activities. However, the limited expressiveness of current GNNs leads to performance degradation for Ethereum account de-anonymization. To address this challenge, we propose an innovative Local-Global Awareness (LGA) framework, which consists of a Local Structure-Aware (LSA) module and a Global Information-Aware (GIA) module. LSA integrates subgraph-level encoding strategies with local attention to enhance the capture of microscopic interactions. As a complementary measure, GIA introduces global attention to facilitate the understanding of macroscopic information. The LGA framework meticulously captures subgraph-level account behavior patterns at a granular level while simultaneously incorporating global contextual insights, demonstrating higher-level expressive power and receptive fields over conventional GNN. The efficacy of the LGA framework is corroborated by experimental evaluations conducted on the lw-AIG dataset. Our framework achieves exceptional performance, significantly outstripping state-of-the-art GNN-based methods in terms of the micro F1 score metric, with relative improvements ranging from 0.14% to 6.63%. Through its detailed and comprehensive analysis of account interactions, the LGA framework aims to provide a potent solution to the complex security challenges faced in the expanding blockchain landscape. The code for LGA is available at https://github.com/deepang-ai/LGA.
Web3 applications, such as on-chain games, NFT minting, and leader elections necessitate access to unbiased, unpredictable, and publicly verifiable randomness. Despite its broad use cases and huge demand, there is a notable absence of comprehensive treatments of on-chain verifiable randomness services. To bridge this, we offer an extensive formal analysis of on-chain verifiable randomness services. We present the first formalization of on-chain verifiable randomness in the blockchain setting by introducing the notion of Verifiable Randomness as a Service (VRaaS). We formally define VRaaS using an ideal functionality$\mathcal{F}\text{VRaaS}$in the Universal Composability model. Our definition not only captures the core features of randomness services, such as unbiasability, unpredictability, and public verifiability, but also accounts for many other crucial nuances pertaining to different entities involved, such as smart contracts. Within our framework we study a generic design of Verifiable Random Function (VRF)-based randomness service - where the randomness requester provides an input on which the randomness is evaluated as VRF output. We show that it does satisfy our formal VRaaS definition. Furthermore, we show that the generic protocol captures many real-world randomness services like Chainlink VRF and Supra dVRF. Moreover, we investigate the minimalism of the frame-work. Towards that, first we show that, the two transactions in-built in our framework are actually necessary for any randomness service to support the essential qualities. We also discover practical vulnerabilities in other designs such as Algorand beacon, Pyth VRF and Band VRF, captured within our framework.
Lanlan Sun, Yinzhen Wei, Zongshan Wang, Hengjun Liu
Privacy protection, establishment of trust, and quality assurance are known as very vital issues faced by the data sharing systems nowadays. This paper introduces a new framework that addresses these core limitations to integrate distributed ledger technology and machine learning practices. The suggested system will have a consortium blockchain design with incorporated neural network modules to provide automatic data validation and anomaly-detecting features. Smart contractbased governance leads to the safety of sensitive information due to the protection laid by multi-layered encryption protocols and transparency in operations. Evaluation of performance makes use of three different domains namely: their medical information systems, financial transaction networks and sensor data networks. The security is enhanced by $23.5 \%$, quality assessment is more accurate by $\mathbf{9 4. 2 \%}$, and the sustained processing capacity values 2,847 transactions per second according to the comparative assessment. The model provides the basis of cross-organizational cooperation with data as well as regulation and operation efficiency needs in distributed computing environments.
Rakesh Reddy Charla, Kalpan Dharamshi, Raj Sonani, Aishwarya Rajkumar Shah
Exponential increase in data in distributed retail and financial settings poses severe challenges to privacy of user’s preservation without compromising the effectiveness of AI-driven insights. Such sensitive fields are usually not easy to keep up with using the conventional centralized data engineering approaches that fail to comply with the regulatory, scalability, and latency constraints. This paper presents a brand-new federated data engineering framework with a focus on privacy-aware AI applications, based on the patterns seen in the real-world retail and finance workflows. The proposed framework benefits from federated learning paradigms, decentralized feature engineering, and privacy-preserving transformation techniques to offer secure, cross-node training of the modelling powered with overcoming data ownership and sector-specific regulation restrictions. The study describes the details of the key system components such as federated schema harmonization, edge-level preprocessing, and secure aggregation mechanisms. Performance evaluations performed on synthetic retail transaction data and federated credit risk datasets show that the framework strikes a balanced performance between privacy guarantees, data utility and computational efficiency. According to the results, the approach to federated data engineering provides an expandable and legally friendly way for implementing AI solutions for data-sensitive fields, which allows creating robust, data-protective intelligence systems for various industries.
Beyhan Adanur Dedetürk, Ahmet Soran, Burcu Bakır-Güngör
Every day, hundreds of gigabytes of data are produced due to the exponential growth of next-generation sequencing and omics technologies. By combining omics data with other data types, such as electronic health record data, panomics research is actively attempting to uncover novel and potentially useful biomarkers. For the effective analysis of high-throughput-derived omics data, it is imperative to establish robust and reliable platforms that prioritize ethical considerations while effectively managing privacy, ownership concerns, and the responsible sharing of data. The GenShare model was proposed to provide an efficient platform that fits these needs. GenShare is a hybrid platform that utilizes blockchain technology. Paillier’s homomorphic encryption scheme in tandem with Intel Software Guard Extension (SGX) serves to enable the sharing of genomic data, execution of count queries, and statistical analysis of genomic data while preserving privacy and avoiding compromise of sensitive information. The objective of this paradigm is to confront security and privacy concerns through the integration of homomorphic encryption and SGX, addressing additional challenges associated with Hyperledger Fabric and Ethereum. In pursuit of this objective, the implementation of the system involved establishing the Hyperledger Fabric network, with various workloads employed to assess the network’s efficiency. Consequently, it was hypothesized that the new GenShare model would enhance the data collection and dissemination cycle and serve as a proficient platform catering to the needs of its users.
S. Aruna Deepthi, Deva Vasista, Ankitha Vundyala, Dileep Kumar Murala · 5 authors
Blockchain, with its decentralized ledger, holds promise for transforming industries by enhancing transparency, efficiency, and security. To achieve widespread adoption, it must address the core principles of accountability, data integrity, and privacy. Accountability in blockchain involves identifying responsibility for actions in a decentralized environment, crucial for scenarios like supply chain management or financial transactions. Data integrity safeguards the accuracy of information on the blockchain, preventing fraud and fostering trust, especially in sensitive domains like finance and healthcare. Blockchain’s immutable nature ensures that recorded data remains unaltered. Privacy is a fundamental human right that blockchain must uphold. While transparency promotes accountability, concerns about personal data protection necessitate a balance. Striking this balance is essential for user acceptance and ensuring individuals control their data. These principles of accountability, data integrity, and privacy are indispensable for blockchain’s successful adoption. By adhering to them, blockchain can foster trust, transparency, and security, paving the way for a more secure digital future. Its potential to revolutionize industries and enhance societal well-being hinges on addressing these principles and establishing a trusted foundation for a decentralized world.
The insurance sector has been moving towards a more digital footing on the backdrop of growing demand for transparency, efficiency and privacy of data. The common way it is built exposes it to subtle policy checks and human intervention for claim reviews and exposes consumer data. Although blockchain technology has been explored as a possible solution, the public log nature of blockchain introduces grave concerns in privacy preservation, especially, in industries with strict regulatory requirements, including healthcare and life insurance. We propose a privacy-preserving insurance system in our paper based on ZKPs and SMPC to handle these issues. Zero-Knowledge Proofs (ZKPs) enable policyholders to prove to third parties’ compliance to their policy conditions without revealing underlying information, while leveraging Secure Multi-Party Computation (SMPC) provides insurance companies, business partners, and third-party auditors to jointly compute premium prices, validate claims, and compute refunds on privatized data inputs. End to End the Privacy of Data is secured, even at point of payment, Claim or Refund settlement. We present experimental results demonstrating real-time (300-950ms) latency, perfect replication on all nodes, and no leakage across transaction types. This work addresses inefficiencies that have plagued the insurance industry for years by presenting an extensible, cryptographically secure technology that can be used in the next era of digital insurance. It proposes a verifiable secret sharing based transaction model to bring in a new trustful and privacy-preserving insurance transaction paradigm.
The European Union's (EU) data strategy aims to create a single market for seamless data flow while ensuring proper governance, privacy, and data protection. In this paper, we present SEDIMARK, an EU project, that builds on this strategy by developing a fully decentralised, secure data marketplace. The goal of SEDIMARK is to build a complete toolbox that enables users to purchase and process data assets. The toolbox includes tools for data cleaning, decentralised machine learning models and secure data exchange. SEDIMARK offers users full control over data assets by enabling them to keep their data locally and thus removing the need for central servers. With customisable pipelines and tools, SEDIMARK supports a wide range of users, from novices to experts, promoting seamless collaboration and fair access to high-quality datasets across Europe. The decentralised connectivity in SEDIMARK is achieved with the use of Distributed Ledger Technology (DLT). Furthermore, SEDIMARK's architecture features a unique Connector component using Self Sovereign Identities (SSI), fostering trust and secure interactions. Transactions in SEDIMARK are stored in a Registry, a decentralised, immutable, non-repudiable and permissionless database. Together the technologies used in SEDIMARK ensure privacy, trust and data quality for secure management, sharing, and monetisation of assets in data spaces.
Bilel Zaghdoudi, Gewu Bu, Maria Potop-Butucaru, Serge Fdida
This paper presents a novel blockchain-based decentralized identity (DI) system, tailored for enhanced digital identity management in Internet of Things (IoT) and device-todevice (D2D) networks. The proposed system features a hierarchical structure that effectively merges a distributed ledger with a mobile D2D network, ensuring robust security while streamlining communication. Central to this design are the gateway nodes, which serve as intermediaries, facilitating decentralized identifier (DID) registration and device authentication through smart contracts and distributed storage systems. A thorough security analysis underscores the system's resilience to common cyber threats and adherence to critical principles like finality and liveness.
We present a secure and efficient federated learning protocol for autonomous vehicles that resists data leaks, redundancy, and adversarial attacks. Our system combines fast zero-knowledge proofs and compressed Bloom filters to verify updates without exposing private data. Compared to traditional approaches, our method reduces proof sizes by 90 % (under 10 KB), memory by up to 75 %, and maintains accuracy with less than 4% degradation under 30% attack rates. The entire update cycle completes in under 600 ms, making it practical for real-time use in vehicles. This work advances trustworthy AI deployment in dynamic, resource-limited networks.
Federated Learning (FL) is an innovative learning paradigm that allows multiple devices to collaboratively train a shared model without uploading the raw data to the cloud, thereby enhancing privacy and security. Leveraging Mobile Edge Computing (MEC), Hierarchical Federated Learning (HFL) can further reduce the communication overhead, thereby increasing the efficiency and scalability of FL systems by enabling model aggregation at the network edge. However, this framework often encounters security challenges, such as single points of failure and the risk of malicious model tampering. To address these challenges, researches have employed blockchain technology to enhance the security of FL systems, but most of these solutions incur significant resource burdens due to the intensive computation demands of blockchain consensus mechanisms, such as Proof-of-Work (PoW). In this work, we aim to explore a lightweight blockchain-enabled federated learning (LBFL) framework that utilizes the Delegated Proof-of-Stake (DPoS) consensus mechanism, which employs a simple voting process to elect a small number of candidate block producers (known as delegates) to aggregate the FL model and produce blocks. This framework significantly reduces the number of consensus nodes, thereby minimizing resource consumption during the consensus process. We study the joint optimization of mobile device association, bandwidth allocation, computing frequency management, and block producer selection, aiming to minimize the overall delay and energy consumption. To address the challenges posed by discrete and continuous decision variables, we decompose the problem into three sequential subproblems and solve them iteratively. Simulation results show that compared with the existing benchmarks, the proposed scheme can reduce overall delay and energy consumption by 15% to 22%.