<p>As Decentralized Finance (DeFi) and Non-Fungible Tokens (NFTs) expand, self-custody wallets have become the primary interface for user sovereignty. However, existing solutions suffer from critical limitations, including static authentication frameworks that compromise usability, a lack of real-time risk awareness, and inadequate key recovery mechanisms that often lead to permanent asset loss or reliance on centralized custodians. Furthermore, current wallets frequently expose transaction metadata, undermining user privacy. To address these systemic flaws, we present a modular self-custody wallet that incorporates a context-aware risk engine for real-time transaction scoring, risk-based adaptive authentication, and a dual-path decentralized key-recovery layer combining DAO-governed Shamir secret sharing with a zk-SNARK-verified fallback. The architecture further includes programmable policy enforcement and a zero-knowledge swap layer with stealth addressing to decouple front-end activity from on-chain data. The design integrates smart contracts on EVM chains and Solana through provider adapters and executes on-device ML inference to minimize latency. Experimental results demonstrate that the proposed system reduces privacy leakage probability to 5% (compared to 85% in standard architectures) and accelerates key recovery from over 24 h to approximately 8 seconds using zk-SNARKs, all while achieving 93.6% risk classification accuracy. The proposed CAPPR-Wallet advances self-custody by combining context adaptivity, privacy, and recoverability without centralized trust.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

CAPPR-Wallet: a context-aware and recoverable wallet architecture with privacy-preserving rules for trustless blockchain ecosystems

  • Mingjun Liu,
  • Huiying Li,
  • Ali Muqtadir,
  • Rubab Osama,
  • Muhammad Farrukh Shahzad

摘要

As Decentralized Finance (DeFi) and Non-Fungible Tokens (NFTs) expand, self-custody wallets have become the primary interface for user sovereignty. However, existing solutions suffer from critical limitations, including static authentication frameworks that compromise usability, a lack of real-time risk awareness, and inadequate key recovery mechanisms that often lead to permanent asset loss or reliance on centralized custodians. Furthermore, current wallets frequently expose transaction metadata, undermining user privacy. To address these systemic flaws, we present a modular self-custody wallet that incorporates a context-aware risk engine for real-time transaction scoring, risk-based adaptive authentication, and a dual-path decentralized key-recovery layer combining DAO-governed Shamir secret sharing with a zk-SNARK-verified fallback. The architecture further includes programmable policy enforcement and a zero-knowledge swap layer with stealth addressing to decouple front-end activity from on-chain data. The design integrates smart contracts on EVM chains and Solana through provider adapters and executes on-device ML inference to minimize latency. Experimental results demonstrate that the proposed system reduces privacy leakage probability to 5% (compared to 85% in standard architectures) and accelerates key recovery from over 24 h to approximately 8 seconds using zk-SNARKs, all while achieving 93.6% risk classification accuracy. The proposed CAPPR-Wallet advances self-custody by combining context adaptivity, privacy, and recoverability without centralized trust.