CrowVault: A Proof-of-Concept Architecture for Authenticated File Encryption, Unequal Fragmentation, and Multi-Nest Storage

CrowVault is a proof-of-concept architecture for protecting digital files through authenticated encryption, unequal fragmentation, multi-location storage, encrypted recovery metadata, and fragment-level integrity verification. The system encrypts each source file using AES-256-GCM before dividing the resulting encrypted container into unequally sized fragments. These fragments are assigned randomized identifiers and distributed across multiple independent storage locations, referred to as digital nests. A separately encrypted Crow Map records the information required to locate, verify, reorder, reconstruct, and decrypt the protected file. The proof-of-concept implementation was developed in C# using .NET 8. It uses PBKDF2-HMAC-SHA256 for password-based key derivation and SHA-256 for fragment integrity verification. The prototype supports file protection, fragment distribution, protected-file restoration, wrong-password rejection, detection of modified or missing fragments, and verification of the restored file against the original SHA-256 digest. CrowVault does not propose a new cryptographic primitive. Instead, it investigates a storage-security architecture that combines established cryptographic techniques with distributed encrypted fragmentation and integrity-controlled recovery. The present version is intended for research, experimentation, architectural evaluation, and future development. The report discusses the system architecture, threat model, container and manifest formats, protection and recovery workflows, prototype implementation, verification procedures, security properties, limitations, and future research directions. This software architecture and prototype have not undergone independent security certification or a formal cryptographic audit. They should not yet be used as the sole protection mechanism for critical, classified, medical, financial, or legally sensitive information. Author:Ahmed Abdulla Ashlam Affiliation:Reverse Innovation Research and Development LabReverse Innovation for IT & AI (RIIA)Tripoli, Libya Email:[email protected] Website:https://riia.ly DOI:https://doi.org/10.5281/zenodo.21633642