Protected-Secret Convergent Encryption with Ciphertext-Token Verification for Secure Cloud Deduplication

Objectives: To develop and evaluate an Advanced Convergent Encryption (ACE) method that preserves exact encrypted deduplication while preventing an external observer from deriving the encryption key from a guessed file alone. The method integrates protected-secret key derivation, deterministic authenticated encryption, ciphertext-token verification, and ownership-aware references. Method: For plaintext file or block P , ACE computes the SHA-256 content digest h . The protected service derives the 256-bit key K by applying HMAC-SHA-256 to a domain label and h under server secret S . AES-SIV produces deterministic authenticated ciphertext, and SHA-256 of the ciphertext forms the repository token. A reproducible single-host experiment used deterministic 1, 5, 10, and 25 MiB files, 30 measured runs per method and size after three warm-up runs, and content-only MLE and 2048-bit RSA blind-OPRF controls. Mean, sample standard deviation, throughput, storage saving, and two-sided Welch tests were evaluated. Findings: Across the four size-level means, ACE required 24.859 ms for end-to-end in-memory upload, compared with 24.004 ms for content-only MLE and 47.485 ms for RSA-OPRF. The 3.56% overhead relative to MLE was not significant at any size ( p \geq 0.071 ), whereas ACE reduced upload time by 47.65% relative to RSA-OPRF ( p < 0.001 at every size). ACE key derivation averaged 6.730 ms, 77.19% below RSA-OPRF. Authorized retrieval averaged 17.559 ms for ACE and 40.726 ms for RSA-OPRF. ACE correctly matched all 120 exact-duplicate conditions and separated all 120 one-byte-modified conditions. In the controlled corpus, eight unique ciphertext objects replaced twelve undeduplicated objects, giving 33.33% ciphertext-storage saving before metadata. Novelty: ACE combines a secret-dependent deterministic key, misuse-resistant authenticated encryption, ciphertext-derived duplicate tokens, and authorization-gated ownership pointers in one symmetric-key path. Unlike content-only convergent encryption, a guessed plaintext is insufficient for offline key reconstruction while the protected secret remains unavailable. Unlike the evaluated RSA-OPRF control, ACE avoids public-key operations. The method does not claim oblivious key derivation and explicitly retains equality leakage inherent to exact deduplication. Keywords: Convergent Encryption, Encrypted Deduplication, Deterministic Authenticated Encryption, Ciphertext Token, Cloud Storage Security, Message-Locked Encryption

Authors

Institutions

Publication Details

Journal
Indian Journal of Science and Technology
Published
2026-11-16
DOI
https://doi.org/10.17485/ijst/v19i40.1358
Primary Topic
Cryptography and Data Security
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Protected-Secret Convergent Encryption with Ciphertext-Token Verification for Secure Cloud Deduplication

S S Manikandasaran, R Surendhar Kumar
Indian Journal of Science and Technology
Cryptography and Data Security
article

Protected-Secret Convergent Encryption with Ciphertext-Token Verification for Secure Cloud Deduplication

S S Manikandasaran, R Surendhar Kumar
article en

Abstract

Objectives: To develop and evaluate an Advanced Convergent Encryption (ACE) method that preserves exact encrypted deduplication while preventing an external observer from deriving the encryption key from a guessed file alone. The method integrates protected-secret key derivation, deterministic authenticated encryption, ciphertext-token verification, and ownership-aware references. Method: For plaintext file or block P , ACE computes the SHA-256 content digest h . The protected service derives the 256-bit key K by applying HMAC-SHA-256 to a domain label and h under server secret S . AES-SIV produces deterministic authenticated ciphertext, and SHA-256 of the ciphertext forms the repository token. A reproducible single-host experiment used deterministic 1, 5, 10, and 25 MiB files, 30 measured runs per method and size after three warm-up runs, and content-only MLE and 2048-bit RSA blind-OPRF controls. Mean, sample standard deviation, throughput, storage saving, and two-sided Welch tests were evaluated. Findings: Across the four size-level means, ACE required 24.859 ms for end-to-end in-memory upload, compared with 24.004 ms for content-only MLE and 47.485 ms for RSA-OPRF. The 3.56% overhead relative to MLE was not significant at any size ( p \geq 0.071 ), whereas ACE reduced upload time by 47.65% relative to RSA-OPRF ( p < 0.001 at every size). ACE key derivation averaged 6.730 ms, 77.19% below RSA-OPRF. Authorized retrieval averaged 17.559 ms for ACE and 40.726 ms for RSA-OPRF. ACE correctly matched all 120 exact-duplicate conditions and separated all 120 one-byte-modified conditions. In the controlled corpus, eight unique ciphertext objects replaced twelve undeduplicated objects, giving 33.33% ciphertext-storage saving before metadata. Novelty: ACE combines a secret-dependent deterministic key, misuse-resistant authenticated encryption, ciphertext-derived duplicate tokens, and authorization-gated ownership pointers in one symmetric-key path. Unlike content-only convergent encryption, a guessed plaintext is insufficient for offline key reconstruction while the protected secret remains unavailable. Unlike the evaluated RSA-OPRF control, ACE avoids public-key operations. The method does not claim oblivious key derivation and explicitly retains equality leakage inherent to exact deduplication. Keywords: Convergent Encryption, Encrypted Deduplication, Deterministic Authenticated Encryption, Ciphertext Token, Cloud Storage Security, Message-Locked Encryption

Indian Journal of Science and TechnologyVol. 19(40)
Bharathidasan University (IN)
Cryptography and Data Security
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon