Geng Academic Fraud Report: "Verifiable Attribute-Based Encryption Scheme with Outsourced Decryption in Fog Computing" (DOI: 10.3969/j.issn.1007-130X.2023.03.009)

Academic fraud report · Geng Detector

Summary

Verdict: Highly Suspicious. This report examines the paper "Verifiable Attribute-Based Encryption Scheme with Outsourced Decryption in Fog Computing" by Duan Yahong, Wang Zheng, Zhao Juanjuan, and Wang Long, published in Computer Engineering & Science, Vol. 45, Issue 3, March 2023 (DOI: 10.3969/j.issn.1007-130X.2023.03.009). The principal finding is a severe internal contradiction between Table 2 and Figure 4: Table 2 lists user-decryption cost as a constant (tE+tM or 2tM+tP) independent of attribute count N, yet Figure 4 shows these same schemes' decryption time rising linearly as attributes scale from 10 to 50. Either the theoretical complexity claims or the experimental measurements are fabricated. Figure 3 additionally shows suspicious curve shapes for the baseline schemes that do not match their published complexity formulas, suggesting plotted rather than measured data. Confidence is moderate-to-high; the inconsistency is mathematically clear, but we cannot determine which side (theory or experiment) is wrong without source code.

Verdict

🟠 Highly Suspicious. Two independent inconsistencies between theoretical complexity statements and experimental figures cast serious doubt on the validity of the reported results. A separate concern about ciphertext structure suggests questionable security proof rigor.

Key findings

Theory–experiment contradiction (Table 2 vs. Figure 4): Table 2 reports user-decryption costs of tE + tM (proposed scheme) and 2tM + tP (Ref. [11]), implying O(1) complexity w.r.t. attribute count N. Figure 4, however, plots both schemes' decryption time with a clear linear upward trend across 10–50 attributes, which is mathematically incompatible with constant-time decryption.
Plausibly fabricated baseline curves (Figure 3 vs. Table 2): Ref. [6]'s encryption cost is listed as (3+N_U−N_P)tE + (2+N_P)tM. With attribute count on the x-axis and N_P held fixed, the dominant growth term in tE would be very steep relative to tM, but the figure's slope for Ref. [6] is reported as anomalously shallow compared to what the formula predicts—suggesting curves may have been hand-drawn rather than measured.
Questionable ciphertext parameter exposure (Section 3.3, Algorithm 5): The scheme publicly embeds the parameter C'_i = ω_i in the ciphertext to enable on-chain verification of Σ c_x C'_x = 0. Releasing such linear-combination material in clear text deviates from standard CP-ABE design and may open the scheme to collusion or secret-leakage attacks; the rigor of the accompanying security proof is therefore doubtful.
OCR / formula corruption (Section 3.3, formula (6)): The KeyGen algorithm renders as K = g^{α+kαb kah bc}, with scrambled tokens such as α+kαb k·s and k·a appearing in formula (6). This is likely a PDF–OCR artifact but obscures whether the underlying derivation is correct.

Evidence highlights

Direct numeric evidence from Table 2: tE + tM and 2tM + tP for user-decryption costs (proposed / Ref. [11]).
Direct numeric evidence from Figure 4: linear growth in user-decryption time over attributes 10–50 for both schemes.
Direct numeric evidence from Table 2: Ref. [6] encryption cost (3+N_U−N_P)tE + (2+N_P)tM.
DOI preserved verbatim: 10.3969/j.issn.1007-130X.2023.03.009.

Notes

The central Table 2 / Figure 4 conflict is a mathematical inconsistency: a constant-complexity operation cannot, even with measurement noise, exhibit the sustained linear growth shown across a 5× range of attribute counts. Without access to the authors' implementation it is not possible to adjudicate whether the theoretical analysis or the experimental numbers are the falsified component.
The Figure 3 observation is suggestive rather than conclusive; we lack the exact axis ranges and numeric values to fully reconstruct the expected slope.
The Section 3.3 parameter exposure is flagged as a design concern, not a definitive proof break.
Formula rendering issues in Section 3.3 are noted for transparency but should be checked against the original PDF before any definitive judgement on the KeyGen derivation.