Integrity Investigation Report: 'Secure MQTT for Internet of Things (IoT)' — Internal Data Inconsistency and Likely Fabricated Performance Curves

Academic fraud report · Geng Detector

Summary

Verdict: Severe internal inconsistency in experimental data indicating likely fabrication of performance figures. The paper reports in Table VI that increasing the CP-ABE key length from 256 to 512 bits raises setup time from 588 ms to 19,177 ms (~32x increase), and KP-ABE setup time from 187 ms to 4,307 ms (~23x increase). Yet the corresponding curves in Figure 6 (Packet Overhead) and Figure 7 (Encryption Performance) for 256-bit and 512-bit variants are visually reported as pixel-level identical, contradicting cryptographic scaling laws. In ECC-based ABE, ciphertext size and pairing operation cost must scale with key length. The internal contradiction between the table and figures is mathematically incompatible. Additional concerns include reliance on only 3 trial iterations without variance reporting. Confidence is moderate-high based on the structural contradiction; absolute confirmation would require the original raw data.

Verdict

🔴 Confirmed structural inconsistency between tabular and graphical experimental data, with strong indicators of fabricated or duplicated figure data. Recommended for editorial investigation and raw-data audit.

Key findings

Mathematical contradiction between Table VI and Figures 6/7: Table VI reports CP-ABE setup time of 588 ms (256-bit) vs 19,177 ms (512-bit) and KP-ABE setup time of 187 ms (256-bit) vs 4,307 ms (512-bit), a 23–32x increase. Figures 6 and 7 nonetheless show 256-bit and 512-bit curves as visually identical.
Violation of cryptographic scaling expectations: In ECC-based Attribute-Based Encryption, ciphertext size and pairing-based encryption time must scale with key length; pixel-level overlap of 256-bit and 512-bit curves is physically implausible.
Suspected data duplication in plotting: The reported perfect coincidence of two distinct parameter sets most plausibly indicates copy-paste reuse of a single data series relabeled as two configurations.
Statistically insufficient evaluation methodology: Only 3 experimental iterations were used to produce averaged results, with no reported variance or standard deviation — inadequate for cryptographic benchmarking and consistent with the absence of expected measurement noise.

Evidence highlights

DOI: 10.1109/CSNT.2015.16
CP-ABE setup: 256-bit = 588 ms; 512-bit = 19,177 ms (Table VI)
KP-ABE setup: 256-bit = 187 ms; 512-bit = 4,307 ms (Table VI)
Figure 6 (Packet Overhead): 'Packet size KP 256' and 'Packet size KP 512' curves reported as fully overlapping across all attribute counts.
Figure 7 (Encryption Performance): 'Enc KP 256' and 'Enc KP 512' curves reported as fully overlapping.
Methodology claim in Section VII.B (Results): experiments repeated three times to compute averages; no variance metrics reported.

Notes

This report is AI-assisted and intended to support academic discussion only. A final determination of misconduct requires formal investigation with access to the authors' raw experimental data and code.
Limitations: the analysis relies on numerical claims from Table VI and visual descriptions of Figures 6 and 7. Independent image analysis (e.g., pixel-comparison tools, EXIF metadata, source-file forensics) was not performed within the scope of this report and could strengthen or refine the conclusion.
If authors provide raw datasets and plotting scripts, the apparent contradiction may potentially be reconciled — however, the magnitude of the table-reported key-length scaling makes coincidental curve overlap highly unlikely.