Forensic Review Report: SD-IoTR: an SDN-based Internet of Things reprogramming framework

Academic fraud report · Geng Detector

Summary

This review assesses concerns of research integrity in the 2020 IET Networks paper "SD-IoTR: an SDN-based Internet of Things reprogramming framework" by Dang Huynh-Van and Quan Le-Trung (DOI: 10.1049/iet-net.2019.0223). The overall verdict is highly suspicious. Three confirmed findings are reported: (1) a missing experimental group (Version 2) in Table 4 despite being described in Section 4.1.1, suggesting selective reporting; (2) internal arithmetic inconsistencies in Table 4, where mean bytes/s values do not reconcile with total bytes divided by time, and packet counts are inconsistent with total bytes divided by average packet size; (3) incomplete statistical reporting, with a 95% confidence interval quoted in Section 4.2.2 without the underlying sample size n. One additional concern (figures) could not be verified due to lack of original image data. The arithmetic errors and missing group are reproducible from the paper text and constitute credible evidence of sloppy or potentially fabricated data handling. Image-based manipulation analysis was not possible.

Verdict

Highly suspicious. Multiple confirmed issues, including a missing experimental group and irreconcilable arithmetic in reported data, warrant a request for the authors' raw experimental logs.

Key findings

Missing experimental group (Version 2): Section 4.1.1 describes three application versions (Version 1, Version 2, Version 3), but Table 4 only reports Version 1 and Version 3, with no explanation of the omission. This raises concerns of selective reporting.
Arithmetic inconsistencies in Table 4:
Version 1: 270,768 Bytes over 101 s yields 2,680.87 Bytes/s, but the table reports 2,663.6 Bytes/s.
Version 3: 246,326 Bytes over 89 s yields 2,767.7 Bytes/s, but the table reports 2,788.037 Bytes/s.
Version 3: 246,326 Bytes ÷ 113 Bytes/packet = 2,180 packets, but the table reports 2,186 packets.
Incomplete statistical reporting: Section 4.2.2 reports "35.85 ± 1.6% CPU (95% confidence)" without stating the sample size n, preventing verification of the interval.
Unverifiable figures: Fig. 5, Fig. 7, and Fig. 8 could not be analyzed at pixel level due to lack of original high-resolution images.

Evidence highlights

Table 4 (Page 312) contains the contradictory numeric entries listed above; all values are reproduced exactly as printed in the paper.
Section 4.1.1 explicitly references three versions, while Table 4 displays only two.
Section 4.2.2 cites a 95% confidence interval without sample size, unlike Section 4.1.2 which states n = 10.
DOI: 10.1049/iet-net.2019.0223

Notes

All numeric discrepancies are reproducible from the manuscript text alone.
Image manipulation analysis (e.g., pixel-level checks of Fig. 5, Fig. 7, Fig. 8) was not feasible in this review; conclusions on figure integrity are limited.
The findings do not, by themselves, prove intentional fabrication; alternative explanations include transcription or calculation errors. However, the combination of a missing group and multiple internal numerical contradictions in the central results table is unusual and concerning.
Recommended follow-up: request raw experimental logs and packet captures from the authors to reconcile Table 4, and clarify why Version 2 is absent.