Offset-normalised Savitzky–Golay Channel Smoothing for Key Agreement in Heterogeneous LoRa V2X: Unified Benchmarking and Ablation Study

Physical-Layer key agreement using RSSI from heterogeneous LoRa transceivers is impaired by a persistent node-dependent RSSI reporting difference of 9.5-12.7 dB between the RFM95 and SX1262 modules, elevating prereconciliation key disagreement rate (KDR).This study proposes an offset-normalised Savitzky-Golay pipeline that corrects the node-dependent bias through median offset calibration followed by polynomial smoothing.The pipeline is evaluated on field measurements from eight LoRa V2V and V2I mobility conditions at 20-30 km/h and benchmarked against four baselines under a common evaluation protocol.An eight-configuration ablation study confirms offset normalisation as an effective first-stage correction (7.5% KDR reduction [50/25/25 split]) with Savitzky-Golay providing a further 22.4% reduction; BiLSTM prediction increases KDR to 0.4150 under the available dataset (50 or fewer training samples in seven of eight conditions).The Offset-SavGol configuration achieves KDR 0.1375 (V2I Straight, [60/20/20 split]) and median KDR 0.2417 (IQR 0.1222, [50/25/25 ablation split]); under a multi-objective criterion (median KDR, inter-condition IQR, degenerate output), Savitzky-Golay and Moving Average (median KDR 0.1302) are both non-dominated smoothing candidates.In the unified benchmark, the Offset-SavGol configuration attains agreed KGR of 0.8734 bps (highest among all seven evaluated pipelines) and total pure processing time of 3.67 ms on Raspberry Pi 5 (excluding acquisition interval) without model training.Security against eavesdropping remains subject to future investigation.

Authors

Institutions

Publication Details

Journal
International journal of intelligent engineering and systems
Published
2026-11-17
DOI
https://doi.org/10.22266/ijies2026.1231.38
Primary Topic
Vehicular Ad Hoc Networks (VANETs)
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Offset-normalised Savitzky–Golay Channel Smoothing for Key Agreement in Heterogeneous LoRa V2X: Unified Benchmarking and Ablation Study

Rini Puji Astutik, Mike Yuliana, Nailul Muna, Reni Soelistijorini et al.
1 links
International journal of intelligent engineering and systems
Vehicular Ad Hoc Networks (VANETs)
article

Offset-normalised Savitzky–Golay Channel Smoothing for Key Agreement in Heterogeneous LoRa V2X: Unified Benchmarking and Ablation Study

Rini Puji Astutik, Mike Yuliana, Nailul Muna, Reni Soelistijorini, Tri Budi Santoso, Wenny Mistarika Rahmawati
article en

Abstract

Physical-Layer key agreement using RSSI from heterogeneous LoRa transceivers is impaired by a persistent node-dependent RSSI reporting difference of 9.5-12.7 dB between the RFM95 and SX1262 modules, elevating prereconciliation key disagreement rate (KDR).This study proposes an offset-normalised Savitzky-Golay pipeline that corrects the node-dependent bias through median offset calibration followed by polynomial smoothing.The pipeline is evaluated on field measurements from eight LoRa V2V and V2I mobility conditions at 20-30 km/h and benchmarked against four baselines under a common evaluation protocol.An eight-configuration ablation study confirms offset normalisation as an effective first-stage correction (7.5% KDR reduction [50/25/25 split]) with Savitzky-Golay providing a further 22.4% reduction; BiLSTM prediction increases KDR to 0.4150 under the available dataset (50 or fewer training samples in seven of eight conditions).The Offset-SavGol configuration achieves KDR 0.1375 (V2I Straight, [60/20/20 split]) and median KDR 0.2417 (IQR 0.1222, [50/25/25 ablation split]); under a multi-objective criterion (median KDR, inter-condition IQR, degenerate output), Savitzky-Golay and Moving Average (median KDR 0.1302) are both non-dominated smoothing candidates.In the unified benchmark, the Offset-SavGol configuration attains agreed KGR of 0.8734 bps (highest among all seven evaluated pipelines) and total pure processing time of 3.67 ms on Raspberry Pi 5 (excluding acquisition interval) without model training.Security against eavesdropping remains subject to future investigation.

International journal of intelligent engineering and systemsVol. 19(12)
Universitas Muhammadiyah Gresik (ID), Politeknik Elektronika Negeri Surabaya (ID)
Vehicular Ad Hoc Networks (VANETs)
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