Machine Learning and Deep Learning-Based Multi-Attribute Physical-Layer Authentication for Spoofing Detection in LoRaWAN

Aug 4, 2026

Upload by Sage

Wireless sensor networks used in environmental monitoring, smart agriculture, and industrial automation face severe security risks because wireless broadcasts enable attackers to tamper with authenticity. This work proposes a multi-attribute physical-layer authentication scheme for LoRaWAN that leverages physical attributes including RSSI, battery level, and altitude. The study focuses on the LoRaWAN join procedure, where early communication involves plain-text transmission without encryption. A partially synthesized dataset supports evaluation using real RSSI measurements and simulated battery and altitude in a forest-fire scenario.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Introduction

IoT and WSN security challenges

LoRaWAN attack surface and spoofing

Proposed Multi-Attribute PLA Scheme

LoRaWAN join procedure risk

Multi-attribute features (RSSI, battery level, altitude)

Dataset and Experimental Setup

Partially synthesized dataset design

Real RSSI source and simulated sensor behavior

ML and DL Models Evaluation

Logistic Regression, Random Forest, KNN

MLP and CNN performance comparison

Results and Discussion

Accuracy and resource-demand trade-offs

Article Machine Learning and Deep Learning-Based Multi-Attribute Physical-Layer Authentication for Spoofing Detection in LoRaWAN Azita Pourghasem *, Raimund Kirner , Athanasios Tsokanos, Iosif Mporas * and Alexios Mylonas Academic Editors: Emanuele De Santis and Francesco Delli Priscoli Received: 6 November 2024 Revised: 14 January 2025 Accepted: 27 January 2025 Published: 6 February 2025 Citation: Pourghasem, A.; Kirner, R.; Tsokanos, A.; Mporas, I.; Mylonas, A. Machine Learning and Deep Learning-Based Multi-Attribute Physical-Layer Authentication for Spoofing Detection in LoRaWAN. Future Internet 2025, 17, 68 . [https://doi.org/10.3390/fi17020068](https://doi.org/10.3390/fi17020068) [Copyright:](Copyright:) © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://creativecommons.org/](https://creativecommons.org/)[ ](https://creativecommons.org/)[licenses/by/4.0/](licenses/by/4.0/)) . Cybersecurity and Computing Systems Research Group, Department of Computer Science, University of Hertfordshire, Hatfield AL10 9AB, UK * Correspondence: [a.pourghasem@herts.ac.uk](a.pourghasem@herts.ac.uk) (A.P.); [i.mporas@herts.ac.uk](i.mporas@herts.ac.uk) (I.M.) Abstract: The use of wireless sensor networks (WSNs) in critical applications such as environmental monitoring, smart agriculture, and industrial automation has created significant security concerns, particularly due to the broadcasting nature of wireless communication. The absence of physical-layer authentication mechanisms exposes these networks to threats like spoofing, compromising data authenticity. This paper introduces a multi-attribute physical layer authentication (PLA) scheme to enhance WSN security by using physical attributes such as received signal strength indicator (RSSI), battery level (BL), and altitude. The LoRaWAN join procedure, a key risk due to plain text transmission without encryption during initial communication, is addressed in this study. To evaluate the proposed approach, a partially synthesized dataset was developed. Real-world RSSI values were sourced from the LoRa at the Edge Dataset, while BL and altitude columns were added to simulate realistic sensor behavior in a forest fire detection scenario. Machine learning (ML) models, including Logistic Regression (LR), Random Forest (RF), and K-Nearest Neighbors (KNN), were compared with deep learning (DL) models, such as Multi-Layer Perceptron (MLP) and Convolutional Neural Networks (CNN) . The results showed that RF achieved the highest accuracy among machine learning models, while MLP and CNN delivered competitive performance with higher resource demands. Keywords: wireless sensor networks; physical-layer authentication; deep learning; machine learning; spoofing; multi-attribute; altitude; radio frequency fingerprinting; battery level; RSSI; LoRaWAN 1. Introduction IoT has become an important technology across multiple sectors, impacting industries such as healthcare, smart cities, agriculture, environmental monitoring, and more. IoT, as defined by Ref. [1], is a complex system of entities—comprising cyber-physical devices, information resources, and people—that exchange information and interact with the physical world through sensing, processing, and actuating. In recent years, the integration of Artificial Intelligence (AI) with IoT has further accelerated technological advancements, created new use cases, and enhanced the capabilities of next-generation connected systems [2] . According to Ref. [3], combining AI and IoT will transform industries by creating smarter and more flexible networks and infrastructure. An essential component within the IoT ecosystem is WSNs, consisting of numerous sensor devices deployed to gather environmental data such as temperature and humidity [4] . Based on Ref. [5], WSNs are experiencing rapid grow

FAQs

Why does LoRaWAN spoofing threaten wireless sensor networks?

LoRaWAN’s broadcasting nature makes networks susceptible to eavesdropping, data forgery, and spoofing. Spoofing lets an attacker impersonate a legitimate sensor by falsifying attributes such as signal strength or location, undermining authenticity and reliability.

Which physical attributes are used for the proposed physical-layer authentication scheme?

The scheme uses multi-attribute physical-layer features including RSSI, battery level (BL), and altitude to strengthen authentication for LoRaWAN communications.

How were machine learning and deep learning models evaluated?

Models including Logistic Regression, Random Forest, and KNN were compared with deep learning models such as MLP and CNN using a partially synthesized dataset. Real-world RSSI values were sourced from the LoRa at the Edge dataset, while battery level and altitude were added to simulate realistic sensor behavior for a forest fire detection scenario.

Aug 4, 2026

Upload by Sage