Sub-1 GHz Indoor RSSI-Based Localization - An Experimental Evaluation of Trilateration, Multilateration, and Machine Learning Fingerprinting Methods

Aug 5, 2026

Upload by Eden

Indoor localization for IoT and wireless sensor networks is evaluated using sub-1 GHz RSSI techniques in a smart home testbed with off-the-shelf 868/915 MHz transceivers. Hardware constraints are experimentally analyzed, including antenna and RSSI radiation patterns and multipath reflections, to identify optimal node placement. A low-overhead RSSI recording and forwarding scheme is implemented with under 420 ms cycle time. Trilateration and multilateration are compared for three and four receivers in LOS and non-LOS links, achieving 46%–89% room prediction accuracy, and minimum mean error of 1.49 m. A multinomial logistic regression fingerprinting approach reaches 97%–100% room classification accuracy from 25–30 RSSI samples, demonstrating feasible low-complexity RSSI-only localization.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Introduction

Indoor localization demand for WSNs and IoT

Limits of GPS indoors

Challenges in 2.4 GHz approaches

Motivation for sub-1 GHz bands

RSSI-based methods and multipath variability

Machine learning for RSSI localization

IEEE JOURNAL OF SELECTED AREAS IN SENSORS, VOL. 2, 2025 121 Sub-1 GHz Indoor RSSI-Based Localization: An Experimental Evaluation of Trilateration, Multilateration, and Machine Learning Fingerprinting Methods Ben McPartlin and Mahmoud Wagih Abstract—As wireless radiofrequency-based localization techniques continue to attract interest, a plethora of approaches including received signal strength indicator (RSSI) trilateration and multilateration, phase, time-ofarrival, and machine learning models have been explored for indoor localization. However, there has been no comprehensive experimental investigations that compared the accuracy of these methods in a practical Internet of Things (IoT) wireless sensor network. Herein, we present a holistic evaluation of localization techniques in an indoor smart home environment, based on off-the-shelf 868/915 MHz transceivers. First, the hardware limitations, such as the antenna and RSSI radiation patterns and the effects of multipath reﬂections are experimentally investigated, identifying the optimal node placement. A practical RSSI recording and forwarding scheme is proposed and implemented using microcontroller units, showing a frugal approach for joint sensing and communication, with under 420 ms cycle time. Using this testbed, we compare multilateration approaches for three and four receivers, in both line-of-sight (LOS) and non-LOS links, achieving between 46% and 89% room prediction accuracy, with a minimum mean error of 1.49 m. A machine learning-based approach, using multinomial logistic regression, is then reported with a peak room classiﬁcation accuracy of 97%–100%, for 25–30 RSSI points. A comparison with state-of-the-art implementations is presented showing a high room localization accuracy at a low hardware complexity, demonstrating the feasibility of RSSI-only localization in resource-constrained IoT networks. Index Terms—Antennas, localization, received signal strength indicator (RSSI), sub-1 GHz, trilateration, wireless sensor networks (WSNs). Received 14 September 2024; revised 10 February 2025; accepted 23 February 2025 . Date of publication 26 February 2025; date of current version 2 April 2025 . This work was supported in part by the U.K. EPSRC under Grant “EDIBLES” EP/Y002008/1, and in part by the Royal Society under Grant “STEMS” RGS/R1/231028 . Recommended by Lead Guest Editor Pai-Yen Chen and Guest Editor Paul C. P. Chao.(Corresponding author: Mahmoud Wagih.) The authors are with the Green RF-Enabled Electronics Lab, James Watt School of Engineering, University of Glasgow, G12 8QQ Glasgow, U.K. ([e-mail: mahmoud.wagih@glasgow.ac.uk](e-mail: mahmoud.wagih@glasgow.ac.uk)). Datasets underpinning this paper will be made accessible through the Universiy repository at DOI 10.5525/gla.researchdata.1867 and Github repository: [github.com/benmcp11/HProject.](github.com/benmcp11/HProject.) [Digital Object Identi](Digital Object Identi)ﬁ[er 10.1109/JSAS.2025.3545784](er 10.1109/JSAS.2025.3545784) I. INTRODUCTION THE emergence of wireless sensor networks (WSNs) and Internet of Things (IoT) has generated a demand for indoor localization. Technologies ranging from smart homes to patient monitoring all necessitate the precise knowledge of the location of nodes on a network [1], [2], [3], [4], [5] . In an outdoor environment, the globalpositioning system(GPS)wouldbe used to accurately determine the location of these nodes. However, GPS receivers(RX)cannot reliably detect satellite transmissions indoors due to signals being blocked by the building’s exterior [1] . Therefore, GPS is not an accurate or reliable tool for indoor localization, and thus other methods must be developed to meet this ever growing demand. In recent years, there has been signiﬁcant interest in indoor localization methods utilizing the 2.4 GHz frequency band, particularly with technologies, such as ZigBee and LoRa [1],[6], [7] . However, these higher frequency bands often face challenges

FAQs

What localization methods are experimentally compared in the smart home testbed?

The paper compares RSSI-based trilateration and multilateration, and a machine learning fingerprinting approach using multinomial logistic regression in the same indoor environment.

How does the work address hardware-related factors affecting RSSI localization?

It experimentally investigates antenna and RSSI radiation patterns and the impact of multipath reflections, then identifies an optimal node placement for improved results.

What localization accuracy is reported for multilateration and for the machine learning approach?

Multilateration with three or four receivers achieves 46%–89% room prediction accuracy with a minimum mean error of 1.49 m. The machine learning fingerprinting method attains 97%–100% peak room class classification accuracy using 25–30 RSSI points.

Aug 5, 2026

Upload by Eden