Descriptor: A Hybrid Indoor and Indoor-Outdoor Positioning Multi-Technology Dataset (HYMN)

Muhammad Ammad , Albrecht Michler , Paul Schwarzbach , Jonas Ninnemann , Hagen Ußler , AND Oliver Michler · 2026

Read Descriptor: A Hybrid Indoor and Indoor-Outdoor Positioning Multi-Technology Dataset (HYMN) by Muhammad Ammad , Albrecht Michler , Paul Schwarzbach , Jonas Ninnemann , Hagen Ußler , AND Oliver Michler (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-20349.txt.

Introduction

Ammad and five colleagues present HYMN, a hybrid multi-tech navigation data set for indoor-outdoor positioning research. Five systems feed it in a factory hall: ultra-wideband, Bluetooth Low Energy, WiFi, 5G, and satellite navigation, all time-synced. Unlike single-tech sets, HYMN covers indoor-outdoor shifts, backing multi-sensor fusion for seamless positioning. Each record carries a unique measurement id with time-stamped notes and ground truth, aiding fingerprinting, cross-tech fusion, and seamless positioning.

Catalog record

CREATORMuhammad Ammad , Albrecht Michler , Paul Schwarzbach , Jonas Ninnemann , Hagen Ußler , AND Oliver Michler
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36094
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.20349

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Descriptor: A Hybrid Indoor and Indoor-Outdoor Positioning Multi-Technology Dataset (HYMN) Muhammad Ammad , Albrecht Michler , Paul Schwarzbach , Jonas Ninnemann , Hagen Ußler , AND Oliver Michler Abstract This article introduces the HYMN ( HY brid M ulti-technology N avigation) dataset: a multi-system, and time synchronized dataset for localization research based on opportunistic signals collected in an indoor-outdoor scenario. HYMN comprises measurement data collected in an industrial hall setting for five different positioning systems including Ultra-Wideband (UWB), Bluetooth Low Energy (BLE), WiFi, 5G, and Global Navigation Satellite System (GNSS). Unlike existing datasets that focus on single technologies or purely indoor/outdoor scenarios, HYMN combines five positioning technologies with explicit coverage of indoor-outdoor transitions, enabling multi-sensor fusion research for seamless localization. Each instance of data is identified through a unique measurement id and it represents time-stamped observations relevant for each system respectively along with the ground truth information. HYMN is designed to support a wide range of localization tasks including multi-sensor fingerprinting, cross-technology fusion, and seamless indoor-outdoor positioning. The synchronized measurements from GNSS and other terrestrial systems enable researchers to investigate how heterogeneous signals complement each other to overcome individual technology limitations such as GNSS degradation in covered areas or terrestrial system variability in dynamic environments. \abstractheadfont DATA DOI/PID DOI: 10.5281/zenodo.17979434 \abstractheadfont DATA TYPE/LOCATION Ranging data; industrial hall; Germany † † corresponding: CORRESPONDING AUTHOR: Muhammad Ammad (e-mail: [email protected]). BACKGROUND Localization has gained significant attention in recent years, however the enabling radio systems are confronted with heterogenous challenges in different scenarios. Outdoor environments rely primarily on GNSS for localization which performs exceptionally well in open skies but its functionality is limited in covered areas for indoor positioning. The indoor-outdoor positioning requires seamless integration of the two technologies used for both the indoor and outdoor environments. Outdoor to indoor transitioning while positioning poses a lot of difficulties since it requires precise switching between the technology being used outside (e.g., GNSS) and inside (e.g, Wi-Fi, 5G, etc.) for positioning [ 30 ] . Accurate positioning within these transitional areas, which bridges indoor and outdoor environments, are often difficult to obtain with standard positioning models. Developing navigation algorithms involving such transitional areas requires extensive sensor readings [ 27 ] . Studies show that conventional GNSS solutions achieve reliable outdoor accuracy but their performance deteriorates in obstructed or indoor areas due to multipath reflections, non line-of-sight (NLOS) interference, and signal blockage, thereby motivating the use of additional sensing modalities for continuous and resilient localization [ 6 , 26 ] . Terrestrial and short-range positioning technologies such as WiFi, UWB, and BLE offer improved coverage and continuity in indoor or partially GNSS-denied environments. However, their performance is constrained by environmental dynamics and signal variability, motivating continued research into sensor fusion, and data-driven adaptation across heterogeneous settings [ 16 , 17 ] . Complementary datasets and benchmarks that span multiple/mixed environments are critical for evaluating observability-aware estimators, multipath/NLOS resilience, and cross-system fusion strategies, thereby advancing robust localization under dynamic environmental conditions, sensor failures, and diverse deployment scenarios [ 7 , 29 ] . However, a systematic review of 119 indoor positioning datasets reveals that 50% lack information on collection area size, 21% omit ground truth protocol documentation, and 13% have undefined environment types, indicating that critical metadata required for reproducible benchmarking remains widely absent [ 3 ] . Collectively, these studies highlight the necessity of multi-modal dataset, and fusion methodologies to ensure reliable performance across environments. TABLE I: Overview of sensor systems and their identifiers System System ID Sensor/Hardware Measurement Principle Anchors GNSS gnss NovAtel PwrPak 7 Pseudorange varying UWB uwb ZigPos RTLS Two-Way Ranging (TWR) / RTT 10 Bluetooth LE ble Metirionic DMK-215 Phase-Based Ranging (PBR) 5 Wi-Fi wifi HPE Aruba 630 Series APs Fine Timing Measurement (FTM, RTT) 6 5G NR nr5g Industrial Small Cell System Fingerprinting (SNR, TDOA-based) 3 Ground Truth ref Leica TS16 Total Station Optical survey (total station) 1 The rapid AI development in the localization domain has fundamentally transformed the landscape of positioning systems, enabling adaptability across diverse environments [ 5 , 25 ] . However, this technological progress has created a requirement for large-scale, high-quality training and benchmarking datasets that support the advancement, validation, and comparison of data-driven localization approaches [ 28 ] . Data-driven methods such as fingerprinting-based approaches, neural network-based ranging, and channel state information learning, require diverse and real-world measurements to generate effective localization results across varied environmental conditions and deployment scenarios. Across indoor positioning benchmarks, ground truth rigor and metadata completeness predict localization performance more reliably than dataset size or access point density [ 4 ] . Several indoor localization public datasets have been developed, including UJIIndoorLoc [ 14 ] , Tampere University datasets [ 19 , 18 ] , and several other WiFi-fingerprinting, BLE RSSI, and UWB ranging datasets [ 24 , 20 , 23 ] . However, these datasets suffer from limitations that restrict their utility for comprehensive algorithm development and evaluation. Notably, signal sparsity in established benchmarks can reach 93–97%, and device-induced measurement variance spans over two orders of magnitude, posing challenges for models that aim for device-independent generalization [ 4 ] . There also exists datasets for GNSS-based localization addressing urban canyon scenarios such as UrbanNav [ 13 ] , and android raw GNSS measurement dataset [ 9 ] . Multi-modal datasets combining radio signals with other sensor modalities such as inertial or visual data have also been proposed [ 1 ] . The existing datasets do not explicitly address seamless indoor-outdoor transitions or mobile user scenarios. Furthermore, the most significant gap in the current localization datasets is the rarity of hybrid or multi-sensor datasets, limiting research on multi-technology fusion approaches. Hybrid and opportunistic localization remains constrained by the lack of open, multi-modal, and time-synchronized datasets that include indoor, and indoor-outdoor transitions in realistic settings. Various radio technologies have been studied for indoor and outdoor positioning. However, these are evaluated individually and there are no datasets combining all these technologies to perform indoor-outdoor localization using signals from multiple systems. Moreover, only 22.4% of positioning publications utilize open data and merely 7.9% provide both open data and code [ 3 ] , further constraining reproducibility and cross-study comparison in the field. To address this gap, this paper presents HYMN, a time-synchronized dataset encompassing five positioning technologies (GNSS, UWB, BLE, WiFi, and 5G) prepared for opportunistic radio-signal availability with explicit coverage of indoor, outdoor, and transitional environments, enabling research on opportunistic multi-technology fusion for seamless localization. COLLECTION METHODS AND DESIGN -A System Architecture The system architecture developed for HYMN data collection is set up inside an industrial hall measuring 44 m x 18 m in Torgau, Germany. The hall has a driveway with gates on each side making it perfect for indoor-outdoor positioning measurements. The architecture incorporates a combination of sensor technologies including GNSS, UWB, BLE, Wi-Fi, and 5G. These technologies when collectively used, provide an opportunistic approach facilitating effective localization. Table I gives a brief overview about the technologies used in the system architecture along with their specifications. UWB technology is utilized for its ability to provide precise distance measurements through the two-way ranging measurement principle. BLE technology is leveraged for its low power consumption and effective short-range communication. The Metirionic DMK-215 [ 10

Read the complete text