Direction-of-Arrival-based Indoor Localization Systems for Massive IoT Networks : An Embedded Implementation Perspective
Troccoli Cunha, Tiago (2024)
Tampere University
2024
Tieto- ja sähkötekniikan tohtoriohjelma - Doctoral Programme in Computing and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Väitöspäivä
2024-09-27
Julkaisun pysyvä osoite on
https://urn.fi/URN:ISBN:978-952-03-3596-0
https://urn.fi/URN:ISBN:978-952-03-3596-0
Tiivistelmä
This thesis investigates cost-effective, accurate, and energy-efficient Directionof- Arrival (DOA) methods for DECT-2020 NR and Bluetooth Low Energy 5.1+ indoor localization solutions. The former lacks positioning capabilities, while the latter provides a protocol for direction finding but does not offer any DOA method, leaving the crucial development aspect to external entities. These standards support Massive Internet of Things (IoT) networks, which interconnect an unparalleled number of low-cost and battery-operated smart sensors and actuators.
IoT-based indoor localization revolutionizes inventory management, and industrial safety, extending its benefits to retail, smart homes, and emergency response by offering location-based services. Additionally, it enhances operational efficiency and safety by directing driverless forklifts in warehouses, navigating mobile robots in healthcare and hospitality, facilitating synchronized processes in industrial automation, and enabling real-time tracking of assets across diverse settings.
However, integrating DOA-based localization in massive IoT networks poses challenges due to devices’ cost, computational, and power constraints, especially those lacking a dedicated Floating-Point Unit (FPU). Our research tackles these challenges by developing innovative DOA methods, including three solutions designed for battery-operated IoT devices that work with array sequential sampling. These cost-efficient DOA solutions employ uniform L-shaped arrays with a single Radio-Frequency (RF) chain, allowing each antenna to sequentially sample the signal through an RF switch.
Our study presents bare-metal DOA implementations on System-on-Chips, evaluating their accuracy, execution time, energy consumption, and memory footprint under multipath and noise channels. The results validate our methods, with the top performers achieving under three milliseconds execution time, sub-degree accuracy at 15 dB SNR and higher, and keeping the memory footprint under 20 kB. We also developed a pioneering fixed-point DOA method that boosts energy efficiency by 5.9 times and computational speed by 4.4 times, marking a significant improvement over the conventional software-based floating-point methods used by IoT devices without a floating-point unit.
IoT-based indoor localization revolutionizes inventory management, and industrial safety, extending its benefits to retail, smart homes, and emergency response by offering location-based services. Additionally, it enhances operational efficiency and safety by directing driverless forklifts in warehouses, navigating mobile robots in healthcare and hospitality, facilitating synchronized processes in industrial automation, and enabling real-time tracking of assets across diverse settings.
However, integrating DOA-based localization in massive IoT networks poses challenges due to devices’ cost, computational, and power constraints, especially those lacking a dedicated Floating-Point Unit (FPU). Our research tackles these challenges by developing innovative DOA methods, including three solutions designed for battery-operated IoT devices that work with array sequential sampling. These cost-efficient DOA solutions employ uniform L-shaped arrays with a single Radio-Frequency (RF) chain, allowing each antenna to sequentially sample the signal through an RF switch.
Our study presents bare-metal DOA implementations on System-on-Chips, evaluating their accuracy, execution time, energy consumption, and memory footprint under multipath and noise channels. The results validate our methods, with the top performers achieving under three milliseconds execution time, sub-degree accuracy at 15 dB SNR and higher, and keeping the memory footprint under 20 kB. We also developed a pioneering fixed-point DOA method that boosts energy efficiency by 5.9 times and computational speed by 4.4 times, marking a significant improvement over the conventional software-based floating-point methods used by IoT devices without a floating-point unit.
Kokoelmat
- Väitöskirjat [4864]