|
Title
|
|
|
|
Using sub-GHz radio frequency communication for device-free detection, tracking and identification
| |
|
Author
|
|
|
|
| |
|
Abstract
|
|
|
|
| The research project that led to the creation of this PhD thesis investigated the use of sub-GHz frequencies within the field of radio frequency based device-free localization (DFL). Localization techniques within this field are capable of detecting, tracking and/or identifying (primarily human) targets, without requiring these targets to wear any kind of device or tag. Instead, the physicial influence the presence of a target has on its environment is used. Within the current state-of-the-art, the vast majority of systems make use of 2.4 GHz signals and studies regarding potential advantages of sub-GHz frequencies tend to be rather limited. First of all, an RF sensor network was created which consisted of multiple transceiver nodes capable of regularly communicating with each other. Communication made use of the DASH7 Alliance Protocol (D7A) and occurred on the 433 MHz and 868 MHz frequency bands. The Received Signal Strengths (RSS) with which signals within this network are received by the nodes, are regularly communicated to a centralized system were this data can be used as input to a DFL algorithm. This system was continuously being developed over the course of the experiments and it formed a key component of this research. I investigated the accuracy of the tagless localization technique Radio Tomographic Imaging (RTI) for the use of 433 MHz and 868 MHz communication. Experiments in an empty classroom environment of approximately 60 m2 indicated that shadowing-based RTI was capable of locating a stationary human target with sub-meter accuracy, provided that the data from both frequency bands was used. The accuracy declined rapidly, however, when a second experiment was performed in a more complex environment which consisted of two office rooms connected by a hallway. This was a clear indication that more advanced RTI approaches were required in complex environments causing more multipath effects. The use of sub-GHz frequencies was also investigated in the context of Passive Fingerprinting. These experiments were not solely focused on determining the position of a target, but also on differentiating between different targets (identification). Although difficult, differentiation based on body type was shown to be feasible and this has opened up several interesting research paths. The previously mentioned setups were all installed in relatively small indoor environments, but experiments were also performed in large-scale festival environments capable of containing thousands of humans. Here, the goal was not to locate all of these individuals. instead, it was to estimate the size of the crowds. The use of sub-GHz frequency signals in this context was quite logical, given their increased range when compared to signals of a higher frequency which are transmitted with similar amounts of power. The average attenuation measured within the RF sensor network when compared to a calibration measurement when the environment was unoccupied, was found to be highly indicative of crowd size. Multiple experiments in a variety of environments showed strong positive correlations between this value and the true amount of humans present. Given the importance of accurate, real-time information regarding the size of crowds for crowd safety management, I believe that this technique can be highly useful for organisers of large-scale events. Finally, an extensive survey of the broader field of DFL research was created as well. Based on this overview, a high-level analysis was performed regarding potential future research directions for the field as a whole. I consider the lack of publicly available datasets to be one of the most important issues within the current state of the art. To lead by example, several of data sets resulting from the experiments described in this thesis were made publicly available. |
| |
|
Language
|
|
|
|
English
| |
|
Publication
|
|
|
|
Antwerpen
:
Universiteit Antwerpen, Faculteit Toegepaste Ingenieurswetenschappen
,
2020
| |
|
Volume/pages
|
|
|
|
245 p.
| |
|
Note
|
|
|
|
:
Weyn, Maarten [Supervisor]
:
Berkvens, Rafael [Supervisor]
| |
|
Full text (open access)
|
|
|
|
| |
|