Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation
The widespread availability of sensors, improved computing, and storage capabilities, and ubiquity of networking services have led to the transformation of the conventional transportation services. Achieving the smart transportation goal can be either via dedicated or non-dedicated methods. The form...
Main Author: | |
---|---|
Other Authors: | |
Format: | Others |
Language: | en |
Published: |
Université d'Ottawa / University of Ottawa
2018
|
Online Access: | http://hdl.handle.net/10393/37961 http://dx.doi.org/10.20381/ruor-22219 |
id |
ndltd-uottawa.ca-oai-ruor.uottawa.ca-10393-37961 |
---|---|
record_format |
oai_dc |
collection |
NDLTD |
language |
en |
format |
Others
|
sources |
NDLTD |
description |
The widespread availability of sensors, improved computing, and storage capabilities, and ubiquity of networking services have led to the transformation of the conventional transportation services. Achieving the smart transportation goal can be either via dedicated or non-dedicated methods. The former denotes utilization of sensors that are explicitly deployed and configured for pre-defined sensing tasks whereas the latter exploits opportunistic and participatory sensing paradigms. With that motivation in mind, the contributions of this thesis are three-fold: \textit{(i)} Sensor emulation, \textit{(ii)} Crowd-sensing based GPS-less tracking, and \textit{(iii)} Reliable data acquisition in crowd-sensed GPS-less tracking. Emulating non-dedicated sensors in a simulation environment enables us to perform large-scale crowd-sensing tasks. We introduce a variety of vehicular crowd-sensing-based frameworks to track public transportation vehicles that move over static routes in a smart city setting without \textit{GPS}-enabled devices because of the major downsides of \textit{GPS} (e.g. high energy consumption, inaccurate localization in certain environments such as indoor, and privacy violation due to direct location sharing). To this end, we propose a novel framework, in our initial approach, to emulate the functionality of a sensor by using multiple available soft sensors and machine intelligence algorithms. As a case study, the localization of city buses in a smart city setting is investigated by using the accelerometer and microphones of the passengers and supervised machine intelligence running in the cloud. In this application, the \textit{GPS} functionality is emulated by using these two soft sensors. We evaluate our proposed scheme through simulations and show that the proposed framework can operate with more than 90\% accuracy in estimating the location of public buses while preserving the actual location privacy of the smartphone users. This approach results in smartphone battery energy savings of 38--46\% (as compared to \textit{GPS}-based approaches) due to the elimination of the power-hungry \textit{GPS} devices. Additional sensor recruitment schemes with various sensor combinations of accelerometer and microphone are developed in an extension work to examine both localization accuracy and energy consumption performance. Furthermore, in a subsequent work, crowd-sensed data undergoes an unsupervised machine learning module that estimates the location of the vehicle. We evaluate our proposed scheme through simulations and show that the proposed framework can operate with 95\% accuracy in estimating the location of public vehicles in the best case and with 80\% accuracy in the worst case. Since data trustworthiness is vital when data is crowd-solicited, assessment and quantification of the trustworthiness of participating sensors play a key role in the accuracy of the acquired information. To this end, we propose a reliability-aware participant recruitment scheme that assesses the trustworthiness of individual participants. Our simulation results lead to approximately 96\% localization accuracy under the reliability-aware recruitment scheme compared to 93\% localization accuracy the reliability-unaware participant recruitment scheme. Moreover, we introduce two trustworthiness-aware recruitment schemes in another work: Reliability-driven naive recruitment (\textit{RDNR}) and Reliability-driven exclusive recruitment (\textit{RDER}). We evaluate the performance of \textit{RDNR}, \textit{RDER}, and non-restrictive recruitment (reliability-unaware). Through simulations, we show that over 85\% and 98\% accuracy can be achieved in the worst and best cases, respectively, while consuming less energy than \textit{GPS}-based localization approaches. |
author2 |
Boukerche, Azzedine |
author_facet |
Boukerche, Azzedine Kaptan, Cem |
author |
Kaptan, Cem |
spellingShingle |
Kaptan, Cem Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation |
author_sort |
Kaptan, Cem |
title |
Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation |
title_short |
Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation |
title_full |
Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation |
title_fullStr |
Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation |
title_full_unstemmed |
Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation |
title_sort |
data analytics-backed vehicular crowd-sensing for gps-less tracking in public transportation |
publisher |
Université d'Ottawa / University of Ottawa |
publishDate |
2018 |
url |
http://hdl.handle.net/10393/37961 http://dx.doi.org/10.20381/ruor-22219 |
work_keys_str_mv |
AT kaptancem dataanalyticsbackedvehicularcrowdsensingforgpslesstrackinginpublictransportation |
_version_ |
1718969835753308160 |
spelling |
ndltd-uottawa.ca-oai-ruor.uottawa.ca-10393-379612019-01-29T06:19:44Z Data Analytics-backed Vehicular Crowd-sensing for GPS-less Tracking in Public Transportation Kaptan, Cem Boukerche, Azzedine Kantarci, Burak The widespread availability of sensors, improved computing, and storage capabilities, and ubiquity of networking services have led to the transformation of the conventional transportation services. Achieving the smart transportation goal can be either via dedicated or non-dedicated methods. The former denotes utilization of sensors that are explicitly deployed and configured for pre-defined sensing tasks whereas the latter exploits opportunistic and participatory sensing paradigms. With that motivation in mind, the contributions of this thesis are three-fold: \textit{(i)} Sensor emulation, \textit{(ii)} Crowd-sensing based GPS-less tracking, and \textit{(iii)} Reliable data acquisition in crowd-sensed GPS-less tracking. Emulating non-dedicated sensors in a simulation environment enables us to perform large-scale crowd-sensing tasks. We introduce a variety of vehicular crowd-sensing-based frameworks to track public transportation vehicles that move over static routes in a smart city setting without \textit{GPS}-enabled devices because of the major downsides of \textit{GPS} (e.g. high energy consumption, inaccurate localization in certain environments such as indoor, and privacy violation due to direct location sharing). To this end, we propose a novel framework, in our initial approach, to emulate the functionality of a sensor by using multiple available soft sensors and machine intelligence algorithms. As a case study, the localization of city buses in a smart city setting is investigated by using the accelerometer and microphones of the passengers and supervised machine intelligence running in the cloud. In this application, the \textit{GPS} functionality is emulated by using these two soft sensors. We evaluate our proposed scheme through simulations and show that the proposed framework can operate with more than 90\% accuracy in estimating the location of public buses while preserving the actual location privacy of the smartphone users. This approach results in smartphone battery energy savings of 38--46\% (as compared to \textit{GPS}-based approaches) due to the elimination of the power-hungry \textit{GPS} devices. Additional sensor recruitment schemes with various sensor combinations of accelerometer and microphone are developed in an extension work to examine both localization accuracy and energy consumption performance. Furthermore, in a subsequent work, crowd-sensed data undergoes an unsupervised machine learning module that estimates the location of the vehicle. We evaluate our proposed scheme through simulations and show that the proposed framework can operate with 95\% accuracy in estimating the location of public vehicles in the best case and with 80\% accuracy in the worst case. Since data trustworthiness is vital when data is crowd-solicited, assessment and quantification of the trustworthiness of participating sensors play a key role in the accuracy of the acquired information. To this end, we propose a reliability-aware participant recruitment scheme that assesses the trustworthiness of individual participants. Our simulation results lead to approximately 96\% localization accuracy under the reliability-aware recruitment scheme compared to 93\% localization accuracy the reliability-unaware participant recruitment scheme. Moreover, we introduce two trustworthiness-aware recruitment schemes in another work: Reliability-driven naive recruitment (\textit{RDNR}) and Reliability-driven exclusive recruitment (\textit{RDER}). We evaluate the performance of \textit{RDNR}, \textit{RDER}, and non-restrictive recruitment (reliability-unaware). Through simulations, we show that over 85\% and 98\% accuracy can be achieved in the worst and best cases, respectively, while consuming less energy than \textit{GPS}-based localization approaches. 2018-08-03T19:07:55Z 2018-08-03T19:07:55Z 2018-08-03 Thesis http://hdl.handle.net/10393/37961 http://dx.doi.org/10.20381/ruor-22219 en application/pdf Université d'Ottawa / University of Ottawa |