THESIS
2022
1 online resource (ix, 45 pages) : illustrations (some color)
Abstract
Bluetooth low energy (BLE) beacon network is commonly adopted to provide many
emerging and smart IoT services. A BLE beacon network is usually formed by either
battery-powered beacon, or energy harvesting beacon recently. The on-going monitoring
of the energy status in these beacon devices is critical for the timely battery replacement
and the reliable application operation. Some earlier works utilize the user smartphones
to collect the energy status of BLE beacons bearing the users and report the data to
the central monitoring platform. However, this approach can induce a lot of reporting
updates that causes heavy loading on the network and server for continuous monitoring.
It may suffer from the poor monitoring performance and severe data loss of energy status
when some beacon devices...[
Read more ]
Bluetooth low energy (BLE) beacon network is commonly adopted to provide many
emerging and smart IoT services. A BLE beacon network is usually formed by either
battery-powered beacon, or energy harvesting beacon recently. The on-going monitoring
of the energy status in these beacon devices is critical for the timely battery replacement
and the reliable application operation. Some earlier works utilize the user smartphones
to collect the energy status of BLE beacons bearing the users and report the data to
the central monitoring platform. However, this approach can induce a lot of reporting
updates that causes heavy loading on the network and server for continuous monitoring.
It may suffer from the poor monitoring performance and severe data loss of energy status
when some beacon devices do not have enough chances of user passing by. This kind of
data loss is even more severe for the energy-harvesting beacons due to its property of
rapid fluctuations of energy status. Hence, the thesis first introduced an efficient reporting
framework of energy status for battery-powered beacons that can significantly reduce
the amount of reporting updates. This framework can remove unnecessary reporting by
smartly extending the report intervals required on every user smartphone without compromising
the monitoring performance. In addition, another data recovery framework of
energy status for energy-harvesting beacons is proposed, which can deal with the severe
data loss even for energy-harvesting beacon. A recurrent architecture of support vector
regression is adopted to learn the rapid changes of the energy status in energy-harvesting
beacons. Both frameworks are validated with the dataset collected from the real beacon
networks. Our proposed reporting framework can reduce the amount of reporting traffic
up to 70% for the 99% of estimation accuracy. Whereas, our recovery framework is also
proved to achieve the 90% of estimation accuracy even under a severe loss rate of data.
Beside the publications, the contributions of this thesis also include the prototypes of
our proposed frameworks, which are possible to support effective and reliable monitoring
of energy status for future BLE beacon networks.
Post a Comment