Abstract:
The drones are now frequently used for many smart
city applications including intelligent transportation to provide
situational awareness for drivers as well as other stakeholders
that manage the traffic. In such situations one of the widely
collected data is video that is recorded by a drone and streamed
in real-time to a remote control center. The data can then
be accessed through cloud services to do further analysis and
take actions. However, this captured video may contain private
information from the passing by citizens and allow recognition
and tracking if it is intercepted by malicious users. While the
video data can be stored as encrypted in the cloud, this still does
not address the privacy problem as the third party providers
still need to decrypt the data to perform any further processing.
To address this issue, we propose using fully homomorphic
encryption (FHE) which will not only provide confidentiality of
the data but also enable processing on the encrypted video data
by cloud providers and other third parties without exposing any
privacy. However, since fully homomorphic systems have a lot
of overhead, in this paper, we propose to conduct background
extraction on video images and transmit only the changing
foreground to minimize data transmission. As we use FHE,
this allows reconstruction of the video at the server without
decrypting the data. We tested the feasibility of the proposed
approach extensively under various conditions including the type
of FHE used, the underlying communication protocols and video
size. The results indicate that our approach can even outperform
AES-based method in terms of total time to complete the video
transmission while additionally enabling privacy features.
