Page 13 - Unlocking the potential of trust-based AI for city science and smarter cities
P. 13
Figure 3. High level architecture of the Trust based Data Governance.
The interactions related to several distributed data subjects, data controllers and data processors
in a smart city context can be handled by a Trust Manager from a security perspective. Different
systems in a smart city may have different local policies for security. In general, there is a staggering
number of resources and services to be accessed in a smart city. Trust manager receives these
access requests together with a set of credentials and determines if the provided credentials for
access request comply with the local security policy to access the intended resource or service
(in this case it can be data that entails personally identifiable information). Hence, it uses a
general-purpose application-independent algorithm and supports features like delegation, policy
specification, refinement at the different layers of a policy hierarchy. So, the Trust Manager solves
the consistency and scalability problems present in traditional mechanisms.
Recent technological innovations of smart edge devices and services which heavily rely on real-
time data processing and localized intelligent decision-making, have created a vacuum for a
novel approach that extends the traditional means of research in cloud computing towards edge
computing. The idea of edge computing refers to fluid data management and decision-making
towards physical things, working as a middle layer between cloud and the users.
Major advantages of doing so include but not limited to (a) minimizing response delay by addressing
the bottom level request at the network edge instead of servicing it at far located cloud data
centers, (b) minimize downward and upward traffic volumes in the network core and (c) maximizing
the support for cross-border applications due to effective resource and security management at
cloud. Complying with edge computing requirements, the proposed approach further breaks down
the so-called middle layer by introducing two layers ROOF computing and Fog Computing which
places just below the cloud as shown in Figure 4 in order to make the system architecture more
feasible and deployable in real-time environment with an ambitious vision for seamless fluid control
and decision-making through harmonize resource management among different layers.
Fog computing layer is implemented with the idea of achieving the second objective of the edge
computing scenario which is improving application performance and resource efficiency by
removing the need for processing all the information in the cloud, thus also reducing bandwidth
consumption in the network. A Fog node can be defined in several ways. It can be regarded as an
Unlocking the potential of trust-based AI for city science and smarter cities - October 2019 7