Page 62 - Proceedings of the 2018 ITU Kaleidoscope

2018 ITU Kaleidoscope Academic Conference

Windows, Linux, etc. They possess the capability to not 4. AIMS: USE CASES
only aggregate data but also to execute pre-processing on
the collected data. Some of these devices can also run some 4.1 Smart City Surveillance Application
embedded AI algorithms as microservices to provide simple
intelligent decision and insights on the data they have One of the key areas where the architecture proposed in this
aggregated. article is most useful is in security surveillance in a Smart
City platform. In a Smart City, there are numerous smart
The ROOF layer consists of devices and nodes such as 5G cameras installed in various parts of the city for different
gNodeBs, home routers, smartphones that provide the purposes ranging from traffic monitoring, security
resources for always-available services, security, privacy in surveillance at train stations, bus stations, airports,
real-time as the next hop for the Things. It can be shopping malls, streets, etc. Imagine that there is
implemented on these devices that serve as Things’ proxies intelligence about an intending terror attacks and the
for connectivity to the network and Cloud. In our proposed pictures of possible suspects have been shared among
architecture, ROOF serves as a proxy for the physical various security monitoring systems in the city. The
Things for connectivity to the Fog and to the Cloud security monitoring system is linked with the smart cameras.
Computing data centers. At the layer of the architecture, AI To report the sighting of a suspect, the smart cameras
agents and other related distributed applications can be should be empowered to carry our real-time analysis of live
deployed as microservices. streams of video data and decide if an individual with a
suspicious bag is one of the wanted terror suspects. To
The third layer is the Fog layer, which is a virtualized layer realize that, different analytical AI algorithms, such as
providing compute, storage and networking services anomaly detection using deep learning, can be deployed as
between the ROOF and the traditional Cloud data centers. microservices to support the surveillance cameras installed
This layer can deliver more powerful 5G application in the 5G based virtualized service infrastructure.
services that can be supported by the ROOF layer. This
layer consists of Fog nodes, which are facilities and AI algorithm analyzes the video data for autonomous local
infrastructure that can provide resources for distributed 5G decision-making. The smart camera can then communicate
application services. In our architecture, base stations and its decision to the appropriate authority for action while the
other core network gateways serve as Fog nodes. cameras keep on monitoring the suspect and if need be
passing control information to nearby cameras should the
The fourth layer is the Cloud layer, which is located in the suspect move away from the current camera. Thus, the
core network and support interoperability and wide-usage system can locally process the streams of live video data
as AIMS modules independent to the data. In addition, it among themselves and thus to reducing traffic overhead,
provides long-term decision making in the smart city latency in 5G networks.
services.

Figure 3 – 5G based Virtualized Service Infrastructure

– 46 –