Page 78 - 5G Basics - Core Network Aspects
P. 78
1 Core network aspects
NDOs in DAN are typically conventional data, such as videos, text files and measurement data. At the same
time, an NDO can also represent a data-processing function. When a need arises to clarify whether an NDO
represents conventional data or a data-processing function, it is specified as NDO-data or NDO-func,
respectively. An end user can request a data-processing function by its name without knowing its location [b-
Tschudin, 2014].
This service scenario takes into account that DAN elements can: 1) execute data-processing functions;
2) move data-processing functions autonomously or manually; and 3) cache not only data, but also data-
processing functions. Thus, DAN elements close to an end user can cache data and data-processing functions,
as well as directly responding to various user requests. In this manner, traffic load in the core network and
the response time for user requests can be reduced. Nevertheless, the deployment of the data-processing
function should be decided carefully considering various functional allocations to optimize the system
[b-Katoh, 2015].
Typical implementations of this service scenario include video analysis and the Internet of things (IoT). For
instance, when a large quantity of video data from a camera is requested for analysis, e.g., to detect lost
children or to monitor unattended properties, the analysis request may be carried out in the DAN element
close to the video data source rather than being forwarded to a remote server. This strategy is adopted
because the transmission time for a large video file to a remote server through a WAN would be the dominant
element in total delay, although data-processing time is also usually relatively large. In addition, if the remote
server is congested with many other tasks, the analysis request may be carried out in the DAN element close
to the requester rather than on the remote server, because the load on the DAN element might be
comparatively low. In these cases, the transmission times of request and response messages can be reduced
effectively within the total delay or data-processing resources can be used more efficiently due to distributed
processing over ubiquitous DAN elements.
Regarding NDOs in DAN caches, there are two cases. The first is when the DAN element has the result of a
data-processing function for an analysis in its cache, which directly returns the data to the requester. The
second is when the DAN element has a function in its cache that processes input data with the function and
returns the result to the requester. Otherwise, the analysis request is forwarded to another nearby DAN
element that holds the function.
Figure 7-15 describes a sample scenario for in-network data-processing service with DAN.
Figure 7-15 – In-network data processing with DAN
68