Image Processing Reference
In-Depth Information
potential network bottlenecks and the delay jitter.
Active probing
or passive
monitoring techniques may be applied to mine the network characteristics.
These metrics may have quite different result, based on the type and charac-
teristics of content that needs to be delivered [
10
].
(b)
In-network Nodes Characteristics.
The nodes (both network nodes and client
nodes participating in the immersive communication) are not equivalent,
either due to their different characteristics (e.g. CPU capacity, storing capac-
ity/memory, buffering queues, network interfaces) or due to their role in the
network. As a result, the combination of network
'
s characteristics and nodes
'
characteristics defines the technical metrics, which could be measured. As the
client nodes are given, we may only select alternative in-network nodes that
have less load or higher capabilities. From these two categories, we can also
extract information about potential network bottlenecks.
(c)
Network Operator recommendations.
Besides performance and technical
characteristics, there may be a set of other parameters and recommendations,
provided directly or indirectly by the network operator/ISP that needs to be
taken into account. For example, by utilizing information from an IETF ALTO
server [
7
], information such as the ISP communication cost, inter-ISP con-
tracts/agreements and network related policies may be taken into account.
(d)
Social Interaction.
As already stated, besides pure network and nodes char-
acteristics, the network overlays may be set-up taking into consideration the
social dimension of the social interaction of the 3D immersive communication
users (e.g. [
9
]); i.e. based on the frequency and duration that specific users
participate in a common tele-immersion session, we may modify clients
point
'
of connection.
In the next section, we focus on the social interaction. It should be emphasized
that the purpose is not to find the best mathematical equation in order to model the
various measurements and probability, but to compromise between valid results and
fast calculations, with low computational complexity.
12.4 Social Network Analytics for Overlay Construction
Optimization
In many network overlay optimization methodologies and algorithms, the overlay
is considered as a graph, where nodes are the elements positioned according to their
location, while links represent communication paths through intra-net or internet
layouts, being quantified by weight values (e.g. capability primitives and commu-
nication cost). One of the key issues of social networking is that social interaction
can be actually recorded and processed using big data analysis frameworks in order
to remap the user population on a social network canvas. Hence, analytics from the
social network can produce behavioural semantic relational graphs and clustering at
the user level. By exploring the social interactions of users, we could unveil their
