Image Processing Reference
In-Depth Information
Fig. 12.1 Overlay networking
In order to cover both static and dynamic content, we consider two functionally
separated overlays: (a) the
Distributed Media Overlay,
focusing on caching and
delivering of static content and (b) the
Real Time Interactive Communication
Overlay
, catering for the exchange of dynamic content and supporting signaling
and monitoring for sessions, 3D meshes and point clouds, along with real-time
video and audio.
Knowledge of the underlying network infrastructure and conditions (capacity,
traffic, delay, load, network bottlenecks) as well as the network requirements of the
3D scene may significantly change the construction of the overlays and, conse-
quently, the performance of the application and the Perceived Quality of Experi-
ence (PQoE). Moreover, the operational cost caused by the applications may be an
issue for an operator, depending on the type of application. For instance, P2P file-
sharing applications cause a considerable load on an operator
s network and this
happens over a long duration. Applications cannot usually judge the costs caused by
an action, such as loading data from one or another location, as these costs are
transparent to the applications.
In order to improve and even control the delivery of the content, the 3D
immersive communications platform should be aware of the network capabilities,
conditions and infrastructure in order to select the most appropriate participating
nodes (i.e. routers, rendezvous points, ALM nodes, caches, application server(s),
and clients) and build the most efficient overlay topology. This is a challenging
objective from multiple points of view [
4
]. However, path selection based on
background monitoring of alternative paths in an overlay of N nodes requires
constant monitoring of
O
(
N
2
) paths to choose the best detour at any given time
[
5
]; as a result the efficiency and adoptability of this approach quickly diminishes
with increasing number of nodes [
6
]. Moreover, having full knowledge of the
network topology is out of scope of a tele-immersive platform. Yet, the smart
selection of the overlay nodes may (a) increase the quality of the data path
'
