Image Processing Reference
In-Depth Information
example, the UE1 (see Fig.
8.1
), previously attached to ER1, may change to the
ER2 for better performance at lower energy cost. As a consequence, the path
ER3,
<
C1, C2, ER1
must be
established to properly connect ER3 and ER2, depending on the route provided by
the routing protocol in use. Hence, while the cooperative approach promises energy
saving in the wireless environment, lack of scalable QoS control solutions can
easily increase the energy consumption in the core network. In order to address
these challenges, resource over-reservation, consisting of reserving more resources
than a CoS may need, must be explored, so several service requests can be
processed without instant signaling as long as the reservation surplus is sufficient
to accommodate new requests [
12
,
15
,
17
]. However, as we explained earlier in
Sect.
8.1
, the approach imposes a trade-off between signaling overhead reduction
and waste of resources as well as QoS violations [
18
]. ITU-T G.1081 [
83
] suggests
five monitoring points in networks, allowing service providers for monitoring
networks and services performance to improve resource utilization. However,
existing proposals are mostly base on path probing techniques [
84
] and show
limitations in terms of complexity, accuracy and undue signaling [
85
].
Pan et al. [
87
] proposed to over-reserve bandwidth surplus as a multiple of a
fixed integer quantity for aggregate flows destined to a certain domain—a Sink-
Tree-Based Aggregation Protocol. This solution also does not comply well with
network dynamics and fails to efficiently utilize the network resources. The work in
[
88
] over-provisions virtual trunks of aggregate flows based on a predictive algo-
rithm (e.g., past history) without an appropriate mechanism to dynamically control
the residual bandwidth between various trunks. Resource over-reservation is also
studied in [
89
] in an attempt to reduce control states and the signaling overhead.
Sofia et al. [
90
] proposed the use of resource over-reservation to reduce excessive
QoS signaling load of the Shared-segment Inter-domain Control Aggregation
Protocol (SICAP). However, these proposals lead to undesired waste of bandwidth.
The Multi-user Aggregated Resource Allocation (MARA) [
17
] proposed functions
to dynamically control bandwidth over-reservation for CoSs to improve system
scalability. However, MARA also shows serious limitations in its resource distri-
bution capability.
In this scope, recent findings proposed new ways for scalable, reliable, cost- and
energy-efficient control design of IP-based network architectures and protocols
[
23
]. In particular, the Self-Organizing Multiple Edge Nodes (SOMEN) [
79
]
enables multiple distributed network control decision points to exploit network
paths correlation patterns and traffic information in the paths (obtained at the
network ingresses and egresses) in such a way as to learn network topology and
the related links resource statistics in real-time without signaling the paths. As such,
SOMEN provides a generic network monitoring mechanism. The Advanced Class-
based resource Over-Reservation (ACOR) [
19
] effectively demonstrated the break-
through that it is possible to design IP-based networking solutions with significantly
reduced control signaling overhead without wasting resources or violating the
contracted quality in the Internet. The Extended-ACOR (E-ACOR) [
81
] advances
the ACOR
must be released and a new one
ER3, C1, C3, ER2
>
<
>
s solution through proposals of multi-layer aggregation of resource
'
