Controller placement problem in the optimization of 5G based SDN and NFV architecture
The fast rise in data traffic and the vast range of services and applications accessible in 5G networks must be addressed effectively. Integrating Software Defined Networking (SDN) with Network Function Virtualization (NFV) is a low-cost way to build a reconfigurable network, reduce operating cos...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2021
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/104026/1/Abeer%20Ibrahim%20B5%20-%20IR.pdf http://psasir.upm.edu.my/id/eprint/104026/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The fast rise in data traffic and the vast range of services and applications accessible in
5G networks must be addressed effectively. Integrating Software Defined Networking
(SDN) with Network Function Virtualization (NFV) is a low-cost way to build a
reconfigurable network, reduce operating costs, and optimize network performance. The
separation of control functionality from forwarding devices brings orchestration and
management to enable 5G network programmability. Although centralized control
facilitated orchestration and administration of 5G services and applications, it could not
handle massive and varied data volumes. 5G networks can avoid performance
degradation, enable diverse network traffic management, and create a flexible and
scalable design by adopting and deploying multi-controllers in the network control layer.
However, for optimum 5G core design and cost-effectiveness, a group of controllers
must be appropriately mapped.
A distributed 5G-SDN-NFV-based network architecture uses the controller placement
problem (CPP) to manage controller placement and number. A heuristic called dynamic
mapping and multi-stage CPP algorithm (DMMCPP) was developed to solve CPP as
resource allocation in a distributed 5G-SDN-NFV-based network. This thesis divides
CPP solutions into three groups based on three objectives: (i) scalability and load
balancing, (ii) reliability and resilience, and (iii) efficient routing for energy-aware
design. First, a dynamic allocation and mapping CPP (DAMCP) is developed to solve
network dynamic resource location problems. It demonstrates a trade-off between
locating a minimum number of controllers and network traffic to maximize resources
and achieve load balancing at minimum costs. Second, the increasing demand for
controllers exposes the network to control planes and connection failures, which are the
most frequent problems in SDN networks. If the control plane fails to improve system
resilience quality, A reliable RAMCP is formulated as an optimal solution for fault
tolerance. Furthermore, the approach is extended with a Particle Swarm Algorithm
(PSO) and presented as a hybrid RASCP to validate the optimal location and number of
controllers. Third, the considered traffic paths across backup nodes and redundancy
lengthen, increasing latency and power consumption in a network. The proposed energyaware
routing algorithm (EARMCP) implements efficient flow routing mechanisms for
network traffic to minimize the number of active links and 5G-DC devices. Extensive
computations utilizing MATLAB 2018a on the Intel Core i7/Gen 10 processor and 16
GB of RAM are used to evaluate the algorithm efficacy.
According to the blueprint of our heuristic method, the allocation and the optimum
number of controllers under an effective decentralized policy could achieve higher
efficiency. The selected control number is picked with a higher efficiency before the
rescheduling is approximately 80 % for optimized controllers up to 90 % of resource
management than other comparable algorithms in such a densified network. In addition,
energy savings of up to 70% are achieved compared to the proposed Dijkstra-based
energy-aware algorithms. |
|---|