|Project title||Resilient controller and hypervisor placement for future communication networks|
|Funding agency||Unity through Knowledge Fund (UKF)|
|Grant type||1C My First Collaboration|
|Project beneficiary||University of Zagreb, Faculty of Electrical Engineering and Computing|
|Project leader||Assistant Professor Ognjen Dobrijević|
|Project partner||KTH Royal Institute of Technology, School of Electrical Engineering and Computer Science|
|Project co-leader||Marija Furdek Prekratić, Docent|
|Duration||16/10/2017 - 15/01/2019 (15 months)|
|UKF funding||300,000.00 HRK|
|Matching funding||168,750.00 HRK|
Communication networks are evolving towards their fifth generation (5G) driven by the need to support the massive growth of traffic demands and the stringent requirements on service availability, communication latency, and cost. To satisfy these demands, 5G solutions will rely on the software defined networking (SDN) and network virtualization paradigms that are expected to increase network agility and flexibility, while reducing the capital and operational expenditures. In SDN networks, the network control intelligence is separated from the traffic forwarding devices and located in SDN controllers, which form the control plane. On the other hand, network hypervisors facilitate virtualization by abstracting the computational and transport resources of physical network devices into isolated virtual networks, which can be tailored to the specific needs of end-users and applications. Guaranteeing resilience of the SDN control plane to network failures is a fundamental prerequisite for accurate operation of such networks and all the critical services they support.
The goal of this project is to develop and demonstrate approaches for the design of a resilient control plane in virtualized SDN-based networks that is distributed across multiple SDN controllers and network hypervisors. In the first phase of this project, we will develop a framework for modelling control plane availability by evaluating the criticality of individual components in the presence of different failure types. A new availability metric will be defined, applicable to the advanced communication scenarios of the distributed control plane. In the second phase, we will model and mathematically formulate different variants of the resilient controller and hypervisor placement problem and develop optimization algorithms to solve them. Solving this placement problem entails deciding on the number and the network locations of SDN controllers and hypervisors, and the paths of control channels between them, while reserving sufficient working and backup resources to ensure correct control plane operation in the presence of failures. In the third phase, the efficiency of the designed approaches will be experimentally validated in a laboratory prototype of 5G network control plane with state-of-the-art equipment.
As existing approaches for control plane resiliency do not apply to virtualized SDN scenarios, the newly defined availability models, the developed optimization approaches, as well as the simulation and experimental results obtained during this project will be an important contribution towards a unified framework for resilient 5G networking solutions.