Auto-scaling in Virtualized and Native Network Functions

This page is being updated — please visit again.

My contribution in this project has been named: IRONMAN-BRAIN, which is a modular and intelligent autoscaling system that leverages reinforcement learning to enable proactive and adaptive scaling of Virtualized and Cloud-Native Network Functions. IRONMAN-BRAIN uses a Deep Q-Neural Network to make autoscaling decisions within Service Function Chains (SFCs) deployed across OpenStack and Kubernetes platforms. Integration with Open Source MANO (OSM) facilitates lifecycle management, while SDN components such as OpenDaylight and Open vSwitch provide programmable control of the network infrastructure [1].

Some projects in this field where I have been working on:

Advanced framework for secure end-to-end management of virtualised networks and provision of cybersecurity services as a Service

Deep Q-network: network state inputs, hidden layers and scaling actions sent to the decision model — **Deep Q-Neural network** implemented for proactive and adaptive scaling of virtualized and cloud-native network functions [1]

Artificial neuron: weighted inputs and a bias are summed and passed through a ReLU activation function — Artificial neural unit with ReLU activation function implemented in the Deep Q-Neural network

References

Advanced framework for secure end-to-end management of virtualised networks and provision of cybersecurity services as a Service, project funded by IRONTEC, Internet y Sistemas sobre GNU/Linux, Regional.