Distributed Deep LearningIt is a distributed algorithm that uses Unsupervised Deep Learning for the efficient computation of the SFC-E problem.The Deep Neural Network of the algorithm is trained by a Genetic Algorithm. A distributed greedy algorithm is also included for comparison with the DL algorithm. source code in eclipse project, jar file run: "jave -jar SFCdistrDL.jar", documentation. |
Genetic AlgorithmThis is a Genetic Algorithm that embodies lots of innovations and provides near optimal solutionsfor the SFC-E problem combining effectiveness an efficiency. source code in eclipse project, jar file run: "jave -jar SFCgaMT.jar", documentation. Baseline greedy algorithmA simple greedy algorithm mostly used for comparison with the Genetic Algorithm.source code in eclipse project, jar file run: "jave -jar SFCbaseline.jar", documentation. |
Network SimulatorIt is an advanced network simulator integrated in every algorithm, designed for SFC embedding simulations.Currently it supports 3-tier fat-tree topologies that are comonly used in modern datacenters. The simulator is written in Java 8 and it is only necessary to add the project or the .jar file in the classpath of any project that uses it; like in the algorithms presented here. The output is a log file describing the status of the network after each request, accepted or not. Use the -Xmx4g or the -Xmx6g option in case of large networks; larger than 5000 nodes. fields description , documentation. |
VNF-graphsA small collection of VNF-graphs in Edge Vector format suitable to be used as requests in embedding algorithms.The directory called "EVgraphs" should be in the same level as the directory that contains the algorithm. |