On the feasibility of using hybrid evolutionary dynamic optimization for optimal monitor selection in dynamic communication networks

In this paper, we propose a technique to optimize dynamic communication network topologies. Due to recent devel­opments in virtualization technologies like hardware virtualization, software defined networking (SDN), and network function virtualization (NFV), a completely new way of automated network optimization becomes possible. We propose a technique to enable hybrid evolutionary dynamic optimization of networking infrastructures based on changes to its topology or state. Networks have numerous, partially competing, attributes, which might be subject to optimization. Here, we focus exemplarily on the problem of finding the optimal amount and position of monitors in a computer network in order to monitor the whole network traffic. This dynamic monitor selection problem is generalizable to the well-known NP-complete vertex cover problem. Experiments are conducted on different real-world networks using an evolutionary search heuristic. We studied the behavior of the proposed approach using 3 different change levels of the problem instances. Experimental results show that our proposed approach provides network configurations having a sufficiently high quality in reasonable time for all problem instances and change levels. Thus, using the proposed technique may be one helpful step towards an automated self-adapting network optimization.