IP Network Management from the Perspective of Distributed Artificial Intelligence: State of the Art

An approach to study Distributed artificial intelligence is through software agent-based intelligent systems. Instead of having a large centralized system that collects complete system information, there is a small number of subsystems or agents, which are focused on the effort to cooperate to solve a given problem. Agents can communicate with each other, exchange views and find various strategies to obtain a specific solution. Due to the exponential growth of network devices and the massive use of broadband networks, current centralized management systems are reaching limits in terms of managing the amount of information to be managed, which is why distributed management based on the paradigm of intelligent agents can be applied in this domain. In the literature, there are several experimental and academic prototypes that attempt to demonstrate how the use of distributed artificial intelligence can be an interesting alternative to the classic centralized network management systems. This article benchmarks these prototypes, emphasizing two aspects (i) the strategy and (ii) the inherent functional blocks of each proposal.