Have you ever wondered how faults are located? It’s a procedure similar to a doctor’s: looking at symptoms, doing tests, and making a diagnosis to find the right treatment.
In this case, the patient is distributed and his organs are electronic systems. It also grows and transforms at great speed. Performing the diagnosis in an automatic way is a technical challenge for which Telefónica I + D has been innovating with two artificial intelligence techniques for five years.
The first of them uses the so-called multi-agents. An agent is a program capable of emulating complex behaviors and coordinating with others of the same nature. Distributed throughout the network, each of them is able to look for symptoms in their environment and share that information with others, until you get a complete picture of the state of the “patient”.
But having the symptoms is not enough. The most important thing is missing: the doctor who treasures the knowledge and experience to know what is the common origin of all. In communications, this role is necessarily played by highly qualified engineers. Is it possible to emulate their behavior? Yes, using Bayesian networks, a technique that appeared in the ‘ 80s. A Bayesian network is a algorithm that allows to discover the cause of the failure in view of the symptoms, with a certain percentage of certainty.
This also works similarly to what happens in medicine. The doctor may conclude that the patient is most likely to suffer from a certain condition, but if he is unsure he will prescribe more specialized tests to confirm it or to seek another explanation. This iterative procedure can be performed automatically by combining multiagents and Bayesian networks. This is one of the fundamentals of KOWLAN, a Telefónica system that was among the most innovative and efficient.
In a multi-agent study carried out by the team of Dr. Jörg Müller at Clausthal University (Germany), the KOWLAN diagnostic system developed has been rated as the second best application in terms of industrial use, out of 152 studied. In addition, the telecommunications sector is the most advanced in the use of this technology and Telefónica I+D is one of the most innovative organizations.
This result has been possible thanks to Telefónica’s investment and commitment to this type of innovation and the collaboration of the Polytechnic University of Madrid, the Complutense University and our partners in Telecom Italia laboratories.
