Research HighLights

FP7 INTEGRIS

La Salle R&D has finished all the tests of the PHY_TRILL (Phyton) implemented for INTEGRIS FP/ European project. We can conclude that the results have come out mostly as expected. If we take into account that SSH tunnels take some time to be reestablished by the operative system, our results are within the margins expected.

We have also tested TRILL with aggressive timers and it works as expected, but it also can create instability and also increases a lot the traffic overhead caused by TRILL making this configuration not recommended for production use. The purpose of monitoring IDEVs (INTEGRIS Device) is to provide information about them and know the state of the system. It is important to know the real state of the network or every IDEV to prevent any problem. When some problems appear, is useful to know which part of the system fails and find quickly where the problem is. For these reasons, is important to monitor devices of the network, because it will help us to save our time. Our task was also to design a system that extracts different parameters of IDEV and send these parameters to I-NMS. I-NMS is a device that monitors the system and it asks periodically the state of each IDEV.

Besides, La Salle R&D has performed all the different tests proposed for the INTEGRIS cognitive system. There were no noticeable problems while doing the tests. One important thing we had taken into account while performing the tests (especially on Field Trial Test) was the availability of the IDEVs. It was very important to avoid losing the connectivity with the IDEVs because it would imply one week of unavailability because the physical access to the IDEVs must be scheduled in advance. Because of this, some parts of the security tests were not tested in the field (SSH service error).

The system gave the expected output in all the tests, but we think that the system can, and should, be improved in order to generate a more accurate output. Probably, we need to train the system with real network data and not only with synthetic cases. Using more parameters would also be interesting in order to create a better and more specific system.