NAME Artificial intelligence and cybersecurity laboratory
The cybersecurity laboratory offers solutions based on data analytics. The tools are devoted to tasks such as intrusion detection and, anomaly detection and predictive analysis among others. Our team is composed of statisticians, computer scientists and mathematicians which have experience both at academia and industry in designing and implementing machine learning tools, statistics for large data volumes, information extraction and prediction.
FIELDS OF APPLICATION
Asset protection
Attack detection
Attack response
MOST OUTSTANDING EQUIPMENT AND COMPONENTS
-
Computational cluster
The high performance computational cluster offers high computational capacity for the experimental evaluations carried out. It has parallel computation and storage nodes of high capacity and performance and InfiniBand connection between nodes for minimum latency.
Our cluster has 8 Intel Xeon processor nodes with the following specifications:
Two dense nodes with the following specs
InfiniBand network and a 1Gb/s administration network.
4 Intel Xeon Gold processors (72 cores)
192 GB RAM
2 local hard drives of 1TB or more and configures in RAID1
InfiniBand connectivity, 56Gb/s
2 Ethernet ports, 1Gb/s (Gigabit Ethernet)
1 RJ45 port for IMM2 management
Redundant power supply
Compatible with Linux CentOS 7.2.1511
2U sizeSix nodes with the following specs:
InfiniBand network and a 1Gb/s administration network.
2 Intel Xeon E5-2683 processors V4 (32 cores)
256GB RAM
2 local hard drives of 1TB or more and configures in RAID1
InfiniBand connectivity, 56Gb/s
2 Ethernet ports, 1Gb/s (Gigabit Ethernet)
1 RJ45 port for IMM2 management
Redundant power supply
Compatible with Linux CentOS 7.2.1511
Compatible with Lenovo NeXtScale n1200 chassisThe job scheduler is the Slurm workload manager, which allows prioritising and the reservation of different resources. This job is the key to reserving resources (memory, CPU and storage) for the BCSC members or related projects.
The software in the cluster has been configured and optimized for computationally expensive operations. The installed software includes Intel and GNU Compilers, Parallel Matlab, GROMACS, FreeSurfer, R and many libraries for parallel and scientific computing.
SERVICES OFFERED BY THE ASSET
Data analysis for cybersecurity
Development of algorithms to detect intrusions in computer or industrial networks, based on the development on demand of prediction and time series forecasting systems.
Development of optimisation and Big Data applications robust to attacks
Big Data algorithms have become a great aid to decision making. In many cases, very significant decisions are made almost entirely based on the calculations of these algorithms. Big Data technologies are fed with data that, in the industry, are frequently generated by sensors. A malicious user could take control of a data source and modify it so that the Big Data algorithms make wrong decisions based on the manipulated data. When it comes to the development of Big Data systems in key components, it is necessary to implement techniques that can detect such manipulations in order to act accordingly.
ENTITY MANAGING THE ASSET
Contact person:
Miguel Aguilera
maguilera@bcamath.org