contact    site map  
    polski   english

Computer Networks

The team involved in scientific issues related to computer networks consists of 4 people:
Research is conducted in the following key areas:
Networking research area.
Fig. 1. Networking research area.
 

Self-configuration networks


The observation of the network behavior and the quantitative description of the network condition is one of the most important activities for network administrators and analysts. Networks are dynamically changing over time, while their structure stays fixed. The traffic or device load can rapidly change causing networking failures, inaccessibilities or delays due to nodes overload. The main aim of the project is the improvement of networking performance by the equalization of resources load and therefore the prevention of the networking failures due to the nodes overstrain.

The project is focused on the active monitoring of network nodes selected parameters (CPU, MEM, interfaces load, device temperature, etc.) using the general common standards (SNMP, NetFlow, TCL / bash scripts, etc.) supported by most hardware manufacturers. The work concerns a design of a decision system based on predefined actions invoking network reconfiguration when observed parameters exceed acceptable limits. The automatic network monitoring mechanisms give the administrator up-to-date quantitative information about the network performance and, coupled with the appropriate decision system, create the powerful system of the network guard improving its efficiency.

  

Fig. 2. Self-configuration networks mechanism concept diagram.

Wojciechowski R., Sierszeń A., Sturgulewski Ł.: Self-configuration networks, Image Processing and Communications Challenges 7, Advances in Intelligent Systems and Computing, vol. 289, pp. 301 - 308, Springer International Publishing, ISBN 978-3-319-23813-5, 2015.

Network behavior analysis system

The variety of the network traffic caused by the wide scope of devices, applications and user needs cause the anomaly traffic hard to detect, making potentially network threat. Most of NIDS (Network Intrusion Detection System) applications are based on the static approach (signatures, thresholds), but the best adaptation to the real networking can be obtained by the application of non-static approach. In that case, the classificator learns on the real-time / reference grabbed traffic. Classification uses advanced algorithms, i.a. the pattern recognition based on distance functions and neural networks, and the decision systems verify each flow using the previously learned classification rules and pass the decision to the execution units or the information to the monitoring applications.

The research concerns the elaboration of the new pattern recognition algorithms for classification, reduction, and selection of traffic properties, the migration to the GPU platforms to significantly improve the efficiency and the development of the decentralized system with the possibility of connecting analysing modules for selected network protocols and services in selected segments of a network. The network traffic filtering using the self-learning algorithms allows better match of decision rules to the reference traffic and unknown events recognition. The research is focused on the classifier learning process implementation, the filtering execution stays in the process in the same form. The target platforms for the project implementation are commercial and open-source systems.

Fig. 3. The networking statistical analysis concept diagram.

Sierszeń A., Sturgulewski Ł.: Diffuse network behavior analysis systems with the use of pattern recognition methods, Information Systems Architecture and Technology – Networks and Networks’ Services, Wrocław, Polska 2010, pp. 223-234 ISBN 978-83-7493-542-5.
 
Sierszeń A., Sturgulewski Ł.: Network Behavior-Analysis systems with the use of learning set and decision rules based on distance, Automatyka, AGH, Kraków, Polska, 2010 ISSN: 1429-3447.

Network simulations

The simulation is a helpful tool for the engineers to learn the system behavior where the system activity is unpredictible or creating a system is temporary impossible (i.a. economical reasons). There is a wide scope of simulation goals: the verification of the correctness of the existing network configurations, the protocols / networking mechanisms engineering, the modification of networking behavior to the vulnerable tests requirements, etc.

We use the following environments:

  • Graphical Network Simulator - GNS3,
  • Cisco Packet Tracer,
  • Omnet++,
  • OPNET IT Guru Academic Edition.

GNS3 is a software tool for creating virtual networks based on the real OS images. It allows for testing the hardware and software configurations before the implementation in the production environment. It is possible to integrate virtual networks with the real existing ones.

Cisco Packet Tracer is an innovative network configuration simulation tool used to demonstrate complex technical concepts and networking systems behavior. Cisco Packet Tracer is an integral part of the Cisco Networking Academy enabling simulation, visualization and assessment in learning of complex networking technologies.

OMNeT++ is an extensible, modular, component-based C++ simulation framework, primarily for simulating network behavior. The OMNet++ environment is used in the ad-hoc wireless networks communication research.

OPNET IT Guru Academic Edition software is used for modeling real cooperation of different networking solutions available in the market.

    

Fig. 4. Example simulation project in GNS3 environment.

Fig. 5. Example simulation project in Cisco Packet Tracer software.
 
Sierszeń A., Sturgulewski Ł.: Implementation of IPS&IDS technology indevices emulated with GNS3 environment, Information Systems Architecture and Technology – Networks and Networks’ Services, Wrocław, Polska, 2010, pp. 247-258, ISBN978-83-7493-542-5.
 
Sierszeń A., Sturgulewski Ł., Janicka J.: The aspect of testing software for monitoring and managing networks in GNS3 environment, Information Systems Architecture and Technology – Services Oriented Distributed System: Concepts and Infrastructure, Wrocław, Polska, 2009, pp. 293-302, ISBN978-83-7493-477-0.

Sierszeń A., Sturgulewski Ł.: Wykorzystanie Cisco Packet Tracer do nauczania podstawowych zagadnień transmisji danych głosowych w sieciach komputerowych, Zeszyty Naukowe Politechniki Łódzkiej, nr 1078, seria Elektryka, z. 121, Polska, 2010, ISSN: 0374-4817.

Sierszeń A., Sturgulewski Ł.: Symulacja komunikacji w sieciach bezprzewodowych typu ad-hoc w środowisku OMNeT++, Zeszyty Naukowe Politechniki Łódzkiej, nr 1078, seria Elektryka, z. 121, Polska, 2010, ISSN: 0374-4817.

Sierszeń A., Sturgulewski Ł.: OPNET - uniwersyteckie spojrzenie na projektowanie sieci; Informatyka i Edukacja, Monografie Politechniki Łódzkiej, Nr 1949, Polska, 2010, str. 155-184, ISBN978-83-7283-376-1.

return to the top


Editors:
Artur Sierszeń
Rafał Wojciechowski

Last modification:
2015-11-06 18:14:47, Rafał Wojciechowski