Open AccessIP Packaging Filtering in Computer Networks Using Artificial Intelligence in the Regulatory Authority of Electronic and Communications Kosovo
Author(s) -
Kastriot Dërmaku,
AUTHOR_ID,
Liridon Hoti,
Selami Klaiqi,
Hifoblina Dermaku
Publication year - 2022
Publication title -
journal of communications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.185
H-Index - 35
eISSN - 2374-4367
pISSN - 1796-2021
DOI - 10.12720/jcm.17.2.134-142
Subject(s) - computer science , firewall (physics) , artificial intelligence , the internet , artificial neural network , router , matlab , machine learning , computer network , world wide web , operating system , entropy (arrow of time) , physics , extremal black hole , quantum mechanics , charged black hole
The security of our data is one of the most important challenges, and it is one of the things we must be careful about what we display on the Internet. For the security of this data, computer and cyber networking experts use different methods to prevent these attacks, prevention is done by Firewall techniques, and then by Router where both of these cases are quite popular nowadays today, however, another innovation is that filtering can also be done through Artificial Intelligence. In this paper, research has been done on the application of Artificial Intelligence methods such as those of deep learning on IP packages through Artificial Intelligence where specifically the method of Support Vector Machine with the training of a model through a simulated input corpus, using MATLAB software. From the results of the research by Artificial Intelligence, the source address, destination address, port number, protocol and package size were used as filters. All the practical work was done in over 350 different IP addresses, where 175 IPs were used for algorithm training, as well as 175 IPs were used for testing by combining them with filtering attributes, where in the end the filtering accuracy gained is quite high (83%).