Detecção de ataque DDoS através da lógica paraconsistente anotada

Abstract

Computational development has innovated the communications network, which is why the computer network was created. The Internet (global computer network) is the largest computer network, in which it allows the exchange of messages in real time. With the increase of devices connected to the Internet, in which most of the time they do not have the essential security implementation inside, which generates vulnerabilities susceptible to invasion to form a botnet in order to carry out a Distributed Denial Attack Service (DDoS). DDoS attacks are divided into categories. Among the categories are flood attacks, which causes the service to stop when network / transport resources or server resources are exhausted. There are several types of DDoS attacks, among them the SYN (Synchronize) flood attack. The freedom in scientific research to seek multidisciplinary methods and / or resources to solve problems is one of the roots and virtues applicable to science. Thus, Paraconsistent Logic (LP) was one of the techniques that aroused interest in the study of attack detection, depending on its characteristic and application. Therefore, this work aims to verify the viability of using Paraconsistent Logic, specifically Annotated Paraconsistent Logic (LPA), in security in computer networks for the detection of attacks, with a focus on SYN flood-type DDoS attacks. Faced with this challenge, this paper presents the implementation of LPAProg-DDoS, in order to contribute to the areas of Computer Networks and in particular Computer Network Security for decision-making. LPAProg-DDoS captures network data or opens a network capture file, then pre-processes the network data, calculates degrees of belief and disbelief,the implementation of the FuzProg-DDoS system is presented, in which it differs from LPAProg-DDoS in the computational step, with FuzProg-DDoS making use of Fuzzy Logic. The tests are carried out in the Bot IoT database and in a real scenario created in UTFPR. The LPAProg-DDoS and FuzProg-DDoS systems performed satisfactorily in the Bot IoT database for all evaluation metrics. In the real world, both systems are capable of detecting SYN flood type DDoS attacks with an overall accuracy greater than 80%, but in the tests carried out, FuzProg-DDoS obtained better assertiveness. Thus, according to the results obtained, in a stage considered still preliminary, it was found that the LPA has a lot of potential for practical application in network security.