ANALYSIS OF ANOMALIES DETECTION APPROACHES IN DECENTRALIZED COMPUTER NETWORKS

Abstract

In the article a review of contemporary approaches to anomaly detection in decentralized peer-to-peer (P2P) networks is presented. The challenges posed by the architectural features of P2P systems are examined, including the absence of centralized control, the diverse behavioral characteristics of nodes, and the necessity to preserve user privacy. The study focuses on analyzing methods used for monitoring network activity and identifying potential threats, such as denial-of-service attacks, botnets, unauthorized access, and malware distribution. Traditional approaches are discussed, including rule-based methods, statistical models, machine learning algorithms, graph theory-based techniques, and blockchain-based methods. The advantages and limitations of each approach, as well as their effectiveness in various scenarios, are highlighted. Rule-based methods are noted for their efficiency in detecting known attacks but lack adaptability to unknown threats. In contrast, ML algorithms can uncover hidden patterns but require substantial computational resources. Graph theory-based methods facilitate the analysis of network structures to detect anomalies, though their implementation in large-scale networks remains challenging. The article also provides examples of successful implementations, such as the decentralized monitoring system GNUnet, and evaluates their strengths and weaknesses. Issues related to scalability, detection accuracy, and adaptability to dynamic network conditions are discussed in detail. The relevance of this research is driven by the growing popularity of P2P technologies and the increasing need to enhance their security. Summarizing the knowledge in this domain will contribute to the development of new methodologies and technological solutions for anomaly detection, ensuring the resilience and reliability of decentralized networks in the modern cyberspace.