An Integrated Approach for Enhancing Internet Security Using Squid Proxy-Based Whitelisting with Machine Learning
摘要
Due to the proliferation of Internet usage, threats like malware, phishing attacks, Denial of Service (DoS), and data breaches are increasing at rapid rate. Thus, to protect data and users from various online malicious activities, necessary robust security measures that amalgamates established and emerging techniques should be adopted. Organizations provide Internet access to users through proxy servers for enforcing policies to block or allow specific content and domains using its content scanning and URL filtering capabilities. Squid proxy server is usually configured with default allow all and block only malicious domains/URL rule, resulting in dependency on latest and updated blacklist of malicious entities. Since the updation of blacklist is a continuous process, new threats cannot be handled effectively. In this paper, we had adopted a whitelisting approach, wherein squid proxy server allows only pre-approved domains and implements a block all rule. This strategy mitigates potential risks associated with unvetted content, leading to heightened security and data protection. Automation scripts are introduced to streamline the process of updating the whitelist with new domains, ensuring real-time adaptability. To enhance the ease of domain management, a web-based graphical user interface (GUI) is introduced. This GUI empowers administrators to oversee whitelisted domains and make modifications in real time, reducing administrative overhead and fostering responsive security control. The whitelist implemented is further checked for maliciousness and divided into different categories using latest machine learning techniques, allowing only categories as per the compliance policy of the organization. Thus, an integrated approach for bolstering Internet security within LAN environments is presented by leveraging the capabilities of squid proxy server using a whitelisting strategy and harnessing machine learning for safe and secure domain categorization.