The global internet operates on a "trust-by-default" architecture. When a network router claims it knows the best path to a specific set of IP addresses, other routers generally believe it. When a Content Delivery Network (CDN) is told that a user owns a specific domain, it often takes them at their word. This inherent lack of built-in authentication in the Internet's foundational protocols—specifically the Border Gateway Protocol (BGP) and the Domain Name System (DNS)—creates a massive security gap that researchers are now working to close through advanced monitoring, data-plane analysis, and verifiable commitments.
The Vulnerability of Implicit Trust
The core of the problem lies in the Border Gateway Protocol (BGP), the "glue" that connects thousands of independent Autonomous Systems (ASes). As noted in Logical peering for interdomain networking on testbeds by Yuanjun Yao and colleagues, BGP lacks inherent security mechanisms. This allows for prefix hijacking, where a malicious or misconfigured AS falsely claims ownership of an IP range.
When this happens, traffic intended for a legitimate destination is diverted. This isn't just a theoretical risk; it leads to large-scale traffic interception and outages. Early attempts to solve this, such as the routing registry language defined in Representation of IP Routing Policies in a Routing Registry (ripe-81++) by Tony Bates and Daniel Karrenberg, provided a way to document policy but lacked the robust authentication needed to prevent unauthorized announcements.
Detecting Anomalies in the Control and Data Planes
To combat hijacking, researchers have developed two primary detection strategies: monitoring the "control plane" (the routing messages themselves) and the "data plane" (the actual traffic flow).
Control Plane Monitoring
Traditional detection involves scanning global routing tables for suspicious changes. Soon-Tee Teoh and Kwan‐Liu Ma pioneered Visual-Based Anomaly Detection, using 2D and 3D patterns to help human operators spot "hole-punching" attacks that statistical models might miss. More recently, Ruoyu Wang and Qi Li have proposed using Large Language Models (LLMs) to understand the "behavioral semantics" of BGP, moving beyond rigid maps to a more nuanced understanding of network intent.
Data Plane Validation
The limitation of control plane monitoring is that an attacker can lie about the path. Marco Chiesa and Laurent Vanbever introduced Oscilloscope, which detects hijacks by looking at Round Trip Time (RTT) signals. If the time it takes for a packet to return suddenly changes in a way that doesn't match normal network congestion, it suggests the traffic has been diverted to a different physical location, exposing the hijack regardless of what the BGP message says.
Beyond Routing: The Naming and Delivery Gap
The trust problem extends upward to the naming and content delivery layers. In the DNS, "glue" records—the IP addresses provided to help resolvers find name servers—are often unverified. Research into The Availability and Security Implications of Glue in the Domain Name System shows that inconsistent handling of these records allows for "glue redirection attacks," where attackers point users to malicious servers.
Similarly, Content Delivery Networks (CDNs) often fail to verify domain ownership properly. Ziyu Lin and Jianjun Chen identified a systemic Absence of Domain Verification (DVA) in major CDNs. This flaw allows attackers to "hijack" high-reputation domains within the CDN's infrastructure. This is closely related to domain fronting, a technique Karthika Subramani and Manos Antonakakis found is still widely possible on 73% of tested CDNs, enabling users to mask malicious traffic behind legitimate domain headers.
Toward Verifiable Infrastructure
The ultimate goal of this research area is to replace implicit trust with cryptographic or behavioral verification.
- Locating the Source: Tools like LOCK, developed by Tongqing Qiu and Dan Pei, help pinpoint the exact AS responsible for a hijack by analyzing how "polluted" paths converge.
- Verifiable Forwarding: Xiaoliang Wang and Qi Li proposed FC-BGP, which uses "forwarding commitments." This ensures that a router doesn't just promise a path in the control plane but provides a verifiable proof that the path was actually followed in the data plane.
- Source Validation: To prevent spoofing, researchers like Huiliang Duan have explored Source Validation Architectures (SVA) where routers mark packets with path information, allowing end systems to verify that a packet actually came from where it claims.
By combining real-time anomaly detection with these new verification frameworks, the research community aims to transform the internet backbone from a system of blind trust into one of verifiable security.