Network routing is fundamentally built on two core assumptions: identity (who owns a destination address) and proximity (which path to that destination is the shortest or most efficient). Whether navigating the global Internet via the Border Gateway Protocol (BGP) or traversing a decentralized peer-to-peer (P2P) overlay network, nodes rely on their neighbors to advertise these metrics honestly.
Because these protocols historically prioritize connectivity over strict verification, they suffer from a shared structural vulnerability. If an adversary can falsely claim ownership of an address space or mathematically manipulate their perceived proximity to a target, they can seamlessly hijack network traffic. Understanding why these attacks work requires looking at how routing algorithms calculate the "best" path across both physical and logical topologies.
The Mechanics of Inter-Domain Hijacking
On the global Internet, autonomous systems (ASes) exchange routing information using BGP. This infrastructure is highly sensitive to prefix hijacking because of a foundational rule in IP routing: longest-prefix matching.
When a router receives multiple advertisements for a destination, it always prioritizes the most specific network mask. For example, a route to a /24 subnet will always override a route to a broader /16 block. As Sun et al. demonstrate, an attacker can exploit this behavior by announcing a highly specific sub-prefix that they do not actually own. Because of the longest-prefix match rule, traffic naturally detours to the attacker's infrastructure without breaking the underlying end-to-end connection, enabling stealthy data interception.
Legitimate AS (Advertises 192.168.0.0/16)
^
| (Traffic naturally flows here normally)
Routers
| (Attacker announces 192.168.1.0/24)
v
Malicious AS (Longest-prefix match wins; traffic hijacked)
To counter basic origin validation defenses (such as the Resource Public Key Infrastructure, or RPKI), sophisticated attackers employ AS path poisoning. Cryptographic origin validation typically only checks if the originating AS is authorized to announce a prefix. To bypass this, Sun et al. explain that an adversary can forge a routing path that appends the legitimate origin AS to the end of their malicious announcement. This trick satisfies origin validation checks while still allowing the attacker to manipulate the path vector and intercept traffic.
This vulnerability is exacerbated by the implicit trust model of inter-domain peering. As shown by Cavedon et al., many production routers are configured to accept unauthenticated connection requests, allowing unauthorized entities to establish peering sessions and inject false reachability information directly into core tables.
The Parallel in P2P Overlays: Distance Manipulation
The vulnerability to proximity manipulation is not unique to physical hardware; it replicates cleanly in logical networks, such as structured P2P overlays and Distributed Hash Tables (DHTs). Instead of using IP prefixes, networks like Ethereum’s node discovery protocol organize nodes using mathematical distance metrics, typically an XOR operation.
In an XOR-based topology, a node routes queries to neighbors that are mathematically "closest" to the target key. This design creates an opening for identity pre-computation. Henningsen et al. and Liu et al. demonstrate that an attacker can generate a large volume of cryptographic keys until they find an identifier that sits numerically adjacent to a target victim.
By returning these forged records during peer lookups, the attacker exploits the distance-based routing logic. The victim's routing table prioritizes these "close" malicious nodes, filling its high-priority buckets with adversarial peers. This effectively executes an eclipse attack, isolating the victim and controlling its view of the network without needing to flood the entire system.
Systemic Consequences Across Applications
When the underlying routing layer is compromised, the security assumptions of higher-level applications collapse. The impact spans multiple domains:
- Anonymity Networks: By executing active BGP hijacks, an adversary can force traffic from a Tor client and its corresponding exit node through an AS they control, enabling asymmetric traffic analysis to deanonymize users (Sun et al.).
- Tiered Security Pipelines: In multi-stage security architectures, the entire pipeline's integrity depends on the initial routing component. Nasr et al. show that if an attacker can manipulate inputs to cause a misclassification at the routing stage, malicious payloads can bypass downstream specialized scanners entirely.
- Physical Fleet Routing: The vulnerability extends to physical-world optimization algorithms. Estornell et al. demonstrate that strategically spoofing locations at critical network junctions tricks high-level routing algorithms into calculating highly suboptimal paths, causing massive physical detours and operational failures.
Defending the Routing Layer
Building resilience against routing manipulation requires moving away from unverified local reports toward structural verification and path asymmetry analysis.
Exploiting Path Asymmetry
While an attacker can easily manipulate downstream routing announcements, they cannot alter the routing information propagated by upstream providers before it reaches them. Qiu et al. leverage this asymmetry in their distributed measurement system, LOCK. By focusing exclusively on the trustworthy, unmodified segments of a network path where polluted paths converge, defenders can pinpoint the exact location of a prefix hijacker with high accuracy.
BGP-Aware Path Selection
Traditional overlay networks and relay systems select paths based purely on performance metrics like bandwidth or latency. However, Sun et al. show that incorporating AS resilience metrics directly into path selection algorithms significantly mitigates hijacking risks. By proactively avoiding paths hosted on vulnerable or poorly protected autonomous systems, applications can defend against network-level adversaries with minimal impact on performance.
Secure Underlays and the Deployment Hurdle
To bridge the gap between legacy BGP infrastructure and next-generation secure protocols, Birge-Lee et al. propose abstracting secure backbones into a virtual AS underlay. This allows legacy networks to protect their traffic from interception without requiring an immediate global upgrade of the Internet's core hardware.
However, deploying cryptographic defenses remains difficult due to operational realities. As Chang et al. observe, benign traffic engineering practices—such as sub-prefix announcements for load balancing or unrecorded provider-customer transfers—frequently conflict with static cryptographic authorizations. This causes high false-alarm rates in route origin validation, creating a persistent tension between strict security enforcement and network stability.
Key Foundational Work
Beware of BGP attacks (2004) — contributed 12 concept elements to this direction.
Origin authentication in interdomain routing (2003) — contributed 3 concept elements to this direction.
Visual-Based Anomaly Detection for BGP Origin AS Change (OASC) Events (2003) — contributed 1 concept elements to this direction.