Blockchain Communication Vulnerabilities

TL;DR

Algorand is vulnerable to packet loss attacks, Aptos is vulnerable to targeted load attacks and leader isolation attacks, Avalanche is vulnerable to transient failure attacks, Redbelly's performance is impacted by packet loss attacks and Solana is vulnerable to stopping attacks and leader isolation attacks.

Abstract

Blockchains are diverse in the way they handle communications between their nodes to disseminate information, mitigate attacks, and agree on the next block. While security vulnerabilities have been identified, they rely on an attack custom-made for a specific blockchain communication protocol. To our knowledge, the vulnerabilities of multiple blockchain communication protocols to adversarial conditions have never been compared. In this paper, we compare empirically the vulnerabilities of the communication protocols of five modern in-production blockchains, Algorand, Aptos, Avalanche, Redbelly and Solana, when attacked in five different ways. We conclude that Algorand is vulnerable to packet loss attacks, Aptos is vulnerable to targeted load attacks and leader isolation attacks, Avalanche is vulnerable to transient failure attacks, Redbelly's performance is impacted by packet loss attacks and Solana is vulnerable to stopping attacks and leader isolation attacks. Our system is open source.

Figures (11)

• Figure 1: The different topologies used by the different blockchains where valid., VF, PubF, AF, PF, gov. and cand. stand for validator, validator full, public full, archival full, pruned full, governor and candidate, respectively, and where cons, sync and req stand for consensus, synchronization and request.
• Figure 2: During an 800-second load attack, Aptos congestion prevents it from committing as fast as when the attack stops (after 800 seconds).
• Figure 3: Aptos node receiving the transactions generates a quarter of total outgoing traffic during the experiment, up to 5 times higher compared to other blockchains.
• Figure 4: Transaction latency distributions of an Aptos network under a constant uniform workload of 200 TPS for 800 seconds.
• Figure 5: Network traffic (TX+RX Rate) and transaction throughput of Avalanche under a transient failure attack. With throttling enabled (red), attempts to sample failed nodes for consensus during the failure window (133-266s) create a large spike in unproductive network traffic. This activity triggers the throttling mechanism, which then suppresses network activity and prevents throughput from recovering post-failure, causing sustained transaction loss. In contrast, the non-throttled system (blue) shows no traffic spike and recovers quickly.