A Survey on Network Partitioning in Blockchain Systems

Abstract

Despite its serious consequences, network partitioning remains one of the most overlooked threats in blockchain systems. When a blockchain network splits into disconnected segments-due to routing failures, deliberate attacks, or infrastructure outages-nodes in each segment continue building their own version of the ledger independently. When connectivity is restored, one chain must be discarded, and any transactions recorded only on the discarded chain are lost. If an attacker deliberately timed payments to both sides of the partition simultaneously, those payments may succeed on both chains-a double-spend. The severity of this threat varies by consensus design. Public proof-of-work chains like Bitcoin face double-spend risk. Proof-of-stake networks like Ethereum risk finality delays and validator penalties. Permissioned systems like Hyperledger Fabric halt block production entirely. Despite these distinct consequences, the research literature treats these platforms in isolation. This survey unifies the analysis. We review over twenty peer-reviewed works (2002-2024), covering distributed systems theory, deliberate attack vectors, consensus protocol behavior, and detection and mitigation strategies. We present a comparative taxonomy, identify four open research problems, and conclude that existing mitigations are largely reactive and platform-specific-a cross-platform approach to partition resilience is still needed.