On Scaling LT-Coded Blockchains in Heterogeneous Networks and their Vulnerabilities to DoS Threats
Harikrishnan K.,J. Harshan,Anwitaman Datta
2024-10-03
Abstract:Coded blockchains have acquired prominence as a promising solution to reduce storage costs and facilitate scalability. Within this class, Luby Transform (LT) coded blockchains are an appealing choice for scalability owing to the availability of a wide range of low-complexity decoders. In the first part of this work, we identify that traditional LT decoders like Belief Propagation and On-the-Fly Gaussian Elimination may not be optimal for heterogeneous networks with nodes that have varying computational and download capabilities. To address this, we introduce a family of hybrid decoders for LT codes and propose optimal operating regimes for them to recover the blockchain at the lowest decoding cost. While LT coded blockchain architecture has been studied from the aspects of storage savings and scalability, not much is known in terms of its security vulnerabilities. Pointing at this research gap, in the second part, we present novel denial-of-service threats on LT coded blockchains that target nodes with specific decoding capabilities, preventing them from joining the network. Our proposed threats are non-oblivious in nature, wherein adversaries gain access to the archived blocks, and choose to execute their attack on a subset of them based on underlying coding scheme. We show that our optimized threats can achieve the same level of damage as that of blind attacks, however, with limited amount of resources. Overall, this is the first work of its kind that opens up new questions on designing coded blockchains to jointly provide storage savings, scalability and also resilience to optimized threats.
Information Theory