TOWARDS SECURE PUBLIC BLOCKCHAIN PROTOCOLS AT SCALE

Abstract:

Emerging cryptocurrencies, such as Bitcoin and Ethereum, embody at their core a blockchain protocol which runs in large open networks with tens of thousands of participants. Moreover, the consensus protocols in public blockchains differ from traditional consensus protocols in that nodes have no established identities nor rely on a centralized PKI system. Currently, these blockchains employ proof-of-work (or PoW) mechanism to periodically agree on a set of new transactions. PoW requires network nodes to spend computational power to solve computationally expensive puzzles in a process called mining. This probabilistically elects a new leader who proposes a new set of transactions in a given period (or block). To compensate the nodes, the leader receives newly minted coins as rewards.

Designing a secure blockchain protocol at large scale is an open challenging problem since the protocol is open to manipulation by Byzantine or financially-motivated malicious participants. Moreover, the participants do not have any inherent identities and can join/leave at any time. As an example, despite having a decentralized design, both Bitcoin and Ethereum end up being operated by a handful of mining pools due to the stable payout that mining pools provide to miners. As a result, the security of pooled mining protocol is critical as one can perpetrate severe attacks to any blockchain systems by exploiting the mining pools. Further, these centralized pools pose the risk of transaction censorship since only these pools can verify and include new transactions. There exists an attempt to build decentralized mining pool to sidestep the mining centralization problem, but the de- sign only works for Bitcoin and poorly scale when more miners join the pool. In addition, at its core, existing blockchain protocol exhibits security, but does not scale: it processes 3-7 transactions per second, irrespective of the available computation capacity at hand. There exists no existing-protocol that scales up the transaction rate of the network as the network size increases.

In this thesis, we first analyse existing pool reward sharing protocols and show that they are vulnerable to an attack strategy called "block withholding attack". Our game- theoretic analysis proves that the attack is always well-incentivized, i.e. attacker gets more profit, in the long-run, but may not be so for a short duration. As a solution to resolve the mining centralization problem, we propose SMARTPOOL, a novel protocol design for a decentralized cryptocurrency mining pool which is efficient in large scale. Our protocol leverages smart contracts, autonomous blockchain programs, to give transaction selection control back to miners while yielding low-variance payouts. As the last contribution, we propose a new distributed agreement protocol for public blockchains called ELASTICO that is scalable and secure even with presence of Byzantine adversary. ELASTICO scales transaction rates almost linearly with available computation for mining: the more the computation power in the network, the higher the number of transaction blocks selected per unit time.