PaperlayerView exact revision on arXiv

Source-linked AI summary

A Survey on Consensus Mechanisms and Mining Strategy Management in Blockchain Networks

Wenbo Wang, Dinh Thai Hoang, Peizhao Hu, Zehui Xiong, Dusit Niyato, Ping Wang, Yonggang Wen, Dong In Kim

arXiv:1805.02707v4cs.CR

TL;DR

The paper addresses the lack of a comprehensive review of decentralized consensus mechanisms in blockchain networks. It surveys implementation organization, consensus and incentive design, game-theoretic node strategies, applications, and open issues, concluding with a broad synthesis of consensus protocols and their effects on participants and applications.

  • Problem

    Existing literature lacks a comprehensive review of decentralized blockchain consensus spanning protocol design, incentive mechanisms, node strategies, applications, and open issues.

  • Method

    The paper conducts a survey organized around blockchain implementation stacks, permissionless consensus protocols, incentive mechanisms, game-theoretic strategies, and emerging applications.

  • Results

    The survey covers BFT-based, Nakamoto, virtual, and hybrid protocols, emphasizing their links to Byzantine agreement, incentive compatibility, and consensus-participant strategies.

  • Takeaways & Limitations

    Consensus mechanisms affect blockchain deployers, consensus participants, and users, making incentive design and protocol design central to evaluating blockchain networks and applications.

Abstract

from arXiv · show

The past decade has witnessed the rapid evolution in blockchain technologies, which has attracted tremendous interests from both the research communities and industries. The blockchain network was originated from the Internet financial sector as a decentralized, immutable ledger system for transactional data ordering. Nowadays, it is envisioned as a powerful backbone/framework for decentralized data processing and data-driven self-organization in flat, open-access networks. In particular, the plausible characteristics of decentralization, immutability and self-organization are primarily owing to the unique decentralized consensus mechanisms introduced by blockchain networks. This survey is motivated by the lack of a comprehensive literature review on the development of decentralized consensus mechanisms in blockchain networks. In this survey, we provide a systematic vision of the organization of blockchain networks. By emphasizing the unique characteristics of incentivized consensus in blockchain networks, our in-depth review of the state-of-the-art consensus protocols is focused on both the perspective of distributed consensus system design and the perspective of incentive mechanism design. From a game-theoretic point of view, we also provide a thorough review on the strategy adoption for self-organization by the individual nodes in the blockchain backbone networks. Consequently, we provide a comprehensive survey on the emerging applications of the blockchain networks in a wide range of areas. We highlight our special interest in how the consensus mechanisms impact these applications. Finally, we discuss several open issues in the protocol design for blockchain consensus and the related potential research directions.

I. INTRODUCTION

Blockchain networks evolved from public transaction ledgers into platforms for decentralized data processing and autonomous organization. This survey addresses the lack of a global consensus-protocol review by examining protocol design, incentives, node strategies, applications, and open research issues.

  • Blockchain networks provide immutable ledgers and platforms for data-driven autonomous organization, with open-access systems emphasizing disintermediation, public functionality access, and tamper-resilience.
  • Consensus design must balance data consistency, finality speed, Byzantine robustness, scalability, throughput, resource consumption, and transaction-confirmation latency.Open-access protocols can improve scalability relative to classical Byzantine consensus while limiting throughput and consuming substantial physical resources.
  • Existing surveys rarely provide a global view of blockchain consensus protocols, motivating this survey’s systematic perspective on consensus formation in open-access P2P networks.The survey relates blockchain consensus to distributed-system state maintenance and Agreement-Validity-Termination properties.
  • The survey reviews blockchain data and network organization, cryptographic consensus mechanisms, rational-node behavior, protocol performance-versus-scalability concerns, and emerging decentralized applications.It also considers virtual block-mining and hybrid consensus as extensions of the protocol-design landscape.
  • The paper distinguishes open-access and permissionless blockchain networks as terms for networks whose functionalities, rather than only data, are publicly accessible.

II. PROTOCOL OVERVIEW AND PRELIMINARIES

Blockchain systems organize data, communication, consensus, and programmable execution into cooperating protocol layers. Cryptographic structures protect identities and records, while consensus maintains a canonical blockchain state across replicated nodes.

  • A. Overview of Blockchain Network Protocols: Blockchain network implementation comprises data and network organization, distributed consensus, and an autonomous-organization framework based on smart contracts and virtual execution.The virtual computer layer interconnects lower-level protocols with applications and abstracts data organization, propagation, and consensus formation.
  • A. Overview of Blockchain Network Protocols: Data-organization protocols establish secured node identities and cryptographic dependencies among records, including hash-linked blocks arranged as a chronological chain.Blockchain data structures can also expand beyond linear chains into trees and graphs to improve processing efficiency, scalability, or security.
  • A. Overview of Blockchain Network Protocols: Network protocols organize P2P discovery, routing, encrypted transmission, and synchronization, enabling the consensus layer to maintain blockchain originality, consistency, and order.Consensus nodes seek agreement on replicated blockchain-state updates despite conflicting inputs and Byzantine behavior.
  • 2) Block Organization, Hash Pointer and Merkle Tree: A blockchain is an append-only, canonically agreed data structure whose local replica has transaction, block, and chain levels requiring distinct integrity and authenticity protections.
  • 2) Block Organization, Hash Pointer and Merkle Tree: Blocks contain transaction subsets, use hash pointers to specify adjacent-block ordering, and can store Merkle roots as compact tamper-evident transaction-set digests.Lightweight Merkle-root storage supports quick validation and synchronization but requires retrieving complete transactions from peers.

C. Blockchain Networks

Permissionless networks allow nodes to join and activate functionalities without identity authentication, requiring P2P protocols and consensus mechanisms that tolerate pseudo-identities, poor synchronization, and Byzantine behavior. Node roles differ by storage, verification, propagation, and consensus capabilities.

  • C. Blockchain Networks: Permissionless nodes may freely join and activate network functionalities, whereas permissioned networks restrict core functions such as consensus participation or data propagation to authorized nodes.
  • C. Blockchain Networks: Permissionless network protocols seek random node topology and efficient information propagation for synchronizing blockchain replicas, commonly using modified P2P discovery and communication protocols.Transactions and blocks are typically broadcast by flooding persistent P2P links in a gossip-like manner.
  • C. Blockchain Networks: Open-access nodes are categorized as lightweight, full, or consensus nodes: lightweight nodes retain block headers, full nodes store complete replicas, and consensus nodes can publish blocks.Consensus nodes therefore have an opportunity to influence the canonical blockchain state.
  • C. Blockchain Networks: Consensus nodes maintain local blockchain replicas and must agree on a common view despite Byzantine failures such as Sybil attacks, double spending, software-induced forks, collusion, and connection errors.
  • C. Blockchain Networks: Blockchain consensus properties include validity, agreement, liveness, and total order, while permissionless protocols prioritize scalability and tolerance of pseudo-identities and poor synchronization.Direct voting-based BFT protocols do not fit permissionless networks because they lack identity authentication; incentive-based schemes such as Nakamoto consensus are widely adopted.

E. Nakamoto Consensus Protocol and Incentive Compatibility

Nakamoto consensus uses a computation-intensive PoW race to select valid blocks in permissionless networks, but its incentive compatibility is questioned because rational nodes may not propagate information. The protocol therefore faces trade-offs among security, efficiency, scalability, decentralization, and fairness.

  • Protocol operation: Nakamoto consensus defines chain validation, chain comparison and extension, and PoW solution as its three major procedures.Nodes compare local and received blockchain views and adopt the longest chain.
  • Proof of Work: PoW searches for a nonce whose candidate-block hash is below a difficulty-dependent target D(h).The candidate string x contains block data such as transactions and reference hash pointers; formally, bh = H(x∥nonce) ≤ D(h).
  • Mining incentives: Computation-intensive mining raises hash-query capacity requirements, deters Sybil identities, and imposes electricity costs on participants.The protocol introduces incentives to support continued participation despite these costs.
  • Forks and convergence: A temporary fork occurs when nodes simultaneously produce different valid block-header proposals, requiring convergence toward one canonical chain.Both competing headers can satisfy the PoW condition, while the protocol relies on majority adherence to the longest-chain rule.
  • Incentive compatibility: Nakamoto consensus assumes individually rational nodes and seeks incentive compatibility, but rational nodes may lack incentives to propagate transactions or blocks.The survey states that additional protocol measures are needed to address forking and related incentive concerns.
  • Design principles: Consensus design should balance processing throughput with network scalability while encouraging information propagation, decentralization, and fairness.The stated design principles also seek to discourage coalitions such as botnets and mining pools.

A. Permissionless Consensus via Zero-Knowledge Proofs

Permissionless consensus can replace authenticated leader coordination with resource-based, verifiable selection and non-interactive proof puzzles. The survey connects these designs to security, efficiency, environmental, and application-level trade-offs.

  • Resource-based leader election: PoW-like consensus simulates leader election by selecting proposers in proportion to verifiable resources such as computational power, memory, or storage.Unlike traditional BFT protocols, nodes follow the longest-chain rule without an all-to-all messaging phase.
  • Proof-of-resource design: Permissionless PoX requires verifiable unpredictability, one-way puzzles, soundness, completeness, and a complexity gap between solving and verification.Non-interactive design avoids the excessive message overhead of verifier-designated interactive challenges.
  • Puzzle objectives: PoX puzzle design should resist computational-resource aggregation or outsourcing, reduce environmental costs, and potentially provide useful services alongside incentives.The survey presents these as design objectives for extending or modifying Nakamoto-style protocols.
  • Primitive PoW: Hash-based PoW creates fresh, unpredictable challenges from each block proposal without a designated verifier or trusted third-party initializer.Collision resistance and pre-image resistance support the hash function’s pseudorandom behavior in this role.
  • Security properties: Combined Poisson processes yield the resource-proportional leader-election distribution, while the reviewed analysis states that the process tolerates any fraction of Byzantine nodes.The corresponding protocol analysis also links sufficiently deep blocks with non-reversible transaction data.
  • Performance trade-offs: Block interval and propagation delay create a security-efficiency trade-off: larger blocks increase validation time, while network scale and channel capacity affect diffusion delay and orphaning risk.K-consistency analysis further gives a sufficient condition K > K0(L)=c log(L), except with negligible probability in K.

C. Proof of Concepts Attached to Useful Resources

Useful-resource PoX schemes replace conventional PoW work with puzzles tied to computation or storage, while differing in soundness, leader-election behavior, and resource incentives.

  • Alternative PoX schemes seek to incentivize useful resource provision and improve blockchain security, fairness, or eco-friendliness.
  • uPoW adds usefulness to completeness, soundness, and hardness by enabling puzzle solutions to reconstruct a target computation F(x).
  • uPoW uses fine-grained-hard problems and Fiat-Shamir-style hashing to produce non-interactive puzzles whose k-OV solutions follow a Bernoulli distribution.
  • Storage-based schemes use retrievability proofs, Merkle proofs, or repeated proofs over time to demonstrate possession or dedication of data.
  • Permacoin’s PoR uses a fixed number of random-oracle queries and therefore does not simulate random leader election, weakening its claim to repurpose PoW.
  • Filecoin’s PoST proves holding time by generating a PoR sequence over a specified period and introduces expiring spacetime storage.

D. Proof of Concepts for Performance Improvement

Performance-oriented PoX schemes target security, fairness, sustainability, ASIC resistance, and human or storage-based alternatives to repeated hash computation.

  • Performance-oriented PoX designs aim to improve security, fairness, and sustainability beyond conventional PoW.
  • Memory-hard PoW schemes such as Equihash and Ethash address ASIC concentration by constraining parallelism through memory or sequential computation.
  • Ethash searches a growing pseudorandom DAG through a fixed-iteration hash loop, making each puzzle attempt inherently sequential.
  • Ethash remains partially sequential within one preimage attempt, so miners can divide searches into subtasks and outsource them to mining workers.
  • Merkle-based nonoutsourceable puzzles can make preimage search independent of block payload, allowing workers to replace the original payload undetected.
  • Proof of Space organizes committed random data as a DAG, whereas Proof of Human-work targets eco-friendliness, usefulness, and resistance to centralization.

IV. STRATEGIES OF RATIONAL NODES IN THE FRAMEWORK OF NAKAMOTO CONSENSUS PROTOCOLS

The survey reviews incentive compatibility in Nakamoto consensus by modeling rational miners’ strategies, while noting that most analyses focus on PoW-based Bitcoin.

  • The survey examines rational consensus-node strategies in block-mining competitions under the assumption that nodes act rationally.

A. Incentive Compatibility of Nakamoto Protocols

Nakamoto protocols use monetary rewards to align miners with blockchain rules, but mining pools, transaction fees, and propagation incentives create strategic compatibility challenges.

  • Block winners receive transaction fees and token issuance rewards, making monetary incentives central to rule-following in Nakamoto protocols.
  • Joining mining pools reduces income variance and stabilizes profits, but concentrates control in pool servers and forms cartels contrary to unstructured-network goals.
  • Without propagation rewards, nodes may withhold fee-paying transactions to reduce mining competition, motivating incentive mechanisms for broadcasting.
  • Fee-sharing protocols can make honest propagation rational by rewarding relayers, including through hop-count-dependent payments or enforceable relay compensation.
  • When block rewards dominate fees, miners may produce empty or nearly empty blocks, and transaction confirmation can be blocked as a DDoS attack.
  • Rational miners prioritize higher-fee transactions; small-value transactions can face delays from 20 minutes to 30 days.

B. Resource Investment and Transaction Selection for Mining under Nakamoto Protocols

The survey examines how miners allocate resources and select transactions under Nakamoto protocols, then reviews strategic deviations that can increase rewards or threaten network operation.

  • Resource investment: Mining-resource investment balances expensive computational costs against the probability of winning block rewards.More resources increase a miner’s chance of winning, but success also depends on other miners’ investments.
  • Profit modeling: The modeled mining profit incorporates both resource investment and transaction selection, supporting evolutionary, hierarchical, and auction-based game formulations.The model captures miners’ strategic choices in different network setups.
  • Transaction selection: Block-size selection trades off computational power against propagation time, and higher-powered miners may prefer larger blocks.The analysis also identifies equilibria in which miners include no transactions in their proposed blocks.
  • Limitations: Propagation-delay assumptions constrain existing mining-strategy analyses because practical miners may not mine synchronously or honestly propagate blocks.The survey identifies further investigation as necessary when these assumptions do not hold.
  • Selfish mining: Selfish mining can raise a pool’s revenue above its relative computational size by withholding blocks and exploiting blockchain forks.For a given μ, a selfish pool of size α earns revenue above its relative size in the range of 1−μ.
  • Selfish mining: Stubborn mining changes private-block publication behavior and achieves up to 13.94% higher gains than typical selfish mining in simulations.The stubborn strategy may reveal private blocks incrementally even after losing the lead.

2) Block Withholding in Pool-Based Mining:

The survey reviews block-withholding strategies in pool-based mining, showing how attackers can alter other pools’ winning probabilities and how protocol or pool changes may mitigate these incentives.

  • Block withholding: Block withholding splits computational power across pools while withholding valid contributions to reduce honest miners’ winning probabilities.The strategy is analyzed as a way for selfish miners to increase revenue.
  • Pool competition: Repeated-game analysis finds a unique equilibrium in which pool size determines the attacking rates between two pools.Pools can adjust their next-round attack rates after observing an opponent’s behavior.
  • Strategic incentives: Block-withholding attacks can benefit uninvolved mining pools, creating incentives for them to sponsor attacks against a victim pool.The attacker’s expected gain can therefore exceed the gain in the basic attack model.
  • Mitigation: Proposed mitigations include known-solution mining tasks, additional block-data fields, and extra rewards for miners who find valid blocks.These mechanisms aim to reduce the revenue of selfish miners and discourage block withholding.
  • Lie-in-wait: Lie-in-wait attacks postpone block submission while redirecting computational power to the pool where the block was found.The attack risks losing both the withheld reward and the cost of mining if another pool finds a block first.

4) Pool Hopping Strategy:

Pool hopping exploits pay-per-share payment timing: miners can leave after early-round shares become especially valuable. Share scoring and alternative payment schemes are proposed, but analytical evidence for their effectiveness remains incomplete.

  • Pool Hopping Strategy: Pool hopping exploits the payment mechanism of mining pools to increase an individual miner’s profits.The strategy is described specifically for pay-per-share mining pools.
  • Pool Hopping Strategy: Early-round shares can yield higher rewards because pay-per-share payments occur immediately while share value declines as more shares accumulate.A miner can therefore mine early in one round and then move to another pool to increase revenue.
  • Pool Hopping Strategy: Slush’s method uses an exponential score s(t) = e^t/δ to reduce the value of early shares and increase the value of later shares.The score depends on submission time t and scoring-rate parameter δ.
  • Pool Hopping Strategy: Pay-per-last-N-shares and payment-contract-based mechanisms are also proposed as possible responses to pool hopping.The survey states that analytical studies of these mechanisms are missing.
  • Broader Protocol Context: The surveyed consensus literature spans PoW, PoS, committee selection, and hybrid designs combining permissionless and classical BFT properties.Pure PoS replaces puzzle competition with asymmetric-key signing and verification, whereas hybrid designs target performance improvement.

B. Issues of Incentive Compatibility in PoS

PoS incentive compatibility depends on assumptions about malicious stake and validation costs, while the literature identifies nothing-at-stake, grinding, and stake-trade vulnerabilities. Hybrid PoW-BFT designs seek stronger consistency and improved finality, but BFT scalability remains limited by message complexity.

  • Issues of Incentive Compatibility in PoS: Honest behavior is a δ-Nash equilibrium in PoS when malicious stakes remain below a threshold and endorsers ignore transaction-validation costs.At δ-NE, each player’s payoff is within δ > 0 of the equilibrium payoff.
  • Issues of Incentive Compatibility in PoS: Nothing-at-stake attacks allow a PoS leader to create conflicting blocks on multiple forks without expending additional resources.Proposed countermeasures include exposing equivocation, blacklisting repeated failures, and confiscating deposits.
  • Issues of Incentive Compatibility in PoS: Grinding attacks let PoS leaders or committees test alternative block headers to influence future leader or committee selection.The attack targets the randomness input used for subsequent elections.
  • Issues of Incentive Compatibility in PoS: Existing PoS analyses often ignore stake trading outside the blockchain, such as trading at exchange markets.A cited study gives a counterexample to persistence under assumptions including unconstrained liquidity.
  • Hybrid Consensus: Hybrid consensus combines Nakamoto-style mechanisms with BFT to pursue stronger consistency and immediate finality.Bitcoin-NG separates leader election from transaction serialization, while PeerCensus separates block creation from transaction commitment.
  • Hybrid Consensus: BFT consensus offers high throughput and immediate finality but incurs high message complexity, restricting the number of consensus replicas.Permissionless protocols provide broader scalability at a different performance trade-off.

A. Off-chain and Side-chain Techniques

Off-chain, side-chain, and sharding techniques improve throughput by reducing global validation or partitioning state and consensus. Their designs range from payment channels and independent side-chains to dynamically formed BFT committees and full sharding.

  • Off-chain Techniques: Payment channels keep balance-state updates off-chain and broadcast only the latest commitment transaction when a channel closes.This reduces the number of transactions requiring network-wide mining and validation.
  • Off-chain Techniques: Lightning-style channels require 2-of-2 multisignatures and smart contracts to address double-spending risk.A funding transaction locks both parties’ tokens in escrow before channel payments proceed.
  • Side-chain Techniques: A side-chain independently validates a subset of transactions and tracks the corresponding assets, introducing parallelism into the network.The side-chain approach assigns only part of the total transaction workload to each chain.
  • Sharding: Sharding partitions global blockchain state into parallel subsets maintained by node committees rather than by the entire network.The approach aims to improve throughput while retaining permissionless membership.
  • Sharding: Aspen organizes transactions into service-specific sub-blockchains and uses periodic checkpoint blocks for synchronization.It is instantiated on Bitcoin-NG and requires PoW for checkpoint generation.
  • Sharding: Elastico dynamically forms multiple BFT committees and achieves roughly O(n) message complexity with almost linear throughput scalability in hash power.Unlike several earlier throughput-scalable protocols, it also supports generic data services with non-spendable transactions.
  • Sharding: OmniLedger and RapidChain use two-level BFT processes to parallelize data storage and network consensus across shards.Their network-level processes generate randomness or coordinate committee formation, while shard-level committees run intra-committee consensus.

C. Nonlinear Block Organization

Nonlinear block organization addresses scalability by allowing trees or DAGs to incorporate more than a single longest-chain path. These protocols use subtree weight, selective inclusion, or virtual voting, although complete liveness proofs remain open for IOTA Tangle.

  • Nonlinear Block Organization: GHOST selects a main chain using subtree weight rather than simply choosing the longest chain.A block with greater subtree weight can replace the longest-chain candidate in the current view.
  • Nonlinear Block Organization: DAG-based protocols allow blocks to reference multiple predecessors and selectively include non-conflicting off-chain descendants in the ledger view.The selected main chain remains a linear ordering extracted from the DAG.
  • Nonlinear Block Organization: SPECTRE allows concurrent DAG growth without a main branch and uses virtual pairwise voting to order blocks.Each block contributes to determining the relative order of block pairs.
  • Nonlinear Block Organization: Conflux is claimed to provide 100% utilization of off-chain blocks and thereby improve network scalability.Its design modifies GHOST by adding reference pointers to existing blocks without descendants.
  • Nonlinear Block Organization: IOTA Tangle requires each new transaction to approve or reference more than two earlier transactions, accumulating future transaction weight for confirmation.The protocol publishes transactions directly onto a transaction DAG rather than packaging them into blocks.
  • Nonlinear Block Organization: Complete theoretical proof of IOTA Tangle’s liveness remains an open issue.A cited study nevertheless indicates that similarly capable self-interested nodes may reach an almost symmetric Nash equilibrium.

A. General-Purpose Data Storage

Blockchain-based applications extend beyond cryptocurrency ledgers to decentralized storage, service brokering, access control, content delivery, auditing, and crowdsourcing. These systems use smart contracts and blockchain records to coordinate interactions, payments, access, and delivery proofs among mutually distrustful participants.

  • Data storage: General-purpose blockchain storage preserves metadata or pointers to off-chain data, but direct on-chain storage is limited by message length and full replication requirements.Off-chain designs use DHTs for IoT data while publishing storage pointers on-chain.
  • Decentralized service provision: A generic blockchain service framework sequences smart contracts for registration, access granting, hosting requests, peer bidding, delivery negotiation, and proof-backed completion.The framework treats the blockchain as a system integrator for decentralized service provision.
  • Application contexts: Blockchain applications coordinate wireless access, vehicular key delivery, content brokering, cloud auditing, and crowdsourcing through decentralized records and smart contracts.Examples include spectrum auctions, regional vehicle access, CDN delivery monitoring, tamper-proof provenance, and nonmanipulable brokering.

C. Consensus Provision and Computation Offloading under Nakamoto Protocols

This section examines consensus provision as a resource-allocation problem in Nakamoto-based blockchains, including computation offloading modeled through optimization and game theory. It also situates these designs within broader open issues involving efficiency, storage growth, privacy, and future consensus research.

  • Consensus provision: Nakamoto-based consensus requires resource consumption for proof-of-work puzzle competitions, motivating studies that treat consensus provision as decentralized resource allocation.These studies view consensus as the objective to achieve rather than as a ready-to-use service.
  • Computation offloading: PoW task offloading can be modeled as a two-stage Stackelberg game in which providers set computational-resource prices and edge devices choose purchased capacity.Utility models account for transaction transmission delay and blockchain-forking probability.
  • Open issues and directions: The survey identifies unresolved trade-offs between blockchain security and resource efficiency, alongside storage growth, pruning, and privacy challenges.It also surveys consensus protocols, incentive compatibility, applications, and related research directions.
Loading 1805.02707v4…