design multi-proposer consensus

In traditional single-proposer blockchains, validators monopolize transaction inclusion and ordering, enabling extraction of Miner-Extractable Value (MEV). Existing application-layer mitigations rely heavily on on-chain auctions, which require the consensus layer to simultaneously satisfy selective censorship resistance and transaction-content hiding—properties not jointly achieved by prior protocols. Method: We propose the first multi-concurrent-proposer consensus protocol that simultaneously achieves both properties: multiple proposers concurrently submit cryptographically committed transaction bundles; a commit-reveal mechanism ensures content-hiding prior to finalization; and parallel validation guarantees that any valid transaction cannot be selectively delayed. Contribution/Results: Our design fundamentally undermines the root cause of MEV—validator monopoly over transaction ordering—by decentralizing sequencing authority at the consensus layer. Without relying on application-layer auctions, it drastically reduces the exploitable MEV surface while preserving security, fairness, and high throughput.

Traditional Byzantine fault-tolerant (BFT) consensus protocols are constrained by a single-leader architecture, which limits block intervals due to network latency and exposes the system to censorship and proposal manipulation risks. This work proposes Cadence, a novel BFT protocol that partitions time into equal-length slots and employs an aggressive pipelining design, enabling multiple proposer instances to operate in parallel within each slot. Cadence achieves short-term censorship resistance and proposal confidentiality under a multi-proposer setting—while preserving the low-latency fast path of single-leader protocols—for the first time. Built upon a partially synchronous model, it integrates Chorus slot-based consensus (featuring three-round fast finality and one-round speculative finality) with a Conductor coordinator to form an efficient BFT system. Experiments demonstrate that, with 200 nodes and five proposers per slot, Cadence attains an average finality latency of 219 ms (167 ms for speculative finality), and transactions wait only 50 ms on average before inclusion in a proposal under a 100 ms block interval.

Existing multi-Byzantine fault-tolerant (Multi-BFT) consensus protocols rely on a global ordering layer to serialize cross-instance blocks, resulting in poor scalability, widespread fault propagation, and complex deployment. This paper proposes the first global-ordering-free Multi-BFT framework, introducing an object-centric execution model that integrates fine-grained transaction partitioning, parallel BFT across multiple instances, lightweight lock-based coordination, and proactive deadlock recovery—ensuring strong consistency while significantly enhancing concurrency. Experimental evaluation demonstrates that, under both LAN and WAN settings with 128 replicas, our framework achieves 1.3×–2.8× higher throughput than state-of-the-art approaches including State Machine Replication and HotStuff-Multi. Moreover, it exhibits superior robustness against high-latency nodes.

This paper addresses a novel class of miner-extractable value (MEV) problems in multi-concurrent-proposer blockchains—arising from delayed execution-order finality—including simultaneous frontrunning, proposer-to-proposer auctions, and availability-proof-driven timing races. We propose a harm-normalized delay-and-inclusion model and derive a closed-form delay envelope ( M( au) ) to characterize equilibrium properties of censorship, replication, and auction games. Our method integrates deterministic priority DAG-based scheduling with repetition-aware payment mechanisms to suppress simultaneous MEV. The approach unifies game-theoretic modeling, latency analysis, protocol-layer incentive design, and equilibrium analysis. Theoretical analysis and experimental evaluation demonstrate that, without relying on centralized block builders, specific protocol configurations effectively mitigate concurrency-specific MEV while preserving security guarantees and achieving high throughput.

Asynchronous Byzantine Fault Tolerant (BFT) consensus has long struggled to simultaneously achieve high throughput and low latency: traditional chain-based protocols (e.g., PBFT) offer low latency but suffer from throughput bottlenecks, whereas emerging DAG-based BFT protocols improve throughput at the cost of significantly increased latency. This paper proposes a novel DAG-BFT protocol that decouples message propagation from consensus decision-making, enables parallel proposal generation by multiple nodes, and introduces a lightweight causal-order verification mechanism. For the first time in leaderless DAG-BFT, it achieves an average message hop count of 4.5 for transaction finality. Experimental results show up to 60% reduction in end-to-end latency while matching state-of-the-art DAG-BFT throughput. The protocol provides formally proven safety, liveness, and rapid fault recovery—approaching the theoretical optimum in the BFT latency-throughput trade-off.

This study addresses the challenge of fairly comparing how different quorum constructions affect client-perceived latency in Byzantine Fault Tolerant (BFT) systems, as implementation-level discrepancies inherently confound such evaluations. To overcome this limitation, we extend BFT-SMaRt by introducing a modular abstraction layer that serves as the first unified evaluation framework. This architecture enables diverse quorum constructions to operate under identical optimization techniques and wide-area network (WAN) simulation environments, effectively eliminating interference from underlying implementation details. Consequently, this work achieves a rigorously fair cross-construction comparison, precisely quantifying and revealing how distinct quorum designs influence client-observed latency in WAN scenarios.

This work addresses the trade-off between scalability and security in existing committee-based sharded parallel Byzantine Fault Tolerance (BFT) protocols, where random committee assignment further limits performance. The paper introduces, for the first time, an optimization-driven committee configuration approach for trusted parallel BFT systems, formulating a mixed-integer programming model that jointly considers node failure rates and communication latency to minimize transaction latency under both normal operation and trusted hardware failure scenarios. By integrating Trusted Execution Environments (TEEs), committee sharding, and parallel BFT mechanisms, the proposed method achieves significant performance gains: experiments on Microsoft Azure virtual machines demonstrate throughput improvements of 15% and 21% over baseline protocols in normal and fallback modes, respectively, while maintaining strong security, scalability, and efficiency.

Existing parallel block dissemination consensus protocols struggle to balance finality latency and fault tolerance while remaining vulnerable to single points of failure. This work proposes a novel mechanism integrating single-round voting consensus with multi-chain data distribution: chain-level voting feedback enables deep block attachment and efficient block ordering, while a chain-level fault-tolerance design ensures that a faulty producer affects only its own chain, allowing other chains to proceed after at most one view delay and preventing honest blocks from being maliciously excluded. Under the assumption $n \geq 5f+1$, combined with relative leader proposals and self-certified block availability, the system achieves block ordering within $t+2\delta$ to $t+3\delta$ with per-view communication overhead of merely tens of kilobytes, independent of transaction volume.

This work addresses the trilemma in asynchronous Byzantine Fault Tolerant (BFT) protocols combined with Proof-of-Stake (PoS)—namely, simultaneously achieving Sybil attack resistance, reward fairness, and mitigation of wealth concentration. The authors propose a dual-channel DAG-BFT protocol that decouples anchor selection from reward distribution: the selection channel employs superlinear stake weighting to ensure Sybil attack returns remain below unity, while the reward channel adopts square-root normalization to effectively curb the rich-get-richer effect. Integrating a lagged reputation mechanism with a 2f+1 strong-support commit rule, the protocol achieves consensus on operational quality without external oracles. Experiments demonstrate a Gini coefficient of 0.149 (versus 0.488 under Pure-PoS), a monotonically decreasing Herfindahl-Hirschman Index (HHI) down to 0.021, an optimal Sybil split factor K* = 1, a low success-rate coefficient of variation (5.2%) under perturbations, and sustained commit rates above 71.1% up to b = 1/3.

This study addresses the inherent tension in consensus protocols, which are efficient under partial synchrony yet prone to stalling in asynchronous environments. To resolve this, we propose a hybrid consensus algorithm built upon a shared directed acyclic graph (DAG). The approach alternates between partially synchronous and asynchronous commit rules, dynamically adjusting execution periods to balance performance with liveness. It enables adaptive mode switching without additional communication overhead and incorporates a hidden leader mechanism to guarantee asynchronous liveness. The protocol is formally verified using Lean 4. Experimental results demonstrate that the proposed method achieves throughput comparable to synchronous protocols under favorable network conditions while preserving liveness approaching that of asynchronous protocols in adverse environments, thereby effectively reconciling efficiency with robustness.