In briefEffective-balance design rules for Ethereum staking rewards
SummaryThis paper treats Ethereum's effective-balance rules as a mechanism-design problem, where the protocol sets a reward coefficient, activation threshold, and maximum effective balance. For a fixed set of validators and identities, it proves a unique Nash equilibrium in validator stakes and analyzes incentives for splitting or merging identities. An illustrative simulation uses synthetic agents calibrated to Beacon Chain, Lido, and Etherscan data, and reports that lowering the reserve price from 32 ETH to 30 ETH raises active identities by 5.922% and total effective stake by 0.098%.
In briefHybrid DeepXAI Fusion Network: A Deep Learning Model for Explainable and Secure Cloud Decision-Making
SummaryThis paper proposes the Hybrid DeepXAI Fusion Network, a deep learning framework that combines temporal, relational, and contextual features with explainable AI mechanisms. It integrates 1D-CNNs, Transformer encoders, and graph neural networks through attention-based fusion, using CAM and an explanation loss. The authors report 98.42% accuracy across healthcare, finance, and energy datasets and above 91.5% accuracy under PGD adversarial perturbations.
B. Nancharaiah, Ramakrishna Kolikipogu, J. Sulthan Alikhan et al.
In briefFlatMerkle: Logarithmic-size proofs over a plain hash stream
SummaryThis paper presents FlatMerkle, a construction that gives Merkle-tree-style inclusion proofs with logarithmic size and verification time over an ordinary linear hash stream. It works by periodically writing the hash function's past internal states back into the data as ordinary content, so the final digest is a single unmodified hash call. The construction supports compact proofs for any element and proofs that a published file is malformed.
In briefA quantum-resistant security protocol for decentralized edge-IoT peer-to-peer energy trading
SummaryThe paper proposes a mutual authentication session key agreement protocol to protect peer-to-peer energy trading between edge IoT smart meters against quantum attacks. It uses CRYSTALS-Kyber to secure the authenticity and confidentiality of session keys and the Hedera Hashgraph public blockchain to provide transparency and efficient revocation. The authors report comparative analyses showing better performance and security than classical and early post-quantum alternatives.
Ali Shahidinejad, Jamshid Aghaei, and Mahdi Jalili
In briefShared Hidden Signals and Misleading Cross-Excitation in Multivariate Hawkes Models
SummaryMultivariate Hawkes models are used to measure how events excite one another, but hidden common signals can create apparent excitation with no real interaction. The paper builds stable Hawkes processes with different mechanisms that match common-signal models in mean intensity and spectrum, showing a form of second-order non-identification. It uses third-order cumulants, simulations, Whittle estimation, and a cryptocurrency application to show what can and cannot be identified.
WhyConstructs Hawkes processes matching common-signal models' second-order structure, revealing non-identification and deriving third-order identification with simulations.
In briefTrustless Bitcoin Vaults: Withdrawing Collateral Without Trusting Third Parties
SummaryExisting Bitcoin bridges require trust in a centralized party or committee to release deposited BTC. This paper proposes Trustless Bitcoin Vaults, a primitive that lets depositors or known counterparties withdraw BTC by following only the application smart contract's logic. The authors implement it for a lending application using BTC as collateral on Aave v4 on Ethereum. Mainnet experiments report deposit fees of about $0.7 and withdrawal fees of at most $7.7 at current prices and fee rates.
WhyProposes trustless BTC vaults, instantiated for Aave v4 lending with mainnet fee measurements; broad impact on Bitcoin collateral.
In briefThe Geography of Concurrent Block Proposing: An Economic Incentives Analysis
SummaryThis paper studies the economic incentives that determine where blockchain validators and proposers choose to operate, both with a single proposer and with multiple concurrent proposers. The authors model it as a game in which proposers first pick locations and then compete in a first-price auction for an opportunity whose common value depends on geographically dispersed information sources. They show that multiple concurrent proposers reward exclusive rather than shared information, creating an economic force toward geographic decentralization.
Ciamac C. Moallemi, Mallesh M. Pai, Dan Robinson et al.
In briefSolBehave: Testing Whether AI-Written Smart Contracts Actually Behave Correctly
SummaryThis paper introduces SolBehave, a benchmark that checks whether large language model-generated Solidity smart contracts behave correctly, not just whether they compile. It uses 250 real Ethereum contracts and three tests: differential fuzzing, historical transaction replay, and DCR-graph-guided synthetic transaction replay. Across seven open-weight models, compilation success ranged from 26.4% to 92.7%, but only 31.4% of compiled function bodies matched the original behavior under differential fuzzing.
Giovanni Pio Delvecchio, Stefano Fantazzini, Giacomo Frisoni et al.
In briefTree-BTE: Batched Threshold Encryption Using Additive Homomorphism
SummaryThis paper proposes a tree-based batched threshold encryption (BTE) scheme that lets a committee of servers jointly decrypt any chosen subset of ciphertexts from a large pool while keeping the rest private. It is built from two additively homomorphic components and can be instantiated under assumptions including DDH, DCR, LWR and LWE. The authors implement the DDH-based version and partially implement a post-quantum RLWE-based version, reporting computation and ciphertext-size measurements.
Amit Agarwal, Sourav Das, Babak Poorebrahim Gilkalaye et al.
