With the growing number of decentralized finance (DeFi) applications, transaction fairness in blockchains has gained much research interest. As a broad concept in distributed systems and blockchains, fairness has been used in different contexts, varying from ones related to the liveness of the system to ones that focus on the received order of transactions. In this work, we revisit the fairness definitions and find that existing fairness definitions are not adapted to blockchains with multiple DApps. We then provide a more generic one calledverifiable fairness. Compared with prior definitions, our notion has two unique features: (i) it relaxes the ordering rules to apredicate; (ii) it enables users to independently verify if their transactions comply with the predicate for concrete applications. We also provide a scheme that achieves verifiable fairness, leveraging trusted hardware. Unlike prior works that usually design a dedicated consensus protocol to achieve fairness, our scheme can be integrated with any blockchain system. Our evaluation results on Amazon EC2 using up to 120 instances across different regions show that our construction imposes only minimal overhead on existing blockchain systems.

Secure multi-party computation (MPC) allows joint computations on sensitive data while guaranteeing privacy and correctness. In recent years, a series of MPC protocols assisted by trusted execution environments (TEEs) have been proposed to reduce overhead brought by costly cryptographic techniques. However, existing protocols either generally assume consistent trust in TEEs among all participating parties, or require dedicated designs for different applications. This prevents the protocols from being deployed in practice. To address these challenges, in this work, we propose a generic MPC protocol without assuming consistent trust in TEEs while fully utilizing heterogeneous TEEs to improve efficiency. To this end, we propose a security model to capture parties' inconsistent trust in TEEs and prove the security of our protocol under a simpler variant of the UC framework (SUC framework). In addition, we instantiate our protocol for secure aggregation based on a state-of-the-art information-theoretically secure protocol SwiftAgg+. Evaluation results among 64 parties deployed on Azure virtual machines show that our protocol reduces the running time of SwiftAgg+ by 66%. The running time of parties in our protocol is reduced by at most 91% compared to that required in SwiftAgg+.

Multivariate time series forecasting with hierarchical structure is widely used in real-world applications, e.g., sales predictions for the geographical hierarchy formed by cities, states, and countries. The hierarchical time series (HTS) forecasting includes two sub-tasks, i.e., forecasting and reconciliation. In the previous works, hierarchical information is only integrated in the reconciliation step to maintain coherency, but not in forecasting step for accuracy improvement. In this paper, we propose two novel tree-based feature integration mechanisms, i.e., top-down convolution and bottom-up attention to leverage the information of the hierarchical structure to improve the forecasting performance. Moreover, unlike most previous reconciliation methods which either rely on strong assumptions or focus on coherent constraints only, we utilize deep neural optimization networks, which not only achieve coherency without any assumptions, but also allow more flexible and realistic constraints to achieve task-based targets, e.g., lower under-estimation penalty and meaningful decision-making loss to facilitate the subsequent downstream tasks. Experiments on real-world datasets demonstrate that our tree-based feature integration mechanism achieves superior performances on hierarchical forecasting tasks compared to the state-of-the-art methods, and our neural optimization networks can be applied to real-world tasks effectively without any additional effort under coherence and task-based constraints.