To satisfy component concentration constraints in crude oil operations, it is necessary to blend different oil types, resulting in a mixed integer nonlinear programming (MINLP) formulation for the scheduling problem of crude oil operations. Because of the intractability of such a nonlinear problem, approximate methods were proposed in the literature. However, by the existing methods, a composition concentration discrepancy may occur, leading to an infeasible solution; or a feasible solution cannot be found even if such a solution exists for some cases. Based on a priority-slot modeling method, this paper copes with the crude-oil scheduling problem suffering from composition concentration discrepancy. To find a solution without composition concentration discrepancy, a valid inequality is added to the MINLP model. Also, the model size is significantly reduced by properly determining the number of slots. Then, a novel solution method is proposed. By this method, the problem is iteratively solved and, at each iteration step, only a reduced MILP problem is solved. Consequently, a solution can be found such that the composition concentration discrepancy is completely eliminated and it is computationally more efficient than the existing ones. Experiments are done to test the performance of the proposed method. Results show that the proposed method outperforms the existing ones.

The architecture of cloud–edge collaboration can improve the efficiency of Internet of Things (IoT) systems. Recent studies have pointed out that using IoT terminal devices as destinations for computing offloading can promote further optimized allocation of computational resources. However, in practice, this idea encounters the problem that participants might lack the motivation to take over computational tasks from others. Although the edge and the terminal are provided with symmetrical positions in collaborative offloading, their computational resources and capabilities are asymmetric. To mitigate this issue, this paper designs a distributed strategy for the trading of computational resources. The most prominent feature of our strategy is its multi-preference optimization objective that takes into account the overall satisfaction with task delay, energy cost, trading prices, and user reputation of participants. In addition, this paper proposes a system architecture based on the Blockchain-as-a-Service (BaaS) design to give full play to the good distributed technology features of blockchain, such as decentralization, traceability, immutability, and automation. Meanwhile, BaaS delivers decentralized identifier (DID) based distributed identity infrastructure for the distributed computational resource trading stakeholders as well. In the simulation evaluation, we compare our trading strategy based on a matching mechanism called multi-preference matching (MPM) to trading using the classical double auction (DA) matching mechanism. The results show that our computational resource trading strategy is able to offload and execut more tasks, achieving a better throughput compared to the DA-based strategy.

No abstract

No abstract