Package delivery via ridesharing provides appealing benefits of lower delivery cost and efficient vehicle usage.Most existing ridesharing systems operate the matching of ridesharing in a centralized manner,which may r...Package delivery via ridesharing provides appealing benefits of lower delivery cost and efficient vehicle usage.Most existing ridesharing systems operate the matching of ridesharing in a centralized manner,which may result in the single point of failure once the controller breaks down or is under attack.To tackle such problems,our goal in this paper is to develop a blockchain-based package delivery ridesharing system,where decentralization is adopted to remove intermediaries and direct transactions between the providers and the requestors are allowed.To complete the matching process under decentralized structure,an Event-Triggered Distributed Deep Reinforcement Learning(ETDDRL)algorithm is proposed to generate/update the real-time ridesharing orders for the new coming ridesharing requests from a local view.Simulation results reveal the vast potential of the ETDDRL matching algorithm under the blockchain framework for the promotion of the ridesharing profits.Finally,we develop an application for Android-based terminals to verify the ETDDRL matching algorithm.展开更多
基金supported by National Natural Science Foundation of China(Grant No.62271073 and 61971066)Beijing Natural Science Foundation(L212003)the National Youth Top-notch Talent Support Program.
文摘Package delivery via ridesharing provides appealing benefits of lower delivery cost and efficient vehicle usage.Most existing ridesharing systems operate the matching of ridesharing in a centralized manner,which may result in the single point of failure once the controller breaks down or is under attack.To tackle such problems,our goal in this paper is to develop a blockchain-based package delivery ridesharing system,where decentralization is adopted to remove intermediaries and direct transactions between the providers and the requestors are allowed.To complete the matching process under decentralized structure,an Event-Triggered Distributed Deep Reinforcement Learning(ETDDRL)algorithm is proposed to generate/update the real-time ridesharing orders for the new coming ridesharing requests from a local view.Simulation results reveal the vast potential of the ETDDRL matching algorithm under the blockchain framework for the promotion of the ridesharing profits.Finally,we develop an application for Android-based terminals to verify the ETDDRL matching algorithm.
