Platoon assisted vehicular edge computing has been envisioned as a promising paradigm of implementing offloading services through platoon cooperation.In a platoon,a vehicle could play as a requester that employs anoth...Platoon assisted vehicular edge computing has been envisioned as a promising paradigm of implementing offloading services through platoon cooperation.In a platoon,a vehicle could play as a requester that employs another vehicles as performers for workload processing.An incentive mechanism is necessitated to stimulate the performers and enable decentralized decision making,which avoids the information collection from the performers and preserves their privacy.We model the interactions among the requester(leader)and multiple performers(followers)as a Stackelberg game.The requester incentivizes the performers to accept the workloads.We derive the Stackelberg equilibrium under complete information.Furthermore,deep reinforcement learning is proposed to tackle the incentive problem while keeping the performers’information private.Each game player becomes an agent that learns the optimal strategy by referring to the historical strategies of the others.Finally,numerical results are provided to demonstrate the effectiveness and efficiency of our scheme.展开更多
基金supported in part by National Key R&D Program of China under Grant 2020YFB1807802in part by National Natural Science Foundation of China under Grants 62001125 and 61971148+4 种基金in part by FDCT-MOST Joint Project under Grant 0066/2019/AMJin part by Science and Technology Development Fund of Macao SAR under Grant 0162/2019/A3in part by FDCT SKL-IOTSC(UM)2021-2023in part by Research Grant of University of Macao under Grant MYRG2020-00107-IOTSCin part by Guangdong Basic and Applied Basic Research Foundation under Grant 2022A1515011287
文摘Platoon assisted vehicular edge computing has been envisioned as a promising paradigm of implementing offloading services through platoon cooperation.In a platoon,a vehicle could play as a requester that employs another vehicles as performers for workload processing.An incentive mechanism is necessitated to stimulate the performers and enable decentralized decision making,which avoids the information collection from the performers and preserves their privacy.We model the interactions among the requester(leader)and multiple performers(followers)as a Stackelberg game.The requester incentivizes the performers to accept the workloads.We derive the Stackelberg equilibrium under complete information.Furthermore,deep reinforcement learning is proposed to tackle the incentive problem while keeping the performers’information private.Each game player becomes an agent that learns the optimal strategy by referring to the historical strategies of the others.Finally,numerical results are provided to demonstrate the effectiveness and efficiency of our scheme.