QoS Aware Task Offloading for Digital-Twin Enabled Connected Vehicular Network

Future autonomous driving would require a plethora of networking resources and high-quality vehicular communication links. As these mobile devices are resource-constrained, thus, offloading computer-intensive operations to the mobile edge servers has been a crucial strategy to meet the stringent quality of service (QoS) requirements. However, various types of smart vehicles with distinct capabilities, dynamic demands, and unpredictable topology pose significant challenges in realizing efficient edge computing services. To cope with these challenges, we incorporate a digital twin (DT)-enabled con-nected vehicular network which can facilitate automated resource management to enhance QoS of vehicular users. Specifically, we aim to allocate computation and spectrum resources for a DT-enabled vehicular network along with designing an optimal offloading policy. In addition, a problem to optimize the vehicular network's utility-cost ratio (UCR) is formulated in order to reduce the overall latency, energy consumption, and to improve the throughput. Further, to obtain an optimal resource allocation framework, we propose a three-step iterative algorithm to optimize the bandwidth, prices for latency, energy consumption, and offloading factor iteratively until convergence. The simulations demonstrate that the suggested approach efficiently reduces the task offloading latency as well as decreases the average energy usage for vehicular network, in addition to improving the throughput. Specifically, the utility-to-cost ratio decreases by 85 % and 12.5 % for sparse and dense traffic scenarios, respectively for a vehicle density of 0.137 vehicle/m.