Adaptive Task Offloading in Vehicular Edge Computing Networks Based on Deep Reinforcement Learning

Edge Computing Task Offloading of Internet of Vehicles Based on Improved MADDPG Algorithm

“…To deliver better services to nearby users and achieve seamless real-time responses to mobile user service requests, Wang S proposed an offline microservice coordination algorithm based on dynamic programming [7]. Shuai suggested an adaptive network routing technique to find a balance between modeling accuracy and optimization efficiency in MEC systems with time-varying connection delays [8]. In heterogeneous vehicle networks with many random jobs, time-varying radio channels, and dynamic bandwidth, Ke proposed an adaptive depth-reinforcement-based computational offloading approach [9].…”

DRL-assisted delay optimized task offloading in automotive-industry 5.0 based VECNs

“…RL/DRL has dramatically anchored state-of-the-art performance in various vehicular environments, particularly the task offloading [3]. Considering the dynamic nature of the vehicular environment, Shuai et al [19] provided a delay optimization scheme. Initially, the transmission latency is minimized using the optimal flow-based routing algorithm, then a deep Q-learning based task offloading strategy selection scheme is used for adaptive task offloading considering the MEC load states.…”

“…The action value function q π ps, aq is the expected accumulated reward after a chooses π on s, given in (18). Both functions exhibit how decent a state and state-action pair is, their relationship is given in (19).…”