Integrating mobile edge computing (MEC) into the Internet of Vehicles (IoV) allows IoV devices with limited computation capabilities and energy to offload their computation-intensive tasks to the network edge, thereby providing high-quality service to vehicles. Due to the flexibility of unmanned aerial vehicles (UAVs) and the capability of non-orthogonal multiple access (NOMA) to support massive connectivity, a NOMA-based and UAV-assisted MEC framework offers flexible and high-performance computing services for vehicles in large-scale access networks. This paper introduces a novel joint optimization framework that incorporates Age of Information (AoI) awareness to enhance task freshness in IoV networks. By leveraging the AoI metric, the optimization problem is formulated to simultaneously minimize total energy consumption and maximize the number of completed tasks. The proposed solution achieves these objectives by jointly optimizing vehicle wireless transmission power using the NOMA communication technique, task offloading decisions, and the UAV's trajectory planning and computing frequency allocation across time slots. Given the non-convex nature of the problem within a hybrid action space, we employ the Twin Delayed Deep Deterministic (TD3) policy gradient algorithm to address it. Experimental results indicate that the TD3 policy gradient algorithm is effective in solving the proposed optimization framework, particularly in terms of convergence speed and the system reward.

The industrial internet of things (IIoT) system based on the multi-access edge computing (MEC) network architecture can significantly enhance industrial production efficiency and drive the advancement of smart manufacturing. However, such a system faces uncertain environmental factors, including dynamic changes in channel conditions and the random generation of tasks. Motivated by these challenges, this paper investigates the problem of task division and offloading decisions for delay-sensitive tasks with prioritization attributes from the perspectives of low latency and high value, and proposes a twin delayed deep deterministic based multi-prioritized task division scheduling (TD3-MPTDS) strategy in the device to device (D2D)-assisted MEC network. This strategy not only divides tasks into smaller chunks, enabling finer-grained scheduling of the entire system, but also intelligently offloads tasks to either the D2D network or the MEC server while considering the overall device load. In addition, the proposed strategy is tailored to optimize the task queue of the MEC server. By considering factors such as priority, waiting time, and expected time of completion, queue adjustments are dynamically made at each time slot. Simulation experiments validate that our proposed strategy quickly converges and outperforms the benchmark strategies in terms of task completion delay and completion value.

The 6G network is expected to accommodate a wide array of connected devices, supporting diverse services from any location at any time. In this article, we introduce an aerial mobile edge computing (MEC) framework composed of high-altitude platforms (HAPs) and low-altitude unmanned aerial vehicles (UAVs), to cater to computing offloading for Internet of Things (IoT) devices, particularly in rural/remote areas or disaster zones. The framework accommodates various types of tasks, each computed by the corresponding Docker container. The objective is to achieve optimal workload fairness for UAVs while simultaneously minimizing the weighted processing costs among IoT devices in terms of task computation latency and energy consumption over the long term. This is achieved by jointly optimizing the flight trajectories and Docker image caching decisions of the UAVs with limited storage capacities, alongside ensuring service fairness for IoT devices. We tailor a multiagent deep deterministic policy gradient (MADDPG)-based approach to solve the long-term joint optimization problem, normalizing continuous actions and sampling discrete actions by generalizing the Gumbel-Softmax reparameterization trick. Experimental results indicate that our approach significantly outperforms benchmark schemes in terms of processing delay, energy consumption, and fairness.