Vehicular fog computing (VFC) enhances compute-intensive task processing by exploiting idle vehicle resources. However, existing offloading mechanisms may fail due to dynamic factors, such as vehicle mobility, unstable links, and service overload. This paper proposes an offloading-failure-aware (OFA) task offloading scheme (OFA-offloading). Although the exact offloading failure probability is difficult to obtain, it is determined by the service capability of the selected service vehicle (SV). Thus a new tractable metric, i.e., vehicle service capability (VSC), is defined to reflect the offloading failure probability, which is a function of vehicle mobility, resource availability, and link status. Based on VSC of each SV and considering that delay is important for VFC networks, an OFA delay utility is designed. Aiming to maximize this utility, a joint offloading SVs selection and computing resource allocation optimization problem is formulated. Since it is NP-hard and the VFC network is highly dynamic, a novel Graph Neural Network based federated Advantage Actor-Critic (GNN-FAC) algorithm is proposed to solve the problem. GNN-FAC can proactively predict environmental dynamics and incorporate VSC as a critical criterion for offloading decisions. Simulation results demonstrate that compared with existing offloading algorithms, OFA-offloading can improve the OFA delay utility by up to 40%.
Yihao Wu, Yanli Qi, Yiqing Zhou et al.· IEEE Transactions on Network...· 0 citations
Mobile Edge Computing (MEC) is promising to enable low delay services with which users can offload computing intensive and delay sensitive tasks to the edge. Considering a multi-cell MEC (MC-MEC) network without sufficient resources to serve all users, user selection and non-orthogonal multiple access (NOMA) should be introduced. Then, to maximize the delay-aware average user service satisfaction degree (DA-AveUSD), user selection and resource allocation are jointly optimized (DA-JUSRA), which is modeled as a mixed integer nonlinear programming (MINLP) problem and proven to be NP-hard. To solve this problem, it is decomposed into two independent subproblems, i.e., the power allocation (PA) problem and the user selection, subchannel scheduling and computing resource allocation (USC) problem. Next, a convex evolutionary alternating optimization (CEAO) algorithm is proposed, which alternately applies the convex optimization method and the Karush-Kuhn-Tucker (KKT)-embedding enhanced elite genetic algorithm (KKT-embedding E2GA) to solve the PA and the USC problem, respectively. Simulations show that compared to the optimal exhaustive search algorithm, the proposed CEAO algorithm converges rapidly within a few iterations, with a gap in DA-AveUSD of less than 1% to the optimum performance. Next, compared to existing user selection schemes, DA-JUSRA with CEAO can enhance DA-AveUSD by more than 50% and yield a higher optimal load.
Ningzhe Shi, Yiqing Zhou, Ling Liu et al.· IEEE Transactions on Mobile...· 1 citation