Key points are not available for this paper at this time.
Vehicular edge computing (VEC) is a new paradigm in smart cities with the potential to provide and manage resources such as computing and storage closer to resource-constrained smart vehicles (SV) to support ultra-reliable low-latency communications and knowledge sharing. However, it is challenging to make optimal resource allocation and offloading decisions due to the mobility, ubiquitous communications, and diverse resource demands of SVs and the dynamic nature of the vehicular network topology. In this paper, we propose an adaptive resource allocation and task offloading scheme for HAPS-assisted Internet of Vehicle networks by exploiting the potential of digital twins (DT), blockchain, and federated multi-agent deep reinforcement learning (FMADRL) technologies to address these issues. The DT network is utilized to intelligently monitor and control the demand and supply of resources in the digital representation of the physical operating environment. We adopt a dynamic pricing-based double auction to model the supplies and demands of resource providers and requesters. This enables resource providers and SVs to make adaptive and optimal resource allocation and task offloading decisions. In addition, we deploy a consortium blockchain to enable distributed and secure resource allocation. The resource allocation and task offloading multi-objective optimization problem is formulated as a multi-agent extension of the Markov decision process and solved using an FMADRL-based multi-agent deep deterministic policy gradient (FMADDPG) algorithm. The numerical results show that the proposed scheme archives efficient resource allocation and maximizes the utility function while minimizing costs compared to the baseline schemes.
Abishu et al. (Tue,) studied this question.