Mobile Edge Computing (MEC) is a promising technology to extend the diverse services to the edge of Internet of Things (IoT) system. However, the static edge server deployment may cause “service hole” in IoT networks in which the location and service requests of the User Equipments (UEs) may be dynamically changing. In this paper, we firstly explore a vehicle edge computing network architecture in which the vehicles can act as the mobile edge servers to provide computation services for nearby UEs. Then, we propose as vehicle-assisted offloading scheme for UEs while considering the delay of the computation task. Accordingly, an optimization problem is formulated to maximize the long-term utility of the vehicle edge computing network. Considering the stochastic vehicle traffic, dynamic computation requests and time-varying communication conditions, the problem is further formulated as a semi-Markov process and two reinforcement learning methods: Q-learning based method and deep reinforcement learning (DRL) method, are proposed to obtain the optimal policies of computation offloading and resource allocation. Finally, we analyze the effectiveness of the proposed scheme in the vehicular edge computing network by giving numerical results.

The lack of the computation services in remote areas motivates power Internet of Things (IoT) to apply unmanned aerial vehicle (UAV)-enabled mobile edge computing (MEC) technology. However, the computation services will be significantly affected by the UAVs' capacities and distinct power IoT applications. In this paper, we firstly propose a cooperative UAV-enabled MEC network structure in which the UAVs are able to help other UAVs to execute the computation tasks. Then, a cooperative computation offloading scheme is presented while considering the interference mitigation from UAVs to devices. To maximize the long-term utility of the proposed UAV-enabled MEC network, an optimization problem is formulated to obtain the optimal computation offloading decisions and resource management policies. Considering the random devices' demands and time-varying communication channels, the problem is further formulated as a semi-Markov process and the deep reinforcement learning based algorithms are proposed in both of the centralized and distributed UAV-enabled MEC networks. Finally, we evaluate the performance of the proposed DRL-based schemes in the UAV-enabled MEC framework by giving numerical results.

This paper proposes a combinatorial auction approach for multi-resource allocation of an energy storage (ES) shared by multiple electricity end users in a residential community. Through the auction, a user buys a group of ES resources, including capacity, energy, charging power, and discharging power, from the ES operator. With the ES resources, users store grid energy during low-electricity-price hours, so that they can consume the cheap stored energy during high-electricity-price hours to reduce their electricity bills. In the auction, users submit their resource demands and corresponding bid prices, based on which the ES operator determines the winners and the final payments that the winners must pay. To solve the NP-hard winner determination problem, a fully polynomial time approximation scheme (FPTAS) is developed, which can optimize social welfare but may violate resource supply constraints. To deal with the constraint violation, the ES operator may buy extra energy outside the system to meet the winners’ actual demands. Further, a distributed implementation of the auction is designed to offload the auction computation onto the users while preventing the users from manipulating the auction outcomes in the course of computation. The proposed distributed auction can ensure that all users faithfully complete the assigned computation tasks in an ex-post Nash equilibrium. A real time-of-use (TOU) electricity tariff and actual home load data are used in the simulation, in which the proposed auction approach is evaluated in terms of social welfare and computational efficiency.

Space-air-ground integrated edge computing is expecting to provide pervasive computation services for Internet of Things, especially in remote areas. However, the offloading process of power-limited IoT devices is a challenge issue due to unreliable communications in aerial environment. In this paper, we propose an energy efficient space-air-ground integrated edge computing network architecture, in which the IoT devices choose the most appropriate LEO satellites or unmanned aerial vehicles (UAVs) for task offloading according to their energy level, communication conditions and computing capabilities. In order to providing efficient task offloading and energy saving policy under uncertainty aerial environment, a constrained Markov decision process is employed to formulate the task offloading decision problem and a deep reinforcement learning (DRL)-based algorithm is devised to solve the proposed problem. An adaptive federated DRL-based offloading method is further proposed to find sub-optimal offloading decisions by considering the privacy protection and communication failure in the proposed network. Numerical results confirm the effectiveness of the proposed schemes on energy saving and computation efficiency.

In 6G systems, space-air-ground integrated network (SAGIN), relying on space/aerial communications to complement terrestrial networks, is developed to achieve worldwide connectivity and multi-service access especially in remote areas. However, the heavy workload of the resource-limited satellites may raise an important issue about reducing service coverage and quality. In this paper, we propose a collaborative edge computing framework for SAGIN-aided 6G system, in which the LEO satellites are considered as both “servers” and “users”. Excepting provide services for ground users/devices, the LEO satellites is able to offload the tasks to nearby aircrafts via one-hop link or offload them to the cloud server along multi-hop satellite path. To minimize the long-term task completion delay of LEO satellites, a stochastic optimization problem is formulated by considering the variation of the space/aerial environment. The Lyapunovbased optimization method is developed to solve the problem and the delayed online learning technique is adapted to predict dynamic task arrival and queue length of satellites and aircrafts. Numerical results confirm the effectiveness of the proposed collaborative offloading scheme for reducing tasks completion delay of LEO satellites while guaranteeing computation efficiency.