Satellite edge computing, as an extension of ground edge computing, is a key technology for achieving seamless global computing coverage. However, the low earth orbit (LEO) satellites have limited computing resources and are moving at a high speed. This naturally poses a challenge to find more suitable computation offloading strategies with minimum network latency and energy consumption, especially when a large number of co-existing users are to offload their tasks. In this paper, therefore, we mainly focus on computation offloading in the satellite edge computing network (SECN) by jointly considering LEO satellites’ mobility and SECN's heterogeneous resource constraints to explore more practical computation offloading strategies. We first formulate the problem of M obility-aware C omputation O ffloading (MCO) in the SECN via specifying the effect of LEO satellites’ high-speed movement on the computation offloading, aiming to minimize the network latency and energy consumption. Considering the MCO problem is discrete and non-convex as the objective function and constraints are associated with the binary decision variables. We then convert the original non-convex problem into a continuous convex problem which is proved to be feasible. To avoid a high computational complexity incurred by the extensive co-existing user offloading, we design MCO-A , a distributed algorithm based on ADMM (alternating direction method of multipliers) to solve the MCO problem efficiently. Finally, the performance of MCO-A is evaluated via extensive experiments including small-scale and large-scale scenarios. The experimental results show that MCO-A can achieve a lower network latency and energy consumption in an efficient way compared with the baseline and state-of-the-art approaches.

Multiple cooperative task allocation (MCTA) is a crucial problem in mobile crowdsensing, where each task requires more than one user to cooperatively complete. As more and more users join sensing tasks in groups, it is indispensable to develop a group-oriented crowdsensing mechanism supporting MCTA. However, existing studies generally focus on a group that can provide sufficient users to accomplish a task. Once these groups no longer exist, the corresponding task will be discarded or be performed with compromised quality. In this paper, we propose a novel three-phase approach named Group-oriented Cooperative Crowdsensing (GoCC) to tackle the MCTA problem in social mobile crowdsensing. This approach exploits real-life relationships in the social network to form compatible groups, which improves the task coverage via group-oriented cooperation while achieving good task cooperation quality. Specifically, phase 1 selects a subset of users on the social network as initial leaders and directly pushes sensing tasks to them. Phase 2 utilizes the leaders to search for their socially connected users to model groups. Phase 3 presents the process of group-oriented task allocation for solving the MCTA problem. Experiments on the real-world dataset validate that our approach significantly outperforms the representative approaches.

As the demand for global computing coverage continues to surge, satellite edge computing emerges as a pivotal technology for the next generation of networks. Unlike ground-based edge computing, Low Earth Orbit (LEO) satellites face distinctive challenges, including high-speed mobility and resource limitations, etc. Therefore, effectively utilizing LEO satellites for global coverage services is crucial but challenging due to their dynamic coverage areas and diverse task requirements. To address these challenges, we introduce a novel dual-cloud edge collaborative task offloading architecture in the satellite network-assisted edge computing environment, namely, Satellite-Ground Task Offloading (SGTO). The architecture employs a Geostationary Earth Orbit (GEO) satellite and a ground cloud computing center as satellite cloud and ground cloud, respectively, and LEO satellites as edge nodes. We formally define the task offloading problem in the SGTO with the aim of minimizing the average latency and average energy consumption. We then propose an adaptive approach named SGTO-A from the perspective of satellites to adaptively solve the problem leveraging deep reinforcement learning. Specifically, we transform the task offloading problem into a Markov decision process and adopt the generalized proximal policy optimization (GePPO) algorithm to solve the problem. Finally, experimental results demonstrate that SGTO architecture and SGTO-A outperform the representative approaches in terms of average latency, average energy consumption and running time.

Mobile edge computing (MEC), as an emerging technology, allows application vendors to deploy application instances on edge servers to deliver low-latency services to nearby end-users. However, due to hardware faults, software exceptions, or cyberattacks, edge servers are prone to failures in the highly distributed and dynamic MEC environment. Hence service reliability must be ensured when failures occur. This raises a critical and open problem - improving service reliability when deploying application instances in the MEC environment. In this article, we jointly consider both user coverage and service reliability when deploying application instances on edge servers with a given application deployment budget $\mathcal {K}$ . We formally define this joint C overage- R eliability for $\mathcal {K}$ - B udgeted E dge A pplication D eployment ( CR-BEAD ) problem and model it as a constrained optimization problem. Next, we propose an optimal approach (named BEAD-O ) based on integer programming to find optimal solutions to small-scale CR-BEAD problems. We also propose a greedy approach named BEAD-G with a constant approximation ratio of $1 - 1/e$ to solve large-scale CR-BEAD problems efficiently. Extensive experimental evaluation against three representative approaches illustrates the effectiveness and efficiency of our approaches.

Mobile-edge computing (MEC), as an emerging computing paradigm, allows app vendors to deploy their mobile and/or IoT applications on edge servers to deliver low-latency services to their app users. However, when an edge server needs to serve excessive app users concurrently, severe interference is incurred, which immediately reduces app users’ achievable data rates and, consequently, impacts their perceived service quality. This is a major challenge to the app vendor’s attempt to minimize the edge resources required for serving its app users with a satisfactory service quality. To tackle this challenge, in this article, we present and formulate this multiple edge application deployment (MEAD) problem in the MEC environment, aiming to maximize app users’ overall service quality at minimum deployment cost, considering application shareability and communication interference. We prove that the MEAD problem is $\mathcal {NP}$ -hard. Then, we propose a heuristic approach, namely, the deployment-priority greedy via the divide-and-conquer strategy (DPG-D&C), to solve the MEAD problem effectively and efficiently. We evaluate our approach extensively by using a widely used real-world data set. The experimental results show that DPG-D&C significantly outperforms state-of-the-art approaches.

Satellite edge computing, as an extension of ground edge computing, is a key technology for providing computing services by deploying resources on low earth orbit (LEO) satellites. However, the temporal and spatial differences in population density and economic levels may lead to unbalanced computation workloads on LEO satellites. Considering the mobility and limited resources inherent in LEO satellites, effectively utilizing the LEO satellite network to meet global competitive demands for task offloading becomes challenging. Therefore, in this paper, we propose an adaptive task offloading approach with spatiotemporal load awareness, named ATO-SLA , in satellite edge computing, aiming to optimize users' perceived-delay and energy consumption. Specifically, to avoid LEO satellite overload, we first introduce the spatiotemporal load factor, formally modelling the spatiotemporal load-aware task offloading problem. Then, the Markov decision process is employed to structure the task offloading decision process. Afterwards, we propose a task offloading algorithm based on proximal policy optimization strategy to adaptively solve the problem. Finally, experimental results demonstrate that ATO-SLA achieves a lower average delay and average energy consumption compared with other approaches.