This article investigates the issue of cooperative real-time trajectory design for multiple unmanned aerial vehicles (UAVs) to support emergency communication in disaster areas. To restore communication links rapidly between mobile users (MUs) and the ground base stations, UAVs equipped with both radio frequency (RF) modules and free space optics (FSO) modules are utilized as relay nodes. Given the challenges of setting up a central controller for the UAVs and the urgency of emergency communication, the trajectory design problem for these UAVs is formulated as a distributed cooperative optimization problem. Based on the enhanced ${K}$ -mean algorithm and multiagent PPO (MAPPO) algorithm, a cooperative trajectory design method, abbreviated as KMAPPO, is proposed for the UAVs to minimize interaction overhead and optimize deployment efficiency. Compared to the state-of-the-art deep reinforcement learning (DRL) methods, simulations reveal KMAPPO’s superior performance. It converges 32% faster, boosts RF allocation efficiency, and augments FSO communication backhaul capacity.

Online video service becomes prevalent nowadays, as people are accustomed to getting information in form of videos. While home users can easily retrieve all kinds of contents onto their laptops or smartphones, vehicular users are constrained by intermittent connectivity to roadside units (RSUs). Besides, as the load of cellular infrastructure increases dramatically, it is envisioned that the cellular network can be offloaded by utilizing RSUs and vehicular relays. In this paper, we propose a cooperative downloading mechanism in heterogeneous vehicular networks that comprises vehicular ad hoc network (VANET) and cellular network. In this mechanism, RSUs act as traffic managers to fetch proper data from the Internet and then distribute to vehicles in an approximately optimal manner. Specifically, based on vehicular mobility prediction and inter-node throughput estimation, a storage time aggregated graph (STAG) is constructed for planning transmission scheme, then an iterative greedy-driven algorithm is designed for a suboptimal solution which has polynomial computational time complexity. Compared with max-throughput and min-delay cooperative downloading (MMCD), simulation results show that our approach outperforms MMCD by 5%~10% with respect to offload fraction.

Implementing either Federated learning (FL) or split learning (SL) over clients with limited computation/communication resources faces challenges on achieving delay-efficient model training. To overcome such challenges, we investigate a novel distributed C lustering-based H ybrid f E d E rated S plit l E arning ( CHEESE ) framework, consolidating distributed resources among clients by device-to-device (D2D) communications, working in an intra-serial inter-parallel manner. In CHEESE , each learning client can form a cluster with its neighboring helping clients via D2D communications to train an FL model collaboratively. Inside each cluster, the model is split into multiple segments via a model splitting and allocation (MSA) strategy, while each cluster member trains one segment. After completing intra-cluster training, a transmission client (TC) is determined from each cluster to upload a complete model to the base station for global model aggregation under allocated bandwidth. Accordingly, an overall training delay cost minimization problem is formulated, involving the following subproblems: client clustering, MSA, TC selection, and bandwidth allocation. Due to its NP-Hardness, the problem is decoupled and solved iteratively. The client clustering problem is first transformed into a distributed clustering game based on potential game theory, where each cluster further investigates the remaining three subproblems to evaluate the utility of each clustering strategy. Specifically, a heuristic algorithm is proposed to solve the MSA problem under a given clustering strategy, while a greedy-based convex optimization approach is introduced to solve the joint TC selection and bandwidth allocation problem. Extensive experiments on practical models and datasets demonstrate that CHEESE can significantly reduce training delay costs.

A cloud network schedules diverse tasks to multi-access edge computing (MEC) or cloud platforms within dynamic industrial Internet of Things (IIoT). The scheduling is influenced by the diverse intents of different parties, including the time-sensitive nature of device-generated tasks and the energy efficiency of servers. The complexity of this problem under dynamic network conditions is underscored by its nature as a Markov state transition process, typically classified as NP-hard. We introduce an intent-driven intelligent task scheduling approach (IITSA), which models a partially observable Markov decision process (POMDP) and introduces a multi-agent proximal policy optimization (MAPPO) method. We introduce a dynamic adaptive mechanism to effectively address conflicts arising from the temporal requirements and energy limitations associated with various tasks on MEC servers. This mechanism enhances the reward function of MAPPO, for which we offer comprehensive mathematical analysis to validate its convergence performance. Simulation results showcase that our proposed IITSA effectively achieves a harmonious trade-off between time-sensitive demands and infrastructure energy efficiency while exhibiting high adaptability. Compared to state-of-the-art algorithms like MADDPG and QMIX, IITSA reduces energy consumption by 11.68% and 7.07%, and enhances on-time completion numbers for time-sensitive tasks by 18.33% and 12.17%, respectively.

In this article, an integrated object detection and communication (IODC) system based on single-stage detection framework is first proposed for synthetic aperture radar (SAR) images in underwater environment. Specifically, the combination of the multiscale feature extraction module and semantic information enhance fusion module for SAR images in underwater environment is designed as the semantic encoder and the results prediction module with anchor-free detection method is designed as the semantic decoder. Considering the multiscale feature, the channel encoder composed by a multiscale fusion module and a redundancy module is designed, and the channel decoder is the inverse of the channel encoder. To adapt to the time-varying and complex wireless environment, an adaptive transmission (AT) mechanism based on attention mechanism and knowledge base is proposed for the IODC system. Moreover, considering the actual application requirements, a lightweight design for the IODC system with the AT mechanism is also conducted. The experiment results on the Sonar Common Target Detection dataset show that the proposed IODC system with AT mechanism can achieve nearly 91% average precision at 25 dB, which means the proposed system can achieve the effective integration of the objection detection and communication for SAR images in underwater environment. In the lightweight design, the model parameters of the proposed system can be reduced by up to 20%, with only a 2.22% sacrifice in performance.

The sixth generation (6G) mobile communication system is expected to provide global seamless network coverages, where a satellite-ground integrated network (SGIN) is seen as one of the typical 6G networking paradigms. In this paper, a dynamic virtual network function (VNF) selection and chaining (DVSC) problem in an SGIN is investigated. We aim to balance the network resource provisioning and VNF migration costs with service performance gain to maximize the long-term network profit. Specifically, we formulate the DVSC problem as a Markov decision process (MDP), by taking into consideration the heterogeneity and time-varying nature of SGINs. A novel VNF selection and chaining scheme is proposed, where a deep Q-learning (DQL) algorithm is designed to dynamically determine a set of VNF selection and chaining policies (VSCPs) based on the evolving network states (e.g., network resources, network topology, and network traffic load). Furthermore, to elaborate the level of computing resource sharing of VSCP sets, a new sharing ratio (SR) is proposed. To efficiently allocate heterogeneous network resources, the action space is built by clustering the historical records of the network load and selecting the VSCP set for each cluster in a greedy manner. Extensive simulation results are presented to demonstrate the effectiveness of the proposed framework in comparison with the state-of-the-art schemes