Ning Chen

The advent of 5G/6G technologies and the pervasive deployment of IoT devices are driving the emergence of demanding applications that necessitate ultra-low latency, high bandwidth, and significant computational power. Traditional cloud computing models fall short in meeting these stringent requirements. To address this, Software-Defined Edge Networks (SDENs) have emerged as a promising architecture, yet efficiently managing their heterogeneous and geographically distributed resources poses substantial challenges for optimal application provisioning. In response, this paper proposes a novel framework for intelligent task offloading, which reframes the intricate multi-component application task offloading problem as a Virtual Network Embedding (VNE) challenge within a SDEN environment. We introduce a comprehensive model where complex applications are represented as Virtual Network Requests (VNRs). In this model, each VNR consists of virtual nodes that demand specific computing and storage resources, as well as virtual links that demand specific bandwidth and must adhere to maximum tolerable delay constraints. To dynamically solve this NP-hard VNE problem in the face of stochastic VNR arrivals and dynamic network conditions, we leverage Deep Reinforcement Learning (DRL). Specifically, a Soft Actor-Critic (SAC) agent is employed at the SDN controller. This agent learns a sequential decision-making policy for mapping virtual nodes to physical edge servers and virtual links to network paths. To guide the agent towards efficient resource utilization, we define the reward for each successful embedding as the long-term revenue-to-cost ratio. By learning to maximize this reward, the agent is naturally driven to find economically viable allocation strategies. Comprehensive simulation experiments demonstrate that our SAC-based VNE approach significantly outperforms other baselines across key metrics, affirming its efficacy in dynamic SDEN environments.

Multibeam satellites (MBSs) exploit multiple spot beams and frequency reuse to achieve high spectral efficiency and flexible coverage, which are key characteristics of modern satellite communication systems. Among them, beam hopping (BH) leverages phased-array antennas to steer onboard beams and employs time-division multiplexing (TDM) to dynamically schedule illumination patterns in the time domain, with electronic beam position switching enabling rapid reconfiguration. However, existing works exhibit two critical limitations: they rely on global decision strategies that incur high inference complexity, which further increases with the number of beams and ground cells, making it difficult to balance performance and computational complexity; in pursuing high spectral reuse, they overlook beam overlap and co-frequency interference (CFI), and thus cannot effectively mitigate interference without sacrificing spectral efficiency. To address these challenges, this article proposes a cell clustering-based BH (CCBH) algorithm with low complexity and interference avoidance. Specifically, we propose a region-growing cell-clustering method based on user demand load balancing in which each beam independently serves a cell cluster, and full-frequency reuse across beams maximizes spectral efficiency. In addition, based on the centralized training and decentralized execution (CTDE) paradigm, we propose a cooperative multiagent soft actor–critic (SAC) framework with parameter sharing and centralized training, called CoopMASAC-PSCT. Among them, an SAC agent is deployed for each beam; Specifically, all actor networks share the same architecture and parameters, and the execution phase remains decentralized, with each actor making decisions solely on its local observations; To prevent intercluster interference, the shared global reward integrates system throughput, queueing delay fairness, and an interference-penalty term; Moreover, only a single global critic network is employed, which accesses the joint observations and actions of all agents during training, thus balancing individual beam performance with influence between all beams. Simulation and comparative analyses demonstrate that CCBH delivers better performance while significantly reducing inference complexity.

In the low earth orbit (LEO) mega-constellation, dynamically reconstructing resources to enhance the on-orbit efficiency and flexibility of payloads remains a critical challenge. Network slicing (NS), a core concept in 5G communications, offers a promising solution by deploying end-to-end customized and logically isolated virtual networks over shared infrastructures. However, existing methods overlook the survivability of satellite nodes and the impact of load balancing of satellites and inter-satellite links (ISLs) on subsequent slicing processes. Therefore, this paper proposes a multi-stage survivable NS (MS-SNS) method based on load balancing for LEO mega-constellations. The method adopts a hybrid heuristic-reinforcement learning (RL) framework, decomposing the problem into multiple stages. First, a multi-constraint “load coefficient” indicator is proposed for constellation target area planning and selection using a greedy algorithm. Second, a soft actor-critic (SAC)-based algorithm is proposed for slice mapping within the selected target region. Finally, a survivable NS remapping algorithm is proposed, where a multi-constraint “remapping cost” indicator is proposed to guide the selection of remappable LEO satellite nodes, ensuring service continuity during node failures. Extensive simulations demonstrate the effectiveness of MS-SNS.

To address the issues of insufficient samples, limited scene diversity, missing perspectives, and low resolution in existing UAV-based pedestrian detection datasets, this paper proposes a novel UAV-based pedestrian detection benchmark dataset named the Novel Surveillance View (NSV). This dataset encompasses diverse scenes and pedestrian information captured from multiple perspectives, and introduces an innovative data mining approach that leverages tracking and optical flow information. This approach significantly improves data acquisition efficiency while ensuring annotation quality. Furthermore, an improved pedestrian detection method is proposed to overcome the performance degradation caused by significant perspective changes in top-down UAV views. Firstly, the View-Agnostic Decomposition (VAD) module decouples features into perspective-dependent and perspective-independent branches to enhance the model’s generalization ability to perspective variations. Secondly, the Deformable Conv-BN-SiLU (DCBS) module dynamically adjusts the receptive field shape to better adapt to the geometric deformations of pedestrians. Finally, the Context-Aware Pyramid Spatial Attention (CPSA) module integrates multi-scale features with attention mechanisms to address the challenge of drastic target scale variations. The experimental results demonstrate that the proposed method improves the mean Average Precision (mAP) by 9% on the NSV dataset, thereby validating that the approach effectively enhances pedestrian detection accuracy from UAV perspectives by optimizing perspective features.