Unmanned Aerial Vehicles (UAVs), as key enablers of 6G-enabled Internet of Things (IoT) ecosystems, facilitate dynamic aerial coverage, seamless edge intelligence, and adaptive routing. However, despite these advantages, the reliability and trustworthiness of UAV swarms in 6G remain critical concerns due to rising quantum threats and limitations of traditional machine learning approaches. The paper presents QuaRTA-6G, a Quantum-Resilient, Trust-aware, and Accountable framework that provides a unified solution for 6G UAV swarms. QuaRTA-6G achieves security, trust, and efficiency by seamlessly combining post-quantum cryptography for secure communication, a decentralized ledger for identity and trust management, and Variational Quantum Federated Learning (VQFL) for efficient swarm intelligence. For quantum-resistant authentication and immutable UAV identity verification, QuaRTA-6G leverages CRYSTALS-Kyber with blockchain. Through simulations, QuaRTA-6G achieves secure authentication handshakes in under 2 ms and scales robustly to swarms of more than 100 UAVs, keeping high integrity even under 20% packet loss. Experiments demonstrate that, even under strong data poisoning attacks, the framework achieves 92% accuracy and 85% mission success rate, while comprehensive resource analysis confirms its feasibility across both standard and resource-constrained UAV platforms. Furthermore, an ablation study demonstrates that each module of QuaRTA-6G is essential for ensuring a responsible and trustworthy 6G UAV framework.

IoT networks often face conflicting routing goals such as maximizing packet delivery, minimizing delay, and conserving limited battery energy. These priorities can also change dynamically: for example, an emergency alert requires high reliability, while routine monitoring prioritizes energy efficiency to prolong network lifetime. Existing works, including many deep reinforcement learning approaches, are typically centralized and assume static objectives, making them slow to adapt when preferences shift. We propose a dynamic and fully distributed multi-objective Q-learning routing algorithm that learns multiple per-preference Q-tables in parallel and introduces a novel greedy interpolation policy to act near-optimally for unseen preferences. The algorithm learns to optimize for energy efficiency, packet delivery ratio, and the composite reward, adapting to changing trade-offs between these metrics without retraining or centralized control. A theoretical analysis further shows that the optimal value function is Lipschitz-continuous in the preference parameter, ensuring that proposed greedy interpolation policy yields provably near-optimal behavior. Simulation results show that our approach adapts in real time to shifting priorities and achieves up to 80–90% lower energy consumption and up to 5 × higher cumulative rewards and packet delivery compared to six baseline protocols, under dynamic and distributed settings. Sensitivity analysis across varying preference window lengths confirms that the proposed DPQ framework consistently achieves higher composite reward than all baseline methods, demonstrating robustness to changes in operating conditions.

Deep Neural Networks (DNNs) are increasingly adopted across various industries, driving the demand for deploying their capabilities on mobile devices. However, current mobile inference frameworks often rely on a single processor to execute each model inference, limiting hardware utilization and leading to suboptimal performance and energy efficiency. Expanding DNN accessibility on mobile platforms requires more adaptive and resource-efficient solutions to meet increasing computational demands without compromising device functionality. Nevertheless, performing parallel inference of multiple DNNs on heterogeneous processors remains a significant challenge. Existing studies have explored partitioning DNN operations into subgraphs to enable parallel execution across heterogeneous processors. However, these approaches typically generate excessive subgraphs based solely on hardware compatibility, increasing scheduling complexity and memory management overhead. To address these limitations, we propose the Advanced Multi-DNN Model Scheduling (ADMS) strategy that optimizes multi-DNN inference across heterogeneous processors on mobile devices. ADMS constructs an offline subgraph partitioning strategy that considers both hardware support for operations and scheduling granularity. It also employs a processor-state-aware scheduling algorithm to dynamically balance workloads based on real-time system conditions. This ensures efficient workload distribution and maximizes the utilization of available processors. Experimental results demonstrate that, compared to vanilla inference frameworks, ADMS achieves a 4.04× reduction in multi-DNN inference latency.

Mobile devices (MDs) are increasingly incorporating deep neural network (DNN) inference into their systems due to the rapid growth of intelligent applications. Mobile edge computing-based distributed DNN collaborative inference has gained popularity due to limited on-device computation and energy budgets. However, the resource competition among MDs, along with the coupling of collaborative inference tasks across MDs and servers, creates significant challenges for efficient resource management. This issue is further exacerbated by the complexity of directed acyclic graph (DAG)-structured DNNs. Most prior studies do not jointly address the dual challenges of partitioning complex-structured DNNs and leveraging advanced optimization for collaborative inference, and their resilience to channel condition fluctuations remains underexplored. To address these challenges, we propose a novel task-aware collaborative inference framework. First, we devise a fine-grained partitioning point search algorithm based on a bidirectional graph linked list, which enables one-dimensional and flexible partitioning of DAG-structured DNNs. We then reformulate the problem of minimizing collaborative inference energy consumption and latency as a task-aware Markov decision process (MDP), which partitions each user's inference task queue into consecutive task windows for resource allocation. Building on this, we propose an Embedded Multi-Agent Hybrid Proximal Policy Optimization (EMH-PPO) algorithm to learn effective policies. Extensive experiments conducted across diverse network scenarios reveal that, compared to local DNN inference on MDs, our proposed method reduces inference latency by up to 64% and energy consumption by up to 46%.

In Wireless Sensor Networks (WSNs), relay sensor nodes can aggregate data from edge sensor node into a summary information before sending to the sink. Due to the vast number of sensor nodes in a distributed edge computing (DEC) network, these relay sensor nodes may receive a high number of aggregation requests. This increases the chance of conflicting transmissions, which further leads to unwanted latency. Designing a conflict-free and minimal latency data aggregation schedule remains an open question. Moreover, existing related works have been conducted in traditional WSNs. By leveraging multiple antennas, the Multiple Input Multiple Output (MIMO) and cooperative MIMO called virtual MIMO (V-MIMO) enable broadband wireless communication, thereby improving the performance of WSNs. However, compared with traditional WSNs, MIMO and V-MIMO introduce distinct interference models requiring careful consideration. The work proposes a solution to an NP-hard problem, addressing three challenges: (i) interference; (ii) latency; and (iii) dynamic changes in network topology. Firstly, to counter interference, we propose a model where multiple nodes can simultaneously send data to the same parent by connecting different antennas. Secondly, to minimize latency, we propose a novel distributed heuristic data aggregation scheduling method, which intertwines the construction of an optimal data aggregation tree and conflict-free scheduling. Finally, to handle dynamic network topology changes, we propose lightweight adaptive strategies that do not increase data aggregation latency. Simulation results and theoretical analysis demonstrate superior performance in reducing data aggregation latency. When compared with state-of-the-art solutions, our proposed method decreases data aggregation latency by at least 2.6× on average.

Internet of Things (IoT) systems increasingly operate in environments where devices must respond in real time while managing fluctuating resource constraints, including energy and bandwidth. Yet, current approaches often fall short in addressing scenarios where operational constraints evolve over time. To address these limitations, we propose a novel Budgeted Multi-Armed Bandit framework tailored for IoT applications with dynamic operational limits. Our model introduces a decaying violation budget, which permits limited constraint violations early in the learning process and gradually enforces stricter compliance over time. We present the Budgeted Upper Confidence Bound (UCB) algorithm, which adaptively balances performance optimization and compliance with time-varying constraints. We provide theoretical guarantees showing that Budgeted UCB achieves sublinear regret and logarithmic constraint violations over the learning horizon. Extensive simulations in a wireless communication setting show that our approach achieves faster adaptation and better constraint satisfaction than standard online learning methods. These results highlight the framework’s potential for building adaptive, resource-aware IoT systems.

Federated learning has emerged as a privacy-preserving technique for collaborative model training across heterogeneously distributed silos. Yet, its reliance on a single central server introduces potential bottlenecks and risks of single-point failure. Decentralizing the server, often referred to as decentralized learning, addresses this problem by distributing the server’s role across nodes within the network. One drawback regarding this pure decentralization is it introduces communication inefficiencies, which arise from increased message exchanges in large-scale setups. However, existing proposed solutions often fail to simulate the real-world distributed and decentralized environment in their experiments, leading to unreliable performance evaluations and limited applicability in practice. Recognizing the lack from prior works, this work investigates the correlation between model size and network latency, a critical factor in optimizing decentralized learning communication. We propose a graph-based gossiping mechanism, where specifically, minimum spanning tree and graph coloring are used to optimize network structure and scheduling for efficient communication across various network topologies and message capacities. Our approach configures and manages subnetworks on real physical routers and devices and closely models real-world distributed setups. Experimental results demonstrate that our method significantly improves communication, compatible with different topologies and data sizes, reducing bandwidth and transfer time by up to circa 8 and 4.4 times, respectively, compared to naive flooding broadcasting methods.

Machine learning (ML) typically relies on the assumption that training and testing distributions are identical and that data are centrally stored for training and testing. However, in real-world scenarios, distributions may differ significantly, and data are often distributed across different devices, organizations, or edge nodes. Consequently, it is to develop models capable of effectively generalizing across unseen distributions in data spanning various domains. In response to this challenge, there has been a surge of interest in federated domain generalization (FDG) in recent years. FDG synergizes federated learning (FL) and domain generalization (DG) techniques, facilitating collaborative model development across diverse source domains for effective generalization to unseen domains, all while maintaining data privacy. However, generalizing the federated model under domain shifts remains a complex, underexplored issue. This article provides a comprehensive survey of the latest advancements in this field. Initially, we discuss the development process from traditional ML to domain adaptation (DA) and DG, leading to FDG, as well as provide the corresponding formal definition. Subsequently, we classify recent methodologies into four distinct categories: federated domain alignment (FDAL), data manipulation (DM), learning strategies (LSs), and aggregation optimization (AO), detailing appropriate algorithms for each. We then overview commonly utilized datasets, applications, evaluations, and benchmarks. Conclusively, this survey outlines potential future research directions.

As a key technology of intelligent satellite-enabled services in B5G or 6G networks, deploying Deep Neural Networks (DNN) models on satellites has been a notable trend, catering to the daily demand for extensive computing-intensive and latency-sensitive tasks. The computing resources are strategically deployed on satellites where sensor data is generated or collected, facilitating the fine-grained computational inference of DNN-based tasks. However, no prior study has comprehensively explored the crucial inference challenges – e.g., the trade-off between the number of tasks completed and accuracy and partitioning models in multi-exit models – in the resource-constrained space environment. Effective scheduling frameworks cater to various streams of inference tasks are scarce because inference performance may deviate from the ideal situation due to changes in task system status, such as task profiles and network state. To this end, we first formulate a gain-aware in-orbit computing inference problem to strike a proper trade-off between inference latency and the number of tasks completed by dynamically selecting optimal early exit points and model partitioning points. We propose an offline dynamic programming-based algorithm that provides an effective solution when comprehensive system details are to be predicted. We have developed an online learning-based method to schedule inference tasks with uncertain and dynamic system statuses in real-world situations. Our evaluation shows that, compared to baseline methods, the online learning-based algorithm can improve task gain by an average of 87.3% across various tasks.

Federated learning stands out as a promising approach within the domain of edge computing, providing a framework for collaborative training on distributed datasets without necessitating data sharing. However, federated learning involves the frequent transmission of machine learning model updates between the server and clients, resulting in high communication costs. Additionally, heterogeneous clients can further complicate the Federated Learning process and deteriorate performance. To address these challenges, we propose Adaptive Self-Knowledge Distillation-based Quality- and Reputation-Aware Cross-Device Federated Learning (ASDQR) - an efficient communication and inference framework designed for heterogeneous clients. ASDQR initiates the process by selecting high-reputation and high-quality clients to be involved in federated learning, significantly impacting communication efficiency and inference effectiveness. ASDQR also introduces a model of adaptive local self-knowledge distillation that incorporates multiple local personalized historical knowledge for more accurate inference, allowing the historical level to be dynamically adjusted across time. Finally, we present an inference-effective aggregation scheme that assigns higher weights to important and reliable local model updates based on clients’ contribution degrees when performing global model aggregation. ASDQR consistently outperforms baseline methods across all datasets and communication rounds, achieving 9.0% higher accuracy than FedAvg, 6.59% higher than MOON, 0.29% higher than FedProx, 0.2% higher than PFedSD, and 0.08% higher than FedMD on the MNIST dataset at 100 communication rounds. Similar improvements are observed on CIFAR, HAR, and WISDM datasets, demonstrating the robustness and efficiency of ASDQR in federated learning with non-IID data.

As a driving force in the advancement of intel-ligent in-orbit applications, DNN models have been gradually integrated into satellites, producing daily latency-constraint and computation-intensive tasks. However, the substantial computation capability of DNN models, coupled with the instability of the satellite-ground link, pose significant challenges, hindering the timely completion of tasks. It becomes necessary to adapt to task stream changes when dealing with tasks requiring latency guarantees, such as dynamic observation tasks on the satellites. To this end, we consider a system model for a collaborative inference system with latency constraints, leveraging the multi-exit and model partition technology. To address this, we propose an algorithm, which is tailored to effectively address the trade-off between task completion and maintaining satisfactory task accuracy by dynamically choosing early-exit and partition points. Simulation evaluations show that our proposed algorithm signif-icantly outperforms baseline algorithms across the task stream with strict latency constraints.

Computing Continuum (CC) systems are challenged to ensure the intricate requirements of each computational tier. Given the system’s scale, the Service Level Objectives (SLOs), which are expressed as these requirements, must be disaggregated into smaller parts that can be decentralized. We present our framework for collaborative edge intelligence, enabling individual edge devices to (1) develop a causal understanding of how to enforce their SLOs and (2) transfer knowledge to speed up the onboarding of heterogeneous devices. Through collaboration, they (3) increase the scope of SLO fulfillment. We implemented the framework and evaluated a use case in which a CC system is responsible for ensuring Quality of Service (QoS) and Quality of Experience (QoE) during video streaming. Our results showed that edge devices required only ten training rounds to ensure four SLOs; furthermore, the underlying causal structures were also rationally explainable. The addition of new types of devices can be done a posteriori; the framework allowed them to reuse existing models, even though the device type had been unknown. Finally, rebalancing the load within a device cluster allowed individual edge devices to recover their SLO compliance after a network failure from 22% to 89%.

The growing amount of data generated at the edge of the network, e.g., by Internet of Things (IoT) devices, made it indispensable to relocate computational power close to the data source. Meanwhile, data tends to accumulate in chunks and is frequently subject to resource-intensive transformations, such as privacy enforcement. These phenomena, which are summed up as “data gravity” and “data friction”, have an impact on data processing and the overall system. However, whereas cloud centers are able to dynamically adapt services, e.g., by provisioning additional resources, edge devices provide fewer options to react to changing workloads. To retain the option to process data locally, we present the idea of controlling data gravity and friction with Service Level Objectives (SLOs). We introduce Markov SLO Configurations (MSCs) as a novel approach to organizing performance metrics and elasticity strategies. MSCs, in conjunction with our presented architecture, enable the evaluation of SLOs, the context-based selection of elasticity strategy (i.e., corrective measures), and the execution of strategies directly on edge devices. Thus, we lay the foundation for a new generation of SLOs that can operate across multiple elasticity dimensions, e.g., by scaling quality of service (QoS).

Energy consumption for large amounts of delay-sensitive applications brings serious challenges with the continuous development and diversity of Industrial Internet of Things (IIoT) applications in fog networks. In addition, conventional cloud technology cannot adhere to the delay requirement of sensitive IIoT applications due to long-distance data travel. To address this bottleneck, we design a novel energy–delay optimization framework called transmission scheduling and computation offloading (TSCO), while maintaining energy and delay constraints in the fog environment. To achieve this objective, we first present a heuristic-based transmission scheduling strategy to transfer IIoT-generated tasks based on their importance. Moreover, we also introduce a graph-based task-offloading strategy using constrained-restricted mixed linear programming to handle high traffic in rush-hour scenarios. Extensive simulation results illustrate that the proposedTSCOapproach significantly optimizes energy consumption and delay up to 12%–17% during computation and communication over the traditional baseline algorithms.

This article presents our vision on the need of developing new managing technologies to harness distributed “computing continuum” systems. These systems are concurrently executed in multiple computing tiers: Cloud, Fog, Edge and IoT. This simple idea develops manifold challenges due to the inherent complexity inherited from the underlying infrastructures of these systems. This makes inappropriate the use of current methodologies for managing Internet distributed systems, which are based on the early systems that were based on client/server architectures and were completely specified by the application software. We present a new methodology to manage distributed “computing continuum” systems. This is based on a mathematical artifact called Markov Blanket, which sets these systems in a Markovian space, more suitable to cope with their complex characteristics. Furthermore, we develop the concept of equilibrium for these systems, providing a more flexible management framework compared with the one based on thresholds, currently in use for Internet-based distributed systems. Finally, we also link the equilibrium with the development of adaptive mechanisms. However, we are aware that developing the entire methodology requires a big effort and the use of learning techniques, therefore, we finish this article with an overview of the techniques required to develop this methodology.