Traditional knowledge distillation relies on simple MSE or KL divergence losses that fail to capture the complex distributional relationships between teacher and student model representations. We propose FlowDistill, a novel distillation framework that employs normalizing flows to model and transfer the intricate knowledge distributions from teacher to student models. Our approach introduces three key innovations: (1) Invertible Knowledge Mapping using continuous normalizing flows (CNFs) to learn bijective transformations between teacher and student representation spaces, enabling precise knowledge transfer without information loss, (2) Flow-Guided Progressive Distillation that gradually increases the complexity of knowledge transfer by learning hierarchical flow transformations from simple to complex distributions, and (3) Conditional Flow Networks that adapt knowledge transfer based on input context and task requirements. Unlike previous diffusion-based distillation methods such as DiffKD that suffer from computational overhead due to iterative denoising processes and information loss during noise addition, our flow-based approach provides exact invertible transformations with significantly reduced computational cost. Extensive experiments on ImageNet classification, COCO object detection, and Cityscapes semantic segmentation demonstrate that FlowDistill achieves superior performance with 2.1% accuracy improvement over DiffKD on ResNet-34 to ResNet-18 distillation while reducing inference time by 3.5×. Our method establishes new state-of-the-art results across multiple distillation benchmarks and provides theoretical guarantees for lossless knowledge transfer through invertible flow transformations.

With the increasing demand for real-time data processing in disaster relief and remote areas, unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) systems have emerged as a promising solution by providing power and computing resources to user devices (UDs). This paper introduces a multi-cluster fair service UAV-assisted wireless powered MEC system aimed at addressing the challenges between energy efficiency and fair service. We formulate a long-term multi-objective energy efficiency optimization problem (LMEEOP) to trade off UD energy consumption, UAV energy consumption, and offline rate. However, this problem is a mixed integer nonlinear programming problem and difficult to solve. Thus, we propose a trajectory limited online joint optimization approach (TLOJOA), which decomposes LMEEOP into three sub-problems: UAV flight trajectory design, offloading decision generation, and resource allocation. Specifically, a Self-Attention combined proximal policy optimization (PPO) algorithm captures long-range dependencies to balance energy consumption and offline rate. Additionally, a fast offloading strategy generation algorithm with offloading importance ranking is proposed to improve UD energy efficiency. Finally, the UAV flight trajectory and offloading decision are combined to achieve efficient resource allocation. The proposed approach provides a robust solution for improving multi-cluster fair service and energy efficiency optimization in UAV-assisted MEC systems, especially for mission-critical scenarios with constrained resources. Simulation results validate that the proposed approach can achieve outstanding system performance compared to other benchmark approaches, especially in fairness of service.

Internet of Underwater Thing (IoUT) stands as promising technology facilitating diverse underwater applications. Nevertheless, IoUT across vast marine regions is challenged by highly diverse and fluctuating channel environments, which results in unreliable point-to-point (PTP) transmissions. Moreover, its multi-hop nature exacerbates severe unreliable end-to-end (ETE) transmissions. Existing methods utilize routing protocols to address the above challenges by independently power control for PTP reliability or multi-path transmission for ETE reliability. However, these methods ignore the interdependencies between power control and multi-path transmission, which fail to guarantee high energy-efficient reliability in resource-constrained and harsh underwater environments. To this end, we propose a joint power Control And Multi-Path routing (CAMP) protocol for IoUTs in varying environments. Specifically, we develop PTP and ETE reliability models by analyzing the interrelation between power control and multi-path routing, incorporating historical, current, and predictive information. A hybrid routing strategy is designed based on the reliability models to accommodate changing environmental conditions, residual energy, and link quality. This strategy initiates multi-path routing at the source and single-path forwarding at relay nodes, combined with power control. Extensive simulations demonstrate that CAMP achieves superior reliability (packet delivery rate) and energy efficiency, while simultaneously improving network performance in terms of latency and throughput.

Increasing demands for versatile applications have spurred the rapid development of Unmanned Underwater Vehicle (UUV) networks. Nevertheless, multi-UUV movements exacerbates the spatial-temporal variability, leading to serious intermittent connectivity of underwater acoustic channel. Such phenomena challenge the identification of reliable paths for high-dynamic network routing. Existing routing protocols overlook the effects of UUV movements on forwarding path, typically selecting forwarders based solely on the current network state, which lead to instability in packet transmission. To address these challenges, we propose a Routing protocol based on Spatial-Temporal Graph model with Q-learning for multi-UUV networks (STGR), achieving high reliable and energy effective transmission. Specifically, a distributed Spatial-Temporal Graph model (STG) is proposed to depict the evolving variation characteristics (neighbor relationships, link quality, and connectivity duration) among underwater nodes over periodic intervals. Then we design a Q-learning-based forwarder selection algorithm integrated with STG to calculate reward function, ensuring adaptability to the ever-changing conditions. We have performed extensive simulations of STGR on the Aqua-Sim-tg platform and compared with the state-of-the-art routing protocols in terms of Packet Delivery Rate (PDR), latency, energy consumption and energy balance with different network settings. The results show that STGR yields 24.32 percent higher PDR on average than them in multi-UUV networks.

A high-performance network deployment strategy supports fundamental network services, such as topology controls, protocol designs, and boundary detections in underwater wireless sensor networks (UWSNs). Existing deployment methods treat nodes within the communication range as connected. However, in addition to internode distance, packet errors and collisions are also significant factors for point-to-point connectivity. Furthermore, when allocating node locations, deployment strategies focus on maximizing coverage, ignoring the tradeoff between coverage and network performance (reliability, latency, and energy efficiency). To this end, an efficient deployment scheme with network performance modeling (EDNPM) is proposed, to provide reliable data transmission in a time-aware and energy-efficient way for UWSNs. Specifically, we first explore sensor locations’ impact on communication and network factors, to improve the point-to-point connectivity and network performance. A network performance evaluation model (NPEM) is established to quantify performance metrics for guiding network deployment. Based on NPEM, network deployment is formulated as a multiobjective optimization problem, and we propose a novel network connection-constraint particle swarm optimization (NCPSO) algorithm to solve this problem. Notably, EDNPM is a unified network deployment framework for various underwater applications. Extensive experiments demonstrate that EDNPM outperforms other deployment algorithms in terms of network performance, and robustness with different network settings.

Various remote sensing applications have widely used domain adaptation (DA) methods. Since it does not need to add human interpretation in the target domain, it can be used in cross-region, multi-temporal, and multi-sensor application scenarios. In order to further optimize the design of the loss function and better address the challenges of DA in remote sensing, in this paper, we propose a new universal DA method named C3DA for scene recognition of remote sensing images. It has a comprehensive C3criterion for recognizing the "unknown" classes by innovatively fusing confidence, consistency, and certainty of samples to make our network training more efficient. We evaluate the performance of our proposed method based on six transfer tasks on three remote sensing datasets. The evaluation results show that our proposed method achieves an average H-score of 58.44%, significantly higher than other SOTA universal DA methods with an average improvement of 2.32~29.43%. Compared to the baseline ResNet-50, it achieves up to 19.92% improvement, demonstrating that the proposed method outperforms in the universal DA scenario. In the future, we also plan to expand the application of this method to more scenarios.

Continuous and remote health monitoring medical applications with heterogeneous requirements can be realized through wireless body area networks (WBANs). Energy harvesting is adopted to enable low-power health applications and long-term monitoring without battery replacement, which have drawn significant interest recently. Because energy harvesting WBANs are obviously different from battery-powered ones, network protocols should be designed accordingly to improve network performance. In this article, an efficient cross-layer media access control protocol is proposed for radio frequency powered energy harvesting WBANs. We redesigned the superframe structure, which can be rescheduled by the coordinator dynamically. A time switching (TS) strategy is used when sensors harvest energy from radio frequency signals broadcast by the coordinator, and a transmission power adjustment scheme is proposed for sensors based on the energy harvesting efficiency and the network environment. Energy efficiency can be effectively improved that more packets can be uploaded using limited energy. The length of the energy harvesting period is determined by the coordinator to balance the channel resources and energy requirements of sensors and further improve the network performance. Numerical simulation results show that our protocol can provide superior system performance for long-term periodic health monitoring applications.

Federated learning (FL) is a promising new technology in the field of IoT intelligence. However, exchanging model-related data in FL may leak the sensitive information of participants. To address this problem, we propose a novel privacy-preserving FL framework based on an innovative chained secure multiparty computing technique, named chain-PPFL. Our scheme mainly leverages two mechanisms: 1) single-masking mechanism that protects information exchanged between participants and 2) chained-communication mechanism that enables masked information to be transferred between participants with a serial chain frame. We conduct extensive simulation-based experiments using two public data sets (MNIST and CIFAR-100) by comparing both training accuracy and leak defence with other state-of-the-art schemes. We set two data sample distributions (IID and NonIID) and three training models (CNN, MLP, and L-BFGS) in our experiments. The experimental results demonstrate that the chain-PPFL scheme can achieve practical privacy preservation (equivalent to differential privacy with ∈ approaching zero) for FL with some cost of communication and without impairing the accuracy and convergence speed of the training model.

Device-to-Device (D2D) communication has attracted extensive researches because of its ability to reduce latency and improve the spectrum resource utilization. This paper studies a D2D Mobile Edge Computing (MEC) system which contains multiple busy smart devices (SDs) and multiple idle smart devices. To minimize the total energy consumption of the MEC system and satisfy the latency constraints of SDs, the computation intensive task of each busy SD can be partially offloaded to one or more idle SDs as helpers. Therefore, a joint optimization problem of helpers selection and communication and computation resources allocation is proposed. The problem is formulated as an integer-mixed non-convex optimization problem which is a NP-hard problem. We thus propose a two-phase iterative approach by jointly optimizing helpers selection and communication and computation resources allocation. In the first phase, we obtain the suboptimal helpers selection policy with convex optimization techniques and block coordinate descent method. In the second phase, the resource allocation strategy is achieved by applying block coordinate descent after obtaining the suboptimal helpers selection policy. The simulation results demonstrate that not only the proposed algorithm achieves fast convergence in both phases, but also the overall energy consumption is less than other benchmarks.

Large-scale tasks processing based on cloud computing has become crucial to big data analysis and disposal in recent years. Most previous work, generally, utilize the conventional methods and architectures for general scale tasks to achieve tons of tasks disposing, which is limited by the issues of computing capability, data transmission, etc. Based on this argument, a fat-tree structure-based approach called LTDR (Large-scale Tasks processing using Deep network model and Reinforcement learning) has been proposed in this work. Aiming at exploring the optimal task allocation scheme, a virtual network mapping algorithm based on deep convolutional neural network and [Formula: see text]-learning is presented herein. After feature extraction, we design and implement a policy network to make node mapping decisions. The link mapping scheme can be attained by the designed distributed value-function based reinforcement learning model. Eventually, tasks are allocated onto proper physical nodes and processed efficiently. Experimental results show that LTDR can significantly improve the utilization of physical resources and long-term revenue while satisfying task requirements in big data.

Recently big data have become a research hotspot and been successfully exploited in a few applications such as data mining and business modeling. Although big data contain a plenty of treasures for all the fields of computer science, it is very difficult for the current computing paradigms and computer hardware to efficiently process and utilize big data to attain what are looked forward to. In this work, we explore the possibility of employing big data in recommendation systems. We have proposed a simple recommendation system framework BDRSF (Big Data Recommendation System Framework), which is based on big data with social context theories and has abilities in obtaining the Recommender based on the idea of supervised learning through big data training. Its main idea can be divided into three parts: (1) reduce the scale of the current recommendation problems according to the essence of recommending; (2) design a rational Recommender and propose a novel supervised learning algorithm to get it; (3) utilize the Recommender to deal with the later recommendation problems. Experimental results show that BDRSF outperforms conventional recommendation systems, which clearly indicates the effectiveness and efficiency of big data with social context in personalized recommendation.

Aiming at the current problems that most physical hosts in the cloud data center are so overloaded that it makes the whole cloud data center'load imbalanced and that existing load balancing approaches have relatively high complexity, this paper has focused on the selection problem of physical hosts for deploying requested tasks and proposed a novel heuristic approach called Load Balancing based on Bayes and Clustering (LB-BC). Most previous works, generally, utilize a series of algorithms through optimizing the candidate target hosts within an algorithm cycle and then picking out the optimal target hosts to achieve the immediate load balancing effect. However, the immediate effect doesn't guarantee high execution efficiency for the next task although it has abilities in achieving high resource utilization. Based on this argument, LB-BC introduces the concept of achieving the overall load balancing in a long-term process in contrast to the immediate load balancing approaches in the current literature. LB-BC makes a limited constraint about all physical hosts aiming to achieve a task deployment approach with global search capability in terms of the performance function of computing resource. The Bayes theorem is combined with the clustering process to obtain the optimal clustering set of physical hosts finally. Simulation results show that compared with the existing works, the proposed approach has reduced the failure number of task deployment events obviously, improved the throughput, and optimized the external services performance of cloud data centers.

The research for dynamic spectrum access (DSA) and cognitive radio network (CRN) becomes one of the focuses in wireless network currently. However, as many other new techniques, in the initialization period, the security factors in CRN are out of focus. This paper describes the special characteristics of cognitive radio (CR) and CRN, and analysis the current and potential security threats that due to their characteristics. Besides some countermeasures and keys need to attention are mentioned. The goal of this paper is to assist CR designers and the CR application engineers to consider the security factors in the initial development period of CR techniques.

Routing protocols play an important role in WSNs in order to gather data and send them to the BS (base station). One of the challenges that the WSNs have to face is to prolong its lifetime since transmitting and receiving interest information are the most energy exhausting phase. In this paper, we introduced a new routing algorithm called SPER (safe and power -efficient routing algorithm in wireless sensor networks) which is enlightened by PEGASIS (power-efficient gathering in sensor information systems). At the same time in SPER we gave out a novel formula to compute the value of weight in which energy consuming factor was introduced considering the power balance problem. Based on the multiple chains structure, a method of cipher keys distribution was designed. A symmetric public key formally installed in the sensors was used to encrypt the initial process. In each data aggregating rounds different cipher keys were used to enhance the security of the whole net.

A novel efficient and effective Internet multicast routing protocol is presented with short delay, high throughput, resource utilization and scalability for a single multicast group g. The protocol has two features: (1) multiple shared-trees (MST) are configured to provide efficient, dynamic and quality multicast routing; (2) an anycasting approach is used to form the tree roots into an anycast group so that the multicast packets can be anycast to the nearest node at one of the shared trees to achieve the best routing service for the multicast packets. The performance of the MST protocol is analyzed through extensive simulations and compared with well-known source tree and shared-tree routing.

A novel efficient and dynamic integrated routing protocol for multicast and anycast messages is presented. The contributions of the protocol differ from well-known shared-tree systems in two aspects: (1) Off-tree anycast configuration and routing: multicast sources use anycast routing to select a better path from the source to one router in the group in order to avoid congestion or any fault in the network. (2) On-tree router anycast configurations: The nodes in the shared-tree are formed into a virtual anycast group. The shared-tree approach is extended with capability of a group cores (anycast group). The simulation data demonstrates the efficiency of the protocol.

In this paper, we study an efficient fault-tolerant CBT multicast routing protocol. With our strategy, when a faulty component is detected, some pre-defined backup path(s) is (are) used to bypass the faulty component and enable the multicast communication to continue. Our protocol only requires that routers near the faulty component be reconfigured, thus reducing the runtime overhead without compromising much of the performance. Our performance evaluation shows that our new protocol performs nearly as well as the best possible global method while utilizing much less runtime overhead and implementation cost.

In this paper, we design and analyze an efficient fault-tolerant multicast routing protocol. Reliable multicast communication is critical for the success of many Internet applications. Multicast routing protocols with core-based tree techniques (CBT) have been widely used because of their scalability and simplicity. We enhance the CBT protocol with fault tolerance capability and improve its efficiency and effectiveness. With our strategy, when a faulty component is detected, some pre-defined backup path(s) is (are) used to bypass the faulty component and enable the multicast communication to continue. Our protocol only requires that routers near the faulty component be reconfigured, thus reducing the runtime overhead without compromising much of the performance. Our approach is in contrast to other approaches that often require relatively large tree reformation when faults occur. These global methods are usually costly and complicated in their attempt to achieve theoretically optimal performance. Our performance evaluation shows that our new protocol performs nearly as well as the best possible global method while utilizing much less runtime overhead and implementation cost.