Federated learning (FL) provides a high efficient decentralized machine framework, where the training data remains distributed at remote clients in network. Though FL enables privacy-preserving mobile edge computing framework using IoT devices, recent studies have shown that this approach is susceptible to poisoning attacks from side of clients. To address on FL, we provide <i>two-phase</i> defense algorithm called <inline-formula><tex-math notation="LaTeX">${\underline{Lo}cal\...

Federated Learning (FL) has been considered as an appealing framework to tackle data privacy issues of mobile devices compared conventional Machine (ML). Using Edge Servers (ESs) intermediaries perform model aggregation in proximity can reduce the transmission overhead, and it enables great potential low-latency FL, where hierarchical architecture FL (HFL) attracted more attention. Designing a proper client selection policy significantly improve training performance, widely investigated...

Federated learning (FL) is a decentralized machine architecture, which leverages large number of remote devices to learn joint model with distributed training data. However, the system-heterogeneity one major challenge in an FL network achieve robust performance, comes from two aspects: 1) device-heterogeneity due diverse computational capacity among and 2) data-heterogeneity nonidentically data across network. Prior studies addressing heterogeneous issue, for example, FedProx, lack...

Augmented reality (AR) has emerged as a promising tool in various disciplines, including language education. While previous studies have unpacked the potential and benefits of AR with regard to students' gains, most them focused on domain English learning. Yet, little is known about role Chinese Foreign Language (CFL) contexts. In this short paper study, we explored impacts CFL learners' reading comprehension by using quasi-experimental design an online course 54 undergraduate students....

Data aggregation is an important method of improving transmitting efficiency WSNs, but existing researches have some disadvantages as follows: several periods delay will be generated when filtering messages; the reduplicated messages ratio low; complex calculations need to executed; and extra hardware should added obtain high performance. To resolve these problems, this paper proposes a real time efficient data scheme (dynamical message list based aggregation, DMLDA) on clustering routing...

Federated learning (FL) is a new machine framework which trains joint model across large amount of decentralized computing devices. Existing methods, e.g., Averaging (FedAvg), are able to provide an optimization guarantee by synchronously training the model, but usually suffer from stragglers, i.e., IoT devices with low power or communication bandwidth, especially on heterogeneous problems. To mitigate influence this paper presents novel FL algorithm, namely Hybrid Learning (HFL), achieve...

Federated learning (FL) is a new machine framework that trains joint model across large number of decentralized computing devices. Existing methods, e.g., Averaging (FedAvg), are able to provide an optimization guarantee by synchronously training the model, but usually suffer from stragglers, i.e., IoT devices with low power or communication bandwidth, especially on problems non-i.i.d distributed data. To mitigate influence this paper presents novel FL framework, namely Hybrid Learning...

Microservice (MS) structures a service application as collection of independently deployable modules, making it particularly suitable for delivering complex applications in distributed computing systems. This paper investigates MS architecture over Mobile Edge Computing (MEC) networks (hereafter referred to EdgeMS) and studies an EdgeMS placement problem that aims deploy modules the MEC network manner maximizes reward providers. A novel algorithm called Dual-GNN Deep Deterministic Policy...

Multi-server Federated learning (FL) has been considered as a promising solution to address the limited communication resource problem of single-server FL. We consider typical multi-server FL architecture, where coverage areas regional servers may overlap. The key point this architecture is that clients located in overlapping update their local models based on average model all accessible models, which enables indirect sharing among different servers. Due complicated network topology,...

The topology of a network is fundamental for building infrastructure functionalities. In many scenarios, enterprise networks may have no desire to disclose their information. this paper, we aim at preventing attacks that use adversarial, active end-to-end inference obtain the information target network. To end, propose Proactive Topology Obfuscation (ProTO) system adopts detect-then-obfuscate framework: (i) lightweight probing behavior identification mechanism based on machine learning...

Audio adversarial examples (AEs) have posed significant security challenges to real-world speaker recognition systems.Most black-box attacks still require certain information from the model be effective (e.g., keeping probing and requiring knowledge of similarity scores).This work aims push practicality by minimizing attacker's about a target model.Although it is not feasible for an attacker succeed with completely zero knowledge, we assume that only knows short (or few seconds) speech...

We revisit the traditional framework of wireless secret key generation, where two parties leverage channel randomness to establish a key. The essence in is quantify into bit sequences for generation. Conducting tests on such has been common practice provide confidence validate whether they are random. Interestingly, despite different settings tests, existing studies interpret results same: passing means that indeed In this paper, we investigate how properly test ensure enough security...

A wireless sensor network (WSN) model by analyzing the limitations of existing localization algorithm is proposed in this paper. Specifically, neither anchor nor ID required model, and node function limited. The hardware our independent MAC protocol. Moreover, based on WSN we design an energy efficient anchor-free for no-identity WSN. Instead multiple measurements, mathematical calculation adopted to acquire position information. Subsequently, effectiveness confirmed rigid theory. simulation...

Previous adversarial audio attacks have mainly focused on ensuring the effectiveness of attacking an signal classifier via creating a small noise-like perturbation original signal. It is still unclear if attacker able to create perturbations that can be well perceived by human beings in addition its attack effectiveness. In this work, we formulate against music signals as new perception-aware framework, which integrates study into design. Specifically, invite participants rate their...

Federated Learning (FL) is a decentralized machine learning architecture, which leverages large number of remote devices to learn joint model with distributed training data. However, the system-heterogeneity one major challenge in FL network achieve robust performance, comes from two aspects: i) device-heterogeneity due diverse computational capacity among devices; ii) data-heterogeneity non-identically data across network. Prior studies addressing heterogeneous issue, e.g., FedProx, lack...

Data aggregation algorithm is an efficient method to filter messages from different sources and merge the reduplicated ones. In wireless sensor networks, it can be used avoid invalid transmitting. By well designed data algorithms, nodes' energy would conserved utilization of channel improved. So designing reasonable one most important research fields improve WSNs' performance. this paper, a new named DQDA (Dynamic Queue Aggregation) proposed which design on base hierarchal routing algorithm....

The task of searching for visual objects in a large image dataset is difficult because it requires efficient matching and accurate localization that can vary size. Although the segment anything model (SAM) offers potential solution extracting object spatial context, learning embeddings local remains challenging problem. This paper presents novel unsupervised deep metric approach, termed collaborative with mixed-scale groups (MS-UGCML), devised to learn varying scales. Following this,...

Federated Domain Generalization (FedDG) aims to train the global model for generalization ability unseen domains with multi-domain training samples. However, clients in federated learning networks are often confined a single, non-IID domain due inherent sampling and temporal limitations. The lack of cross-domain interaction in-domain divergence impede domain-common features limit effectiveness existing FedDG, referred as single-source FedDG (sFedDG) problem. To address this, we introduce...