There is an increasing interest in a new machine learning technique called Federated Learning, which the model training distributed over mobile user equipments (UEs), and each UE contributes to by independently computing gradient based on its local data. Learning has several benefits of data privacy potentially large amount participants with modern powerful processors low-delay mobile-edge networks. While most existing work focused designing algorithms provable convergence time, other issues...

There is an increasing interest in a fast-growing machine learning technique called Federated Learning (FL), which the model training distributed over mobile user equipment (UEs), exploiting UEs' local computation and data. Despite its advantages such as preserving data privacy, FL still has challenges of heterogeneity across physical resources. To address these challenges, we first propose FEDL, algorithm can handle heterogeneous UE without further assumptions except strongly convex smooth...

Fifth-generation cellular mobile networks are expected to support mission critical URLLC services in addition enhanced broadband applications. This article first introduces three emerging applications of and identifies their requirements on end-to-end latency reliability. We then investigate the various sources delay current wireless by taking 4G LTE as an example. Then we propose evaluate several techniques reduce from perspectives error control coding, signal processing, radio resource...

Recent advances in deep neural networks (DNNs) have substantially improved the accuracy and speed of a variety intelligent applications. Nevertheless, one obstacle is that DNN inference imposes heavy computation burden to end devices, but offloading tasks cloud causes transmission large volume data. Motivated by fact data size some intermediate layers significantly smaller than raw input data, we design surgery, which allows partitioned processed at both edge while limiting transmission. The...

The joint user association and spectrum allocation problem is studied for multi-tier heterogeneous networks (HetNets) in both downlink uplink the interference-limited regime. Users are associated with base-stations (BSs) based on biased received power. Spectrum either shared or orthogonally partitioned among tiers. This paper models placement of BSs different tiers as spatial point processes adopts stochastic geometry to derive theoretical mean proportionally fair utility network coverage...

Horizontal and vertical handoffs are important ramifications of user mobility in multitier heterogeneous wireless networks. They directly impact the signaling overhead quality calls. However, they difficult to analyze due irregularly shaped network topologies introduced by multiple tiers cells. In this paper, a stochastic geometric analysis framework on is proposed, capture spatial randomness various scales cell sizes different tiers. We derive theoretical expressions for rates all handoff...

Cloud computing has become the de facto platform for application processing in era of Internet Things (IoT). However, limitations cloud model, such as high transmission latency and costs are giving birth to a new paradigm called edge (a.k.a fog computing). Fog aims move data close network so reduce traffic. since servers at layer not powerful ones cloud, there is need balance between cloud. Moreover, besides offloading issue, energy efficiency nodes an increasing concern. Densely deployed...

Equipped with easy-to-access micro computation access points, the fog computing architecture provides low-latency and ubiquitously available offloading services to many simple cheap Internet of Things devices limited energy resources. One obstacle, however, is how seamlessly hand over mobile IoT among different points when in action so that service not interrupted -- especially for time-sensitive applications. In this article, we propose Follow Me Fog (FMF), a framework supporting new...

This paper investigates the optimal resource allocation of a downlink non-orthogonal multiple access (NOMA) system consisting one base station and users. Unlike existing short-term NOMA designs that focused on for only current transmission timeslot, we aim to maximize long-term network utility by jointly optimizing data rate control at layer power among users physical layer, subject practical constraints both consumptions. To solve this problem, leverage recently developed Lyapunov...

With the proliferation of Internet Things (IoT) and rising interconnectedness devices, network security faces significant challenges, especially from anomalous activities.While traditional machine learning-based intrusion detection systems (ML-IDS) effectively employ supervised learning methods, they possess limitations such as requirement for labeled data challenges with high dimensionality.Recent unsupervised ML-IDS approaches AutoEncoders Generative Adversarial Networks (GAN) offer...

We study joint spectrum allocation and user association in heterogeneous cellular networks with multiple tiers of base stations. A stochastic geometric approach is applied as the basis to derive average downlink data rate a closed-form expression. Then, expression employed objective function jointly optimizing association, which non-convex programming nature. computationally efficient Structured Spectrum Allocation User Association (SSAUA) proposed, solving optimization problem optimally...

A promising solution to address the spectrum shortage in 5G cellular systems is deployment of millimeter wave (mmWave) heterogeneous networks (HetNets). However, a key challenge for mmWave HetNets manage user mobility and handovers among small cells exploiting highly directional antennas conventional microwave macro cells. In this paper, we propose new efficient handover decision algorithm based on Markov Decision Process (MDP) optimize overall service experience users HetNets. By utilizing...

The steep rise of Internet Things (IoT) applications along with the limitations Cloud Computing to address all IoT requirements promotes a new distributed computing paradigm called Fog Computing, which aims process data at edge network. With help transmission latency and monetary spending caused by can be effectively reduced. However, executing in fog nodes will increase average response time since processing capabilities is not as powerful cloud. A tradeoff issue needs addressed within such...

Federated learning (FL) is a fast-developing technique that allows multiple workers to train global model based on distributed dataset. Conventional FL (FedAvg) employs gradient descent algorithm, which may not be efficient enough. Momentum able improve the situation by adding an additional momentum step accelerate convergence and has demonstrated its benefits in both centralized environments. It well-known Nesterov Accelerated Gradient (NAG) more advantageous form of momentum, but it clear...

Recent advances in deep neural networks have substantially improved the accuracy and speed of various intelligent applications. Nevertheless, one obstacle is that DNN inference imposes a heavy computation burden on end devices, but offloading tasks to cloud causes large volume data transmission. Motivated by fact size some intermediate layers significantly smaller than raw input data, we designed surgery, which allows partitioned be processed at both edge while limiting The challenge...

Federated Learning (FL) enables many clients to train a joint model without sharing the raw data. While byzantine-robust FL methods have been proposed, remains vulnerable security attacks such as poisoning and evasion due its distributed adversarial environment. Additionally, real-world training data used in are usually Non-Independent Identically Distributed (Non-IID), which further weakens robustness of existing (such Krum, Median, Trimmed-Mean, etc.), thereby making it possible for global...

For transmission control protocol (TCP), CUBIC is a TCP-friendly high-speed variant, in which the window size cubic function of time since last loss event. TCP implemented Linux operating systems and performs well wired networks with large bandwidth-delay product. Most evaluations are conducted via simulations or experiments. Analytical models for few. In this paper, we propose Markovian model to determine steady state throughput wireless environment. The proposed considers both congestion...

User-centric base station (BS) cooperation has been regarded as an effective solution for improving network coverage and throughput in next-generation wireless systems. However, it also introduces more complicated handoff patterns, which may potentially degrade user performance. In this paper, we aim to theoretically quantify the tradeoff between cost data rate. Two user-centric clustering modes are investigated: number-based (NBC), is easier implement, distance-based (DBC), gives higher...

As cyber attacks to Industrial Internet of Things (IIoT) remain a major challenge, blockchain has emerged as promising technology for IIoT security due its decentralization and immutability characteristics. Existing designs, however, introduce high computational complexity latency challenges which are unsuitable IIoT. This paper proposes Xyreum, new high-performance scalable enhanced privacy. Xyreum uses Time-based Zero-Knowledge Proof Knowledge (T-ZKPK) with authenticated encryption perform...

There is growing interest in applying distributed machine learning to edge computing, forming <i>federated learning</i> . Federated faces non-i.i.d. and heterogeneous data, the communication between workers, possibly through distant locations with unstable wireless networks, more costly than their local computational overhead. In this work, we propose <inline-formula><tex-math notation="LaTeX">${{\sf DONE}}$</tex-math></inline-formula> , a approximate Newton-type algorithm fast convergence...