Federated learning (FL) has emerged in edge computing to address limited bandwidth and privacy concerns of traditional cloud-based centralized training. However, the existing FL mechanisms may lead long training time consume a tremendous amount communication resources. In this paper, we propose an efficient mechanism, which divides nodes into K clusters by balanced clustering. The one cluster forward their local updates header for aggregation synchronous method, called aggregation, while all...

In the data flow models of today's center applications such as MapReduce, Spark and Dryad, multiple flows can comprise a coflow group semantically. Only completing all in is meaningful to an application. To optimize application performance, routing scheduling must be jointly considered at level rather than individual flows. However, prior solutions have significant limitation: they only consider scheduling, which insufficient. this end, we present Rapier, coflow-aware network optimization...

Network Function Virtualization (NFV) is a promising technique to greatly improve the effectiveness and flexibility of network services through process named Service Chain (SFC) mapping, with which different are deployed over virtualized shared platforms in data centers. However, such an evolution towards software-defined functions introduces new challenges require high availability. One effective way protecting use sufficient redundancy. By doing so, however, efficiency physical resources...

In Mobile Edge Computing (MEC), many tasks require specific service support for execution and in addition, have a dependent order of among the tasks. However, previous works often ignore impact having limited services cached at edge nodes on (dependent) task offloading, thus may lead to an infeasible offloading decision or longer completion time. To bridge gap, this article studies how efficiently offload with (and predetermined) caching. We formally define problem caching (ODT-SC), prove...

In Mobile Edge Computing (MEC), many tasks require specific service support for execution and in addition, have a dependent order of among the tasks. However, previous works often ignore impact having limited services cached at edge nodes on (dependent) task offloading, thus may lead to an infeasible offloading decision or longer completion time. To bridge gap, this paper studies how efficiently offload with (and predetermined) caching. We formally define problem caching (ODT-SC), prove that...

Federated learning (FL) has emerged in edge computing to address the limited bandwidth and privacy concerns of traditional cloud-based training. However, existing FL mechanisms may lead a long training time consume massive communication resources. In this paper, we propose an efficient mechanism, namely FedCH, accelerate heterogeneous computing. Different from works which adopt pre-defined system architecture train models synchronous or asynchronous manner, FedCH will construct special...

Federated learning (FL) has been widely adopted to train machine models over massive distributed data sources in edge computing. However, the existing FL frameworks usually suffer from difficulties of resource limitation and heterogeneity. Herein, we design implement FedMP, an efficient framework through adaptive model pruning. We theoretically analyze impact pruning ratio on training performance, propose employ a Multi-Armed Bandit based online algorithm adaptively determine different...

Network function virtualization is a promising technique to greatly improve the effectiveness and flexibility of network services through process named service chain (SFC) mapping, with which functions are deployed over virtualized shared platforms in data centers. However, failures quite common Therefore, practical yet theoretically challenging issue SFC mapping such an environment manage availability requests. In this paper, we present framework provision requests center multiple layers...

In current data centers, an application (e.g., MapReduce, Dryad, search platform, etc.) usually generates a group of parallel flows to complete job. These compose coflow and only completing them all is meaningful the application. Accordingly, minimizing average Coflow Completion Time (CCT) becomes critical objective flow scheduling. However, achieving this goal in today's Data Center Networks (DCNs) quite challenging, not because schedule problem theoretically NP-hard, but also it tough...

Quantum communication using qubits based on the principle of entangled photons is a promising solution to improve network security. However, it difficult successfully create an entanglement link or connection between two nodes, especially when they are far apart from each other. In addition, only one qubit can be exchanged over established connection, resulting in low throughput.In this paper, we propose Redundant Entanglement Pro-visioning and Selection (REPS) maximize throughput for...

This paper studies reliable teleportation of quantum bits (called qubits) in a data network with multiple sources (S) and destinations (D) as well repeaters. To teleport qubits for SD pair reliably, not only an entanglement path the pair, but also appropriate purification links along is required to ensure that end-to-end (E2E) fidelity established connections high enough.This first work on quantifying E2E fidelity, using this determine critical achieve most resource efficient purification. A...

Data generated at the network edge can be processed locally by leveraging paradigm of computing. To fully utilize widely distributed data, we concentrate on a wireless computing system that conducts model training using decentralized peer-to-peer (P2P) methods. However, there are two major challenges way towards efficient P2P training: limited resources (e.g., bandwidth and battery life mobile devices) time-varying connectivity due to device mobility or channel dynamics, which have received...

TCP in cast congestion which can introduce hundreds of milliseconds delay and up to 90% throughput degradation, severely affecting application performance, has been a practical issue high-bandwidth low-latency data enter networks. Despite continuous efforts, prior solutions have significant drawbacks. They either only support quite limited number senders (e.g., 40-60), is not sufficient, or require non-trivial system modifications, impractical incrementally deployable. We present PAC, simple...

In current data centers, an application (e.g. MapReduce) usually generates a collection of parallel flows sharing common goal. These compose coflow and only completing them all is meaningful. Accordingly, minimizing the average completion time (CCT) becomes critical objective for flow scheduling. this topic, state-of-the-art centralized method, Varys, achieves good CCT; but it has scalability problem. Alternatively, existing decentralized Baraat, suffers from head-of-line blocking To solve...

Network function virtualization (NFV) provides an effective way to reduce the network provider's cost by allowing multiple virtual networks (VNs) share underlying physical infrastructure. In NFV environment, especially when supporting multicast services over VNs, reliability is a critical requirement since failure of one node can cause malfunction nodes that receive multicasting data from it. this paper, we study for first time best our knowledge how efficiently map VNs both general IP and...

Quantum Data Networks (QDNs) typically leverage teleportation to reliably send data quantum bits (called qubits) their destinations. To teleport a qubit from Alice Bob, one entanglement connection between and Bob needs be established. Accordingly, we have establish as many connections possible with limited resources in order maximize the network throughput. Conventional methods assume known traffic matrix calculate paths batch using centralized algorithm. However, these are not scalable...

Network Function Virtualization (NFV) introduces a new network architecture that offers different services flexibly and dynamically in the form of Service Chains (SFCs), which refer to set Functions (VNFs) chained specific order. However, service latency often increases linearly with length SFCs due sequential execution VNFs, resulting sub-optimal performance for most delay-sensitive applications. In this paper, novel Parallel VNF Placement (PVFP) approach is proposed real-world networks via...

In this paper, we study the tradeoff between two important traffic engineering objectives: load balance and energy efficiency. Although traditional commonly used multi-objective optimization methods can yield a Pareto efficient solution, they need to construct an aggregate objective function (AOF) or model one of objectives as constraint in problem formulation. As result, it is difficult achieve fair these objectives. Accordingly, induce Nash bargaining framework which treats virtual players...

There are two conventional methods to establish an entanglement connection in a Quantum Data Network (QDN). One is create single-hop links first and then connect them with quantum swapping, the other for-warding one of entangled photons from end via all-optical switching at intermediate nodes directly connection. Since photon easy be lost during long distance transmission, all existing works adopting former method. However, room size network, success probability delivering across multiple...

Quantum computing holds great promise and this work proposes to use new quantum data networks (QDNs) connect multiple small computers form a cluster. Such QDN differs from existing key distribution (QKD) in that the former must deliver bits (i.e., qubits) reliably between different computers. Two families of QDNs are studied, one using teleportation, named Tele-QDN, other tell-and-go (TAG), TAG-QDN. In order provide reliable delivery qubits, while addressing QDN-specific constraints imposed...

In surgical training and experimental research, brain tissues immersed in cerebrospinal fluid often exhibit complex deformation strain rate effects that can compromise their reliability stability. Therefore, it is essential to develop a high-fidelity human tissue simulant material serves as physical surrogate model understand its mechanical behavior, such traumatic injury (TBI). However, the existing materials have failed meet required properties. This study presents composite hydrogel...

This paper studies how to determine an optimal order of recovering interdependent Cyber Physical Systems (CPS) after a large scale failure. In such CPS, some failed devices must be repaired first before others can. addition, require certain amount repair resources and may take multiple stages repair. We consider two scenarios: 1) reserved model where all the required should prepared at beginning repairing device; 2) opportunistic we can partially device with only part resources. For each...