Federated Learning (FL) enables knowledge sharing among distributed edge devices while ensuring data privacy. However, implementing the FL technique in dynamic networks like the drone network, is challenged by the hig...
详细信息
ISBN:
(纸本)9798350384482;9798350384475
Federated Learning (FL) enables knowledge sharing among distributed edge devices while ensuring data privacy. However, implementing the FL technique in dynamic networks like the drone network, is challenged by the high communication cost among the clients and slow convergence rate of the global model due to the straggling clients in the network. Moreover, the straggler effect can exacerbate attack propagation in a security network, as attackers might exploit the delay of the intrusion detection model designed in the presence of stragglers. The semi-asynchronous FL (SAFL) method has displayed commendable performance in mitigating the straggler effect. However, existing SAFL techniques do not consider a holistic approach that improves the performance of the global model at a reduced communication cost. This study proposes an agnostic straggler-resilient SAFL (ASR-Fed) algorithm that prioritizes the updates of high-performing and efficient clients while circumventing the updates of straggling clients during the FL process. Simulation experiments performed under different scenarios evaluate the effectiveness of ASR-Fed. The results validate the robustness of ASR-Fed in enhancing the detection performance of the cybersecurity model achieving an accuracy above 98.5% within the least communication round. Outperforming existing state-of-the-art FL aggregating protocols.
The exploration into the evolution of architectural structures in 6G edge computing nodes through time-series-based dynamic interactions offers a compelling investigation within complex systems. In the realm of 6G edg...
详细信息
ISBN:
(纸本)9798350384482;9798350384475
The exploration into the evolution of architectural structures in 6G edge computing nodes through time-series-based dynamic interactions offers a compelling investigation within complex systems. In the realm of 6G edge computing, where nodes play a pivotal role, existing approaches primarily emphasize the locality of nodes or clustering relationships between networks. Traditional time-series network modeling tends to fixate on local static relationships, overlooking the dynamic interactions between real network nodes. To address this limitation, we present a model based on time-series interactions, specifically crafted for 6G edge computing networks. Our model extends beyond traditional boundaries, facilitating a comparative analysis of network formation across diverse datasets, presenting a valuable methodology for conducting evolutionary studies. The model's validity is demonstrated through evaluations on two real network datasets. Notably, within the 6G edge, a discernible structure emerges when the preference for high-level nodes surpasses a critical threshold.
Homography estimation serves as a fundamental technique for image alignment in a wide array of applications. The advent of convolutional neural networks has introduced learning-based methodologies that have exhibited ...
详细信息
ISBN:
(纸本)9798350318920;9798350318937
Homography estimation serves as a fundamental technique for image alignment in a wide array of applications. The advent of convolutional neural networks has introduced learning-based methodologies that have exhibited remarkable efficacy in this realm. Yet, the generalizability of these approaches across distinct domains remains underexplored. Unlike other conventional tasks, CNN-driven homography estimation models show a distinctive immunity to domain shifts, enabling seamless deployment from one dataset to another without the necessity of transfer learning. This study explores the resilience of a variety of deep homography estimation models to domain shifts, revealing that the network architecture itself is not a contributing factor to this remarkable adaptability. By closely examining the models' focal regions and subjecting input images to a variety of modifications, we confirm that the models heavily rely on local textures such as edges and corner points for homography estimation. Moreover, our analysis underscores that the domain shift immunity itself is intricately tied to the utilization of these local textures. (1)
The synchronization of clocks used at different transceivers across space is of critical importance in next-generation wireless networks. Experimental testbeds must provide synchronization services so that users can t...
详细信息
ISBN:
(纸本)9798350384482;9798350384475
The synchronization of clocks used at different transceivers across space is of critical importance in next-generation wireless networks. Experimental testbeds must provide synchronization services so that users can test new efficient communication and accurate localization technologies;yet reconfigurability can sometimes introduce unintended asynchronicities. We introduceWATCH, a tool to monitor the time synchronization of nodes in a geographically distributed software-defined radio (SDR) testbed. WATCH estimates clock differences using cross-correlation between transmitted and received packets and considers both propagation delay and noise in the channel. We model pairwise measurements as a linear function of the difference between the nodes' local clocks from the network's global clock, as well as each node's additional delay for packet transmission. This model allows WATCH to simultaneously estimate all of the clock offsets and transmit delays. We compare using transmissions of modulated data packets and pseudo-noise (PN) code signals to experimentally test and verify the performance of WATCH on the Platform for Open Wireless Data-driven Experimental Research (POWDER), where it detected a firmware problem that led to a time synchronization error between SDR nodes on the platform.
Network traffic was first determined to be self-similar through the analysis of core Ethernet traces and has since been shown to demonstrate this characteristic under various connection protocols. However, the variati...
详细信息
ISBN:
(纸本)9798350300741;9798350300734
Network traffic was first determined to be self-similar through the analysis of core Ethernet traces and has since been shown to demonstrate this characteristic under various connection protocols. However, the variation in self-similarity of network traffic due to changing connection protocol (e.g. WiFi to Ethernet) has not been explored, nor has it been shown how self-similarity behaves as data from different sources aggregates physically from the edge to a network's core. We establish that the increased variability inherent to wireless communications increases self-similarity when measured at the edge due to the conditions of the transmission medium and the protocols in place for wired and wireless connections. Furthermore, once aggregated on a wired link, the stability of wired protocols is shown to have the most significant impact on the estimated self-similarity, and the core traffic is less bursty.
Recent advances in Generative Adversarial networks (GANs) have enabled photo-realistic synthesis of single object images. Yet, modeling more complex distributions, such as scenes with multiple objects, remains challen...
详细信息
ISBN:
(纸本)9781665493468
Recent advances in Generative Adversarial networks (GANs) have enabled photo-realistic synthesis of single object images. Yet, modeling more complex distributions, such as scenes with multiple objects, remains challenging. The difficulty stems from the incalculable variety of scene configurations which contain multiple objects of different categories placed at various locations. In this paper, we aim to alleviate the difficulty by enhancing the discriminative ability of the discriminator through a locally defined self-supervised pretext task. To this end, we design a discriminator to leverage multi-scale local feedback that guides the generator to better model local semantic structures in the scene. Then, we require the discriminator to carry out pixel-level contrastive learning at multiple scales to enhance discriminative capability on local regions. Experimental results on several challenging scene datasets show that our method improves the synthesis quality by a substantial margin compared to state-of-the-art baselines.
In recent years, depth recovery based on deep networks has achieved great success. However, the existing state-of-the-art network designs perform like black boxes in depth recovery tasks, lacking a clear mechanism. Ut...
ISBN:
(纸本)9798350307184
In recent years, depth recovery based on deep networks has achieved great success. However, the existing state-of-the-art network designs perform like black boxes in depth recovery tasks, lacking a clear mechanism. Utilizing the property that there is a large amount of nonlocal common characteristics in depth images, we propose a novel model-guided depth recovery method, namely the DC-NLAR model. A non-local auto-regressive regular term is also embedded into our model to capture more non-local depth information. To fully use the excellent performance of neural networks, we develop a deep image prior to better describe the characteristic of depth images. We also introduce an implicit data consistency term to tackle the degenerate operator with high heterogeneity. We then unfold the proposed model into networks by using the half-quadratic splitting algorithm. This proposed method is experimented on the NYU-Depth V2 and SUN RGB-D datasets, and the experimental results achieve comparable performance to that of deep learning methods.
The application of deep learning in change detection (CD) for remote sensing has greatly strengthened CD networks. Although convolutional neural networks (CNN) can capture local features well, they are inefficient at ...
详细信息
ISBN:
(纸本)9798350386783;9798350386776
The application of deep learning in change detection (CD) for remote sensing has greatly strengthened CD networks. Although convolutional neural networks (CNN) can capture local features well, they are inefficient at extracting global features, reducing the detection accuracy. To address this limitation, we introduced a transformer to solve the constrained receptive field issue. We proposed a Bi-SCNet backbone network that integrates convolution and transformer interactions to mine both local and global features simultaneously. Building on this, we introduced an Attention Fusion Module (AFM) that enhances feature extraction in both spatial and channel dimensions through AFM, focusing more on edge areas and small targets. Finally, the bitemporal features obtained are fed into the Transformer module to acquire difference information, resulting in a difference feature map. Extensive experiments conducted on the LEVIR-CD and WHU-CD datasets have proven that our algorithm is more competitive.
To meet stringent performance requirements, communication networks are becoming increasingly programmable and flexible, supporting fast and frequent adjustments. However, reconfiguring networks in a dependable and tra...
详细信息
ISBN:
(纸本)9798350383515;9798350383508
To meet stringent performance requirements, communication networks are becoming increasingly programmable and flexible, supporting fast and frequent adjustments. However, reconfiguring networks in a dependable and transiently consistent manner is known to be algorithmically challenging. This paper revisits the fundamental problem of how to update the routes in a network in a (transiently) loop-free manner, considering both the Strong Loop-Freedom (SLF) and the Relaxed Loop-Freedom (RLF) property. We present two fast algorithms to solve the SLF and RLF problem variants exactly, to optimality. Our algorithms are based on a parameterized integer linear program which would be intractable to solve directly by a classic solver. Our main technical contribution is a lazy cycle breaking strategy which, by adding constraints lazily, improves performance dramatically, and outperforms the state-of-the-art exact algorithms by an order of magnitude on realistic medium-sized networks. We further explore approximate algorithms and show that while a relaxation approach is relatively slow, with a local search approach short update schedules can be found, outperforming the state-of-the-art heuristics. On the theoretical front, we also provide an approximation lower bound for the update time of the state-of-the-art algorithm in the literature. As a contribution to the research community, we made all our code and implementations publicly available.
Improving the energy efficiency of Internet Service Provider (ISP) backbone networks is an important objective for ISP operators. In these networks, the overall traffic load throughout the day can vary drastically, re...
详细信息
ISBN:
(纸本)9798350300741;9798350300734
Improving the energy efficiency of Internet Service Provider (ISP) backbone networks is an important objective for ISP operators. In these networks, the overall traffic load throughout the day can vary drastically, resulting in many backbone networks being highly overprovisioned during periods of lower traffic volume. This paper proposes a new Segment Routing (SR)-based traffic engineering algorithm to reduce power consumption of a backbone network in times of low utilization. The traffic steering capabilities of Segment Routing are utilized to remove traffic from as many links as possible. This is used to turn off the corresponding hardware components. Furthermore, it simultaneously ensures that solutions comply to additional operator requirements regarding the overall Maximum Link Utilization in the network. Based on data from a Tier-1 ISP and a public available dataset, we show that our approach allows for up to 70 % of the overall linecards to be switched off, corresponding to an around 56 % reduction of the overall energy consumption of the network in times of low traffic demands.
暂无评论