搜索结果：Routers

Under the general trend of global energy transformation, the proportion of renewable energy in the power sector continues to increase. Power routers are of great significance for improving energy utilization efficiency and ensuring the stable operation of power systems. However, the intermittent and uncertain nature of distributed energy makes energy management of power routers difficult, and traditional optimization methods are also difficult to adapt. Therefore, this study proposes the integration of Proximal Policy Optimization with a multi-agent framework, combined with a Generative Adversarial Imitation Learning based on a double-buffer mechanism. The double-buffer mechanism is used to improve data utilization efficiency and training stability, and to optimize communication and collaboration among multiple agents, thereby realizing energy collaborative optimization of power routers. Experimental results show that after 420 trainings, the average round reward of the improved algorithm is stable at about -410, and the strategy loss function is the first to stabilize after 500 times. In practical scenarios, the proposed model maintains a DC bus voltage fluctuation range between 728V and 732V. Additionally, its electricity cost amounts to 3846.36 yuan, and its total runtime is 53.32 seconds-both of which are lower than those of the other two models. Overall, the enhanced algorithm and model notably improve the energy collaboration optimization of power routers, offering a practical solution to energy management issues and significantly advancing the progress in this area.

Named Data Networking (NDN) represents a paradigm shift toward content-centric architectures but remains critically vulnerable to Interest Flooding Attacks (IFAs), where malicious actors overwhelm router Pending Interest Tables with spurious requests, causing service degradation and denial-of-service. To address the limitations of existing approaches, including high false positives in threshold-based methods and substantial overhead in centralized learning, we propose FL-IFAshield, a novel federated learning framework for adaptive IFA mitigation. Our solution integrates dynamic Poisson-EMA thresholding for accurate flood detection, entropy-aware federated aggregation to handle non-IID traffic distributions across edge routers, and Byzantine-robust mechanisms with differential privacy guarantees. Comprehensive evaluation on the FIT/IoT-LAB testbed with 100 routers demonstrates exceptional performance: 93.1% F1-score in attack detection, only 5% false positives, 28 ms average end-to-end latency ([Formula: see text]), and over 90% legitimate Interest Satisfaction Ratio under sophisticated collusive attacks, while maintaining minimal computational overhead (<9% CPU utilization on ARMv8 routers). FL-IFAshield significantly improves security performance, offering 35% higher accuracy than static thresholding and 60% lower communication overhead than centralized approaches. While simpler heuristic baselines naturally incur marginally lower computational footprints, our solution delivers the optimal overall operational balance among high precision, low end-to-end latency ([Formula: see text]), and resource efficiency in constrained edge computing environments.

Metasurfaces integrated onto guided-wave photonic systems have been investigated for enabling advanced functionalities such as point-by-point optical extraction and manipulation of amplitude, phase, and polarization. However, achieving full control over the spectrum (i.e., wavelength/frequency) of on-chip light remains a challenge, limiting their widespread application in integrated photonics. Here, we propose and experimentally demonstrate an on-chip metasurface color router by leveraging symmetry-broken quasi-bound states in the continuum (q-BICs) mode. By precisely engineering the on-chip meta-diatom pairs with controlled scaling and asymmetry, we simultaneously achieve modulation of both extraction intensity and narrowband spectral extraction of the out-coupled lightwave. As a proof of concept, we realize several on-chip multiplexed color routers through spatial mapping and cascading of distinct q-BIC-assisted meta-diatom pixels, capable of selectively guiding and routing primary wavelengths into free space from different spatial positions along the waveguide. Crucially, due to the on-chip optical propagation scheme, these color routers, enabled by nonlocal metasurfaces, exhibit spatial multiplexing but with a significant improvement in the energy utilization efficiency (EUE) compared with conventional designs. We envision that such on-chip q-BIC-assisted metasurface color routers, with their potential for miniaturized integration, could open new avenues for advanced applications in multiplexed information routing, intelligent integrated photonic systems, and next-generation wearable display technologies.

All digital Wireless Communication (WC) electromagnetic field (EMF)/radiation (EMR) signals (from mobile/"smart" phones and corresponding base antennas, cordless domestic phones, Wireless Fidelity (Wi-Fi) routers, "Bluetooth" wireless connection among electronic devices, etc.) are emitted discontinuously, in the form of on/off pulses repeated at various Extremely Low Frequency (ELF) rates. Yet, many scientists ignore/underestimate these ELF pulsations, and characterize all WC emissions simply as Radio Frequency (RF)/Microwave (MW) signals. Here, we provide recordings of ELF pulsations with respect to time, emitted by the most common WC devices, specifically Wi-Fi router, 4th and 5th Generation (4G, 5G) mobile phones. We used a broadband antenna, connected to an RF spectrum analyzer (SA), calibrated the SA at the signal's carrier MW frequency and recorded the power of the final emitted RF/MW signal with respect to time. We recorded emissions at 10&#x2009;ms, 100&#x2009;ms, 1 s, and 2 s sweep times, capturing the pulses repeated at various ELF rates, clearly showing the ELF pulsing emissions from the WC devices. As in all real WC EMF signals emitted by commercially available devices and corresponding antennas, there is intense variability in the amplitude, shape, duration, and repetition frequency of the pulses. The present study, in combination with the Ion Forced Oscillation and Voltage-Gated Ion Channel (IFO-VGIC) mechanism of non-thermal EMF-bioeffects, imply that the non-thermal biological and health effects of WC EMFs are induced by the ELF pulsation, modulation and variability, and not by the standalone (non-modulated) RF carrier wave EMFs which can produce only heating. Electromagnetic fields/radiation (EMFs/EMR) emitted by digital wireless communication (WC) devices (mobile/&#x201c;smart&#x201d; phones, cordless domestic phones, Wireless Fidelity (Wi-Fi) routers for connection to the Internet, etc.) and corresponding base antennas, are usually referred to simply as Radio Frequency (RF: 300&#x2009;kHz&#x2013;300&#x2009;GHz) EMFs/EMR. Yet, as we have repeatedly declared before, this reflects only one part of the reality. The other part is that the RF signals that carry the transmitted information (text, speech, music, images, video, etc.) are contained within on/off pulses which are emitted/repeated at various Extremely Low Frequency (ELF: 3&#x2013;3000&#x2009;Hz) rates. Moreover, the RF signal within the pulses is modulated mostly by ELF EMFs, and the final signal contains random variability especially in amplitude which lies mainly in the Ultra-Low Frequency (ULF: 0&#x2013;3&#x2009;Hz) band. Therefore, all WC EMFs are a combination of high (RF) and low (ELF/ULF) frequencies. To record the ELF pulses, specific methodology and instrumentation are required. Here, we have recorded these pulses (as power with respect to time) from the most common WC EMFs/EMR, namely Wi-Fi, 4th, and 5th Generation (4G, 5G) WC EMFs/EMR. The present study, in combination with a widely accepted biophysical mechanism, the Ion Forced Oscillation and Voltage-Gated Ion Channel (IFO-VGIC) mechanism, of non-thermal EMF-bioeffects, imply that the non-thermal biological and health effects associated with exposure to WC EMFs are induced by the ELF/ULF pulsation, modulation and variability, and not by the standalone (non-modulated) RF carrier wave EMFs which can produce only heating at adequately high intensities rarely found in the environment.

Alzheimer's disease (AD) is a progressive neurodegenerative disease with high inter-patient variance in rate of cognitive decline. AD progression prediction aims to forecast patient cognitive decline and benefits from incorporating multiple neuroimaging modalities. However, existing multimodal models fail to make accurate predictions when many modalities are missing during inference, as is often the case in clinical settings. To increase multimodal model flexibility under high modality missingness, we introduce PerM-MoE, a novel sparse mixture-of-experts method that uses independent routers for each modality in place of the conventional, single router. Using T1-weighted MRI, FLAIR, amyloid beta PET, and tau PET neuroimaging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI), we evaluate PerM-MoE, state-of-the-art Flex-MoE, and unimodal neuroimaging models on predicting two-year change in Clinical Dementia Rating-Sum of Boxes (CDR-SB) scores under varying levels of modality missingness. PerM-MoE outperforms the state of the art in most variations of modality missingness and demonstrates more effective utility of experts than Flex-MoE.

Previous studies on topological dual-band frequency routers have focused on routing signals separately within two distinct frequency bands. However, the potential of dual topological bands has not been fully explored, particularly the possibility of each band supporting multiple frequency channels. Here, we propose a dual-band multi-frequency wave routing based on photonic square-root topological insulators. By introducing tailored perturbations into a decorated honeycomb lattice, we open two band gaps and design two distinct topological interfaces to obtain multi-frequency topological edge states. As a result, our single structure enables the routing of four different frequency ranges. This work provides a new approach for advanced integrated devices based on photonic square-root topological insulators.

We propose a versatile on-chip mode manipulation architecture based on arbitrary-to-single-mode power dividers (ASPDs), enabling flexible mode conversion and routing in multimode photonic systems. The architecture comprises two oppositely cascaded ASPD units arranged in a Mach-Zehnder-interferometer-like configuration, leveraging standard single-mode components such as phase shifters and mode routers to achieve highly flexible multimode manipulation. This approach offers two key advantages: a compact, scalable design based solely on ASPD units and standard single-mode components, and direct decomposition of higher-order modes into the fundamental mode, simplifying routing and removing the need for complex multimode elements. As a proof of concept, several ultra-compact ASPDs were designed using subwavelength grating structures optimized via intelligent algorithms. For commonly used low-order modes (TE0-TE3), the devices exhibit footprints below 9.5&#x2005;&#xb5;m and maintain excess losses (ELs) below 0.35&#x2005;dB across a 300&#x2005;nm bandwidth (1400-1700nm). Notably, TE0/TE1 ASPDs achieve ELs < 0.15&#x2005;dB over a record-wide 500&#x2005;nm range (1300-1800nm). Based on these ASPDs, we experimentally demonstrate flexible mode conversion and switching, confirming their effectiveness for multimode signal control. Furthermore, numerical results show that the scheme can be extended to higher-order modes (TE4-TE7) while maintaining excellent performance and compact footprints (8&#x2005;&#xb5;m), and can be combined with standard single-mode crossings for flexible multimode interconnections. These results validate the feasibility and scalability of the proposed ASPD-based platform, offering a promising route toward reconfigurable multimode photonic integrated circuits.

Engineering synthetic gauge fields to realize Landau levels for neutral particles has emerged as a pivotal strategy for advanced classical wave manipulation. While out-of-plane pseudomagnetic field (PMF) generates conventional, non-chiral Landau levels with flat-band zeroth modes, their in-plane counterparts unlock valley-locked chiral Landau levels (CLLs) with linear dispersion and topologically protected unidirectional propagation. Here, we propose and demonstrate the chiral Landau rainbow effect in a 2D honeycomb photonic crystal by synergistically engineering an in-plane PMF and a parallel pseudoelectric field (PEF). PMF, from a spatially graded Dirac mass, generates CLLs. PEF, from a graded band structure shifting, induces frequency-dependent spatial dispersion of these CLLs, forming discrete chiral Landau rainbows. Their confinement and trapping position can be independently tuned by controlling the strengths of the PMF and PEF. This work offers a new paradigm for multifunctional topological wave control, with applications in robust slow-light devices, valley-multiplexed routers, and fault-tolerant integrated photonic circuits.

In practical signal processing and optical computing, multiple signals in different frequency bands usually require nonreciprocal transmission. However, to date, most proposed nonreciprocal systems based on nonlinear resonators only support a single working band. Here, we demonstrate that nonreciprocal light transmissions can be realized in multiple distinct spectral bands, solely via a single nonlinear multimode nanocavity triggered by a control pulse. More importantly, our results further show that the multiband nonreciprocal light transmission is reconfigurable, with high contrasts and reasonably broad per-band bandwidths. The underlying physical mechanism is revealed. The approach presented here is promising in the fields of optical signal processing, quantum computing, optical routers, and rectifiers.

Mode-division multiplexing has emerged as an effective approach to increase the capacity of on-chip optical interconnects. Efficient routing and add-drop manipulation of different modes are therefore essential for scalable multimode photonic integrated circuits. However, implementing compact mode routers with low loss, low inter-mode crosstalk, and high-speed signal compatibility remains challenging on the silicon photonic platform. In this work, we propose and experimentally demonstrate a compact four-port optical mode router on a silicon-on-insulator platform. The device enables mode-selective and bidirectional add-drop routing for multiple modes by integrating multimode waveguide crossings, multimode waveguide bends, and mode-selective microring resonators. Low-loss and low-crosstalk operation is achieved through carefully designed mode-matching and multimode waveguide engineering. Experimental results show correct routing for three supported modes with inter-mode crosstalk below -15.4&#x2009;dB and an insertion loss below 6.6&#x2009;dB. High-speed data transmission at 40&#x2009;Gbit/s is demonstrated for all routing paths, resulting in an on-chip aggregate data capacity of 480&#x2009;Gbit/s. The proposed architecture provides a scalable solution for high-capacity mode-division-multiplexed optical interconnects on a silicon photonic platform.

Visual perception of nighttime images is often impaired by degradations: low-light, noise, motion blur, and low-resolution. While recent methods have made progress in jointly solving these degradations, the diversity of patterns and intensities in degradation has not been properly considered, leading to inconsistent illumination and unintended artifacts. In response, we propose to integrate perceptual cues with mixture-of-experts (IPCMoE) to achieve flexible processing for low-light low-quality images. By exploiting the perceptual cues, we strategically combine dedicated experts with the selective collaboration approach for feature enlightening and texture restoration. To this end, we develop perceptual-integrated MoEs by designing customized routers and task-dependent experts. Specifically, the texture memorial MoE is developed to preserve valuable features to restore high-fidelity details, and the enhancement MoE that adaptively integrates enlightening cues and texture cues is designed to formulate the relationship between feature enlightening and texture restoration, thereby achieving dynamic image processing. Compared to state-of-the-art models, our IPCMoE achieves superior performance on various benchmarks for handling complex low-light scenes.

Asymmetric light transmission at the nanoscale constitutes a fundamental challenge and a key goal in integrated photonic circuits. Polaritons in van der Waals (vdWs) materials offer a promising platform for such nanoscale light confinement and control. However, achieving complex and functional vdWs polaritons control across interfaces remains a challenge. Here, we demonstrate grating-driven on-chip asymmetric steering of phonon polaritons (PhPs) in hyperbolic vdWs crystal &#x3b1;-MoO3 bilayers that incorporate twisted stacks and engineered interfaces. By combining twist-induced dispersion engineering and polaritonic grating diffraction, we achieve tunable deflection and the asymmetric transmission of PhPs. PhP transmission can be switched between bidirectional and unidirectional modes by varying either the excitation frequency or the grating period. We further designed an asymmetric polaritonic lens that focuses forward-propagating PhPs while allowing normal backward transmission. These results provide a novel strategy for polaritonic steering and lay a solid foundation for designing on-chip optical isolators, diodes, and nonreciprocal routers.

In recent years, there has been increasing investment in the deployment of massive commercial Low Earth Orbit (LEO) constellations to provide global Internet connectivity. These constellations, now equipped with inter-satellite links, can serve as low-latency Internet backbones, requiring LEO satellites to act not only as access nodes for ground stations, but also as in-orbit core routers. Due to their high velocity and the resulting frequent handovers of ground gateways, LEO networks highly stress mobility procedures at both the sender and receiver endpoints. On the other hand, a growing trend in networking is the use of technologies based on the Information Centric Networking (ICN) paradigm for servicing IoT networks and sensor networks in general, as its addressing, storage, and security mechanisms are usually a good match for IoT needs. Furthermore, ICN networks possess additional characteristics that are beneficial for the massive LEO scenario. For instance, the mobility of the receiver is helped by the inherent data-forwarding procedures in their architectures. However, the mobility of the senders remains an open problem. This paper proposes a comprehensive solution to the mobility problem for massive LEO constellations using the Named-Data Networking (NDN) architecture, as it is probably the most mature ICN proposal. Our solution includes a scalable method to relate content to ground gateways and a way to address traffic to the gateway that does not require cooperation from the network routing algorithm. Moreover, our solution works without requiring modifications to the actual NDN protocol itself, so it is easy to test and deploy. Our results indicate that, for long enough handover lengths, traffic losses are negligible even for ground stations with just one satellite in sight.

Large Reasoning Models (LRMs) significantly improve the reasoning ability of Large Language Models (LLMs) by learning to reason, exhibiting promising performance in solving complex tasks. However, their deliberative reasoning process leads to inefficiencies in token usage, memory consumption, and inference time. Thus, this survey provides a review of efficient inference methods designed specifically for LRMs, focusing on mitigating token inefficiency while preserving the reasoning quality. The overview structure of this paper is shown in Figure 1. First, we introduce a taxonomy to group the recent methods into two main categories: (a) explicit compact Chain-of-Thought (CoT), which reduces tokens while keeping the explicit reasoning structure, and (b) implicit latent CoT, which encodes reasoning steps within hidden representations instead of explicit tokens. Meanwhile, we discuss their strengths and weaknesses. Then, we conduct empirical analyses on existing methods from reasoning scenarios, object functions, and performance & efficiency aspects. Besides, we present open challenges in this field, including human-centric controllable reasoning, trade-off between interpretability and efficiency of reasoning, ensuring the safety of efficient reasoning, and broader applications of efficient reasoning. In addition, we highlight key insights for enhancing LRMs' inference efficiency via techniques such as model merging, new architectures, and agent routers. We hope this work serves as a valuable guide, helping researchers overcome challenges in this vibrant field. A collection of efficient reasoning methods for LRMs (papers and codes) is provided at this link: https://github.com/yueliu1999/Awesome-Efficient-Inference-for-LRMs.

Network traffic prediction is a critical technology for next-generation intelligent routers, enabling network managers to effectively plan resources and address bandwidth and latency challenges posed by rapidly growing data applications and video traffic. While time series models are commonly employed for traffic matrix (TM) prediction, existing spatial modeling approaches-including Convolutional Neural Networks (CNNs), Graph Convolutional Networks (GCNs), and Graph Attention Networks (GATs)-exhibit significant limitations. These models often fail to accurately capture complex nonlinear spatial relationships between network nodes, primarily because current attention score calculation methods rely on input-dependent feature similarities that impose linear constraints in log-space. To address these fundamental limitations, we propose the Graph Self-learning Attention Scores Network (GSASN), which autonomously learns attention scores between same-origin or same-destination nodes through a learnable parameter matrix, effectively breaking the linear constraints of conventional approaches. Experimental results on two real-world datasets demonstrate that GSASN achieves substantial improvements over GAT: up to 24.7% in Root Mean Square Error (RMSE), 36.5% in Mean Absolute Error (MAE), and 7.3% in coefficient of determination (R2). Furthermore, we present a novel spatio-temporal model (ST-GSASN) that integrates GSASN with temporal learning components through gated fusion. Extensive experiments reveal that ST-GSASN outperforms both baseline and state-of-the-art methods, including recent transformer-based approaches (ASTGNN, STTN), while maintaining superior computational efficiency with only 5% of the parameters required by comparable models. Our source code is publicly available at https://github.com/iamtcs/GSASN.

Precise control of a single photon transport in broadband, multi-mode waveguides is a fundamental challenge for scalable quantum networks. We propose a theoretical scheme for on-demand control of single-photon scattering using a driven &#x39b;-type emitter coupled to a rectangular waveguide. By employing the Lippmann-Schwinger formalism, we derive the exact analytical scattering matrix and reveal two key interference mechanisms: electromagnetically induced transparency for complete transmission and Fano resonance for complete reflection. We demonstrate that the single-photon scattering is dynamically engineered by the driving field, enabling a switch between complete transmission and dual-frequency complete reflection. Crucially, in the multi-mode regime, we show that the scattering is governed by quantum interference between modes, making it critically dependent on the input photonic state. By preparing the photon in a specific coherent superposition state, the multi-mode interference is harnessed to achieve Fano resonance-mediated complete reflection. Conversely, a single-mode input suppresses complete reflection. This input-state-dependent scattering establishes a general framework for multi-mode quantum photonics, paving the way for broadband dual-frequency filters, multi-mode quantum routers, and on-chip spectrometers.

Existing on-link IPv6 address scanning technologies based on IPv6-only information, such as multicast Ping6 scanning, invalid extension header scanning, multicast listener discovery (MLD) scanning, and stateless address auto-configuration scanning, rely on protocol features such as ICMPv6, MLD, and other protocol features to induce responses. However, these scanning packets are easily intercepted by the default security mechanisms of modern OSs (Operating Systems), leading to issues such as low OS coverage and incomplete IPv6 address scan results of alive nodes. On-link IPv6 address scanning technologies based on dual-stack correlation information (e.g., FScan6, LLMNR6, and LinkScan6) utilize local domain names, hostnames, and other dual-stack correlation information to enhance IPv6 address discovery capabilities. However, these technologies overly rely on IPv4 networks, resulting in low IPv6 address scanning efficiency and inability to operate in IPv6-only networks. To this end, we propose HFinder6, an efficient IPv6 address scanning technology for IPv6-only network based on hostname correlation. HFinder6 combines the passive capture of DHCPv6 Solicit messages from active IPv6 nodes with an active triggering mechanism. By constructing Router Advertisement messages of NDP protocol, it forces nodes to proactively send Solicit messages, enabling rapid hostname acquisition. Subsequently, the IPv6 address information is queried in parallel using the mDNS protocol and LLMNR protocol, thereby achieving efficient scanning of IPv6 addresses for alive nodes in IPv6-only network. A typical IPv6-only and IPv4/IPv6 dual-stack network environment was established, comprising 20 versions of OSs such as Windows and Linux. HFinder6 was tested in this environment and compared with 4 IPv6-only network IPv6 address scanning scripts from the Nmap tool and 3 dual-stack network IPv6 address scanning tools (e.g., LinkScan6, LLMNR6, and FScan6). Experimental results show that HFinder6 can discover 43 IPv6 addresses across 18 OS versions in an average of just 10.29 seconds within IPv6-only network. In terms of OS coverage and IPv6 address scanning completeness, HFinder6 performs on par with FScan6 and outperforms 6 other scripts and tools, successfully identifies at more 14 additional OS versions and scans 35 more IPv6 addresses, thereby enhancing the completeness of IPv6 address scanning by up to 5.37 times. Moreover, in terms of IPv6 address scanning efficiency, HFinder6 can identify 3.67 more IPv6 addresses per second, outperforming these 7 scripts and tools.

Low-overlap indoor RGB-D point cloud registration remains vulnerable to hard failures because robust recovery and deployment latency are rarely achieved by one registrar. We present a budget-aware rescue-routing framework that keeps PointDSC+FCGF as the fast primary path and separates deployable pre-rescue gates from frozen-candidate selector analysis. On 3DLoMatch, the frozen selector DRACO-Stack reaches strict success of 0.5205 vs. 0.4278 for PointDSC+FCGF, while the deployable DRACO-Gate reaches 0.4801; a matched accuracy-only CoFiNet reconstruction on the same 1781 pairs reaches 0.5390. On 3DMatch, DRACO-Route activates 369/1623 pairs and reaches 0.8885 at 721.18 ms, compared with 0.8694 for PointDSC+FCGF and 0.9082 at 8310.48 ms for always-on RegTR. Redwood is used as public-transfer validation, where PointDSC+FCGF reaches 0.1425 vs. 0.1043 for PointDSC+FPFH. The results support selective indoor hard-tail rescue under an explicit runtime budget, without claiming a universal scene-free router or a new backbone.