151398 research outputs found
Sort by
Spectral efficiency analysis of multi-user pinching-antenna systems
No full textThis paper investigates a multi-user pinchingantenna (PA) system, where a single PA is activated on eachwaveguide. With the maximum ratio transmission (MRT) beamforming, the system spectral efficiency (SE) is studied, wherethe inter-user interference term complicates the analysis of theSE. To overcome this obstacle, the stationary phase point method(SPPM) is applied to obtain an analytically tractable form of theSE. The analysis reveals that the average inter-user interferencecan be negligible with a large waveguide spacing even usingthe MRT. This insight makes the simple MRT appealing forPA-based multi-user communications. Finally, the theoreticalanalysis is verified through simulations. Our numerical resultsconfirm that 1) with the aid of SPPM, the approximation of thesystem SE is accurate; 2) and while increasing the waveguidespacing helps reduce the average inter-user interference, it mightdegrade the SE due to the increased signal propagation path loss
Deterministic equivalent-based spectral efficiency of cell-free massive MIMO
No full textThis paper studies a cell-free massive multiple-inputand multiple-output (CF-mMIMO) architecture operating in anopen radio access network, where edge distributed units (EDUs)and user-centric distributed units (UCDUs) jointly performthe physical-layer functions—such as channel estimation andprecoding—while each open radio unit (ORU) solely handlesradio frequency transmission and reception toward the userequipment (UE). Leveraging large-dimensional random matrixtheory, we obtain a deterministic equivalent (DE) expressionfor the ergodic sum spectral efficiency (SE) under imperfectstatistical channel state information (S-CSI). This result enablesus to formulate an ergodic sum SE maximization problemthat is solved by tuning the regularization parameter in alocal partial regularized zero-forcing (LP-RZF) precoder andapplying large-scale fading (LSF)-based deployment of EDUORU pairs together with ORU-UE association. Numerical resultsvalidate that our DE-based result is tight and that the LP-RZFstrategy outperforms benchmark alternatives. Interestingly, thereduced computational complexity is achieved with acceptableperformance loss
'The consequences of your actions': political apology and the mommy myth as discursive punishment in the Shamima Begum case
No full textShamima Begum – one of three schoolgirls who travelled to Syria to join ISIS in 2015 – has been the subject of many intersectional analyses that have highlighted the gendered, racist, and Orientalist frames that are deployed in relation to her story. What has been unravelled, so far, is the positioning of Begum within a narrative that functions to strip her of political agency. Building on this literature, I seek to take this examination of Begum’s case further and argue that this gendered treatment also functions to punish Shamima. This article traces two specific discourses through which this punishment is operationalised. Firstly, the insistence on a political apology from Begum for the crimes of the so-called Islamic State, positions her as a proxy for the terrorist organisation. These apologies, which are doomed to fail, leave Begum as culpable for the atrocities of ISIS. Secondly, alongside the demand for apology, I argue that a discourse of the mommy myth is weaponised to further seal Begum’s punishment. Thus, this article shows that even though her political agency is erased, Begum is still firmly placed as a threat via this legitimation of punishment, the strength of which has, ultimately, rendered her stateless.
ConsensusXAI: A framework to examine class-wise agreement in medical imaging
No full textExplainable AI (XAI) is essential for trust and transparency in deep learning, especially in medical imaging. Existing local explanation methods provide per-instance insights but fail to show whether similar explanations hold across samples of the same class. This limits global interpretability and demands time-consuming manual review by clinicians to trust models in practice. We introduce the Consensus Alignment Score (CAS), a novel metric that quantifies consistency of explanations at the class level. We also present ConsensusXAI, an open-source, model- and method-agnostic framework that evaluates explanation agreement quantitatively (via CAS) and qualitatively (through consensus heatmaps) per class. Unlike prior benchmarks, ConsensusXAI uses a latent-space clustering approach, Latent Consensus, to identify dominant explanation patterns, exposing biases and inconsistencies towards certain classes. Evaluated across two different medical imaging modalities for both correct and incorrect predictions on two different backbones, our method consistently reveals meaningful class-level insights, outperforming traditional consensus method i.e. SSIM, and enabling faster, more confident clinical adoption of AI models
A comparative study towards designing a hybrid architecture of microservices and LLM-based multi-agent systems
No full textCompressive DoA estimation based on a reconfigurable metacavity antenna with selected measurement modes
No full textIn this paper, we present a compressive direction-of-arrival (DoA) estimation technique using a single-port reconfigurable metacavity antenna (RMA) with selected measurement modes. The RMA is an electrically over-sized metacavity with its back wall replaced by an 1-bit reconfigurable metasurface and its top surface etched with leaky cross-shaped irises. By manipulating the on/off states of the PIN diodes loaded on the metasurface, the excitations to the leaky cross-shaped irises are altered, resulting in spatio-temporally random radiated fields that can serve as measurement modes in compressive DoA estimation. The physical layer compression achieved by the RMA substantially reduces the number of channels required for DoA estimation. Simulated experiments are conducted to validate the feasibility of the proposed RMA-based compressive DoA estimation method with three far-field sources. Additionally, a comparison experiment using both the selected and randomly generated measurement modes is conducted. It is demonstrated that employing the measurement mode selection method can significantly improve the DoA estimation accuracy without sacrificing the efficiency
Robust and resilient satellite communication
No full textThis paper presents a modular framework for secure and resilient satellite communication in Low Earth Orbit (LEO), integrating post-quantum cryptography (PQC), anomaly detection, and trust-aware federated learning (FL). We implement a hybrid cryptographic stack combining Hamming Quasi-Cyclic (HQC) and dual-mode Advanced Encryption Standard (AES) for both confidentiality and side-channel resilience. To detect cyber–physical anomalies, we construct a synthetic telemetry dataset and evaluate lightweight models, finding that a combined convolutional neural network (CNN) and gated recurrent unit(GRU) architecture (CNN–GRU) achieves the best trade-off between accuracy and minority-class recall. We then deploy the anomaly detector in a federated learning setting and propose a trust-based aggregation method that gradually downweights poisoned updates. Compared to static defences, our approach maintains high accuracy even when malicious clients are in the majority. These results highlight the importance of a multi-faceted security strategy grounded in cryptography–machine learning (ML) integration for autonomous and trustworthy space systems
Optimal scheduling of digital product innovation: a case study of Yonyou
No full textDigital product innovation is crucial for firms' competitive advantages in digitalisation. This process is inherently multi-staged with continuously new module introduction beyond existing scope in response to market demand. While methods for decision-making and planning in digital product innovation are provided, the scheduling of concrete module introduction for effective innovation remains underexplored. We formulated an optimal scheduling model based on a digital product: Yonyou U8 software. The model, with the optimisation goal of minimising innovation cost for a timely market response, aims to output a reasonable module introduction sequence under constraints on capability resources and man-hours. The goal is fully utilisation of capability resources with the least idle man-hour time. Using an applied mixed-integer programming algorithm, it has been proven effective through empirical data. We identify key decision variables and provide methods for planning product innovation actions. It offers practical value for similar firms to to conduct digital product innovation.<br/