CubeSats have gained significant attention due to their low cost and versatility, yet their low inertia makes them particularly vulnerable to disturbances, such as magnetic torques and residual atmospheric drag, that induce strong nonlinearities in their rotational dynamics. While nonlinear controllers can address these challenges, their high computational burden and intricate stability analyses often forces practitioners to rely on linearization-based techniques, which fail to capture the full extent of the system’s nonlinear behavior. Moreover, existing Koopman-based approaches for attitude control have not exploited the quaternion representation intrinsic to CubeSat dynamics, nor have they adequately addressed the impact of environmental disturbances and sensor noise, an important gap in current research. In this paper, we propose a neural Koopman framework that lifts the nonlinear dynamics, represented in quaternion form, into a higher-dimensional linear space. By training deep neural networks as lifting functions using a control-aware loss function, our method enables the design of efficient linear controllers such as LQR. Comparative results demonstrates that the Koopman model preserves the unit-norm property of quaternions under sensor noise and disturbances and delivers superior reference tracking performance compared to the conventional linearized model, which suffers from renormalization issues. These findings highlight the potential of the proposed framework for achieving robust and computationally efficient CubeSat attitude control.

Accurate traffic prediction plays a vital role in intelligent transportation systems by enabling efficient routing, congestion mitigation, and proactive traffic control. However, forecasting is challenging due to the combined effects of dynamic road conditions, varying traffic patterns across different locations, and external influences such as weather and accidents. Traffic data often consists of several interrelated measurements—such as speed, flow and occupancy—yet many deep-learning approaches either predict only one of these variables or require a separate model for each. This limits their ability to capture joint patterns across channels. To address this, we introduce the Multi-Channel Spatio-Temporal (MCST) Mamba model, a forecasting framework built on the Mamba selective state-space architecture that natively handles multivariate inputs and simultaneously models all traffic features. The proposed MCST-Mamba model integrates adaptive spatio-temporal embeddings and separates the modeling of temporal sequences and spatial sensor interactions into two dedicated Mamba blocks, improving representation learning. Unlike prior methods that evaluate on a single channel, we assess MCST-Mamba across all traffic features at once, aligning more closely with how congestion arises in practice. Our results show that MCST-Mamba achieves strong predictive performance with a lower parameter count compared to baseline models.

Federated learning (FL) has gained attention for enabling efficient distributed learning while maintaining data privacy. However, the data privacy constraint reduces the transparency in the agents’ model update making the learning process vulnerable to Byzantine attacks. In this paper, a mathematical proof is provided to show that when the traditional model-combining scheme is used, the model will eventually diverge to non-useful solutions in the presence of Byzantine agents independently from their number or their contributions. A low complexity norm-control based aggregation approach is also proposed and shown to converge to the optimal and sub-optimal solutions in the absence or presence of Byzantine nodes, respectively. Monte-Carlo simulations are also conducted to verify and validate the mathematical derivations and the efficiency of the proposed approach in protecting the FL model.

The identification of malicious users within a large set of participants poses a significant challenge in the domains of cybersecurity, data integrity, user management, and particularly within federated learning (FL) environments. FL, a distributed machine learning approach, necessitates rigorous mechanisms for safeguarding data integrity, model accuracy by effectively managing and identifying malicious participants. Traditional methods require the sequential removal and evaluation of users to determine their impact on the system’s overall error rate or loss function, fall short in terms of efficiency and scalability, especially in FL contexts where data is distributed across multiple clients. To address these limitations, we propose the Refine and Identify Algorithm, a two-phased approach that efficiently narrows the search space for identifying malicious users by initially evaluating users in groups rather than individually and iteratively focusing on those groups with the highest potential for containing malicious users. A rigorous mathematical framework, including a proof of convergence and a detailed analysis of iteration necessities, underpins the algorithm’s efficacy. The convergence proof and analysis of iteration requirements provide a solid mathematical foundation for the proposed method’s effectiveness, paving the way for further optimization and application-specific tuning. Simulation results depict the efficiency of the proposed technique and show a significant reduction in computational resources and time required for identifying malicious users.