stochastic channel model design

This work addresses the challenge that traditional machine learning methods struggle to model time-varying bidirectional wireless channels with a dynamic number of multipath components and often lack statistical consistency. To overcome these limitations, the authors propose a statistics-informed hybrid TimesNet-TimeFilter model that constructs a learnable graph structure by selecting the top-M strongest multipath components and integrates channel statistical characteristics into the training process. This approach effectively circumvents the constraint of fixed input and output dimensions inherent in conventional methods. Evaluated on both synthetic stochastic channels and ray-tracing datasets, the proposed method significantly outperforms current state-of-the-art baselines, achieving high-fidelity and statistically consistent bidirectional channel prediction.

This work addresses wireless network topology inference solely from anonymized node transmission timing observations—without signaling interaction or packet content. Methodologically, it models the problem using state-visit sequences of multiple anonymous Markov chains and, for the first time, formalizes topology inference as consistent estimation of transition matrices under operator norm convergence. It proposes a statistically consistent estimator with provable error bounds, overcoming limitations of conventional causal measures such as transfer entropy. The approach integrates discrete-time finite-state Markov modeling, joint statistical analysis of anonymized multi-chain observations, and structured matrix parameter estimation. Experiments demonstrate that the method significantly outperforms transfer-entropy-based baselines in topology recovery accuracy across networks of varying scales and under high congestion; moreover, its sample complexity scales nearly linearly with the number of nodes.

Wireless channel modeling for communications and radar systems suffers from heavy reliance on high-quality labeled data, poor generalization, and limited physical interpretability. To address these challenges, this paper proposes a Sparse Bayesian Generative Modeling (SBGM) framework that explicitly incorporates physical priors. Specifically, it is the first to embed the inherent compressibility of wireless channels into a generative model, enabling online learning from compressed measurements. Physical constraints—derived from electromagnetic propagation principles—and parameterized channel representations are integrated to ensure model transparency and interpretability. Moreover, the method supports zero-shot transfer across antenna configurations and frequency bands without retraining. Experimental results demonstrate that SBGM achieves high-fidelity reconstruction of channel parameter distributions using only a small number of compressed measurements. This significantly reduces data acquisition and labeling overhead while markedly improving environmental adaptability and cross-scenario generalization performance.

This work addresses three key challenges in wireless communications: difficulty in modeling complex signal distributions, severe hardware impairments due to non-ideal transceivers, and poor generalizability of conventional constellation shaping techniques. To this end, we propose the first AI-native, robust physical-layer framework leveraging denoising diffusion probabilistic models (DDPMs). Our method jointly learns channel estimation, signal denoising, and constellation optimization via end-to-end generative modeling of realistic channel and hardware distortion distributions. The core contribution lies in exploiting DDPMs’ implicit distribution modeling capability to enhance out-of-distribution robustness. Experimental results demonstrate a 30% reduction in bit error rate under non-ideal hardware conditions and superior performance over traditional approaches in dynamic constellation shaping tasks—highlighting strong generalization and resilience to hardware-induced distortions.

Telecommunications engineering graduate students often lack foundational knowledge in probability theory and stochastic processes, hindering their mastery of teletraffic analysis. Method: This work develops a balanced theoretical–practical pedagogical framework centered on classical queueing models (e.g., M/M/1, M/G/1) and stochastic processes (e.g., Poisson processes, Markov chains, steady-state analysis), integrated with contemporary telecommunications use cases—including traffic modeling, resource allocation, and flow management—and reinforced through numerical simulation exercises. A structured background remediation module addresses prerequisite gaps. Contribution/Results: The resulting textbook has been adopted as a core course resource at multiple universities worldwide, demonstrably enhancing students’ practical competencies in performance modeling and optimization of communication systems.

This work addresses the limitations of conventional channel knowledge graphs (CKGs), which capture only static environments and thus struggle to model time-varying channels induced by dynamic scatterers, terminal orientation changes, and radio-frequency impairments—leading to prohibitively high overhead in acquiring high-dimensional channel state information. To overcome this, the paper proposes a Dynamic Channel Knowledge Graph (Dynamic CKG), establishing for the first time a systematic theoretical framework that serves as an intermediate representation layer bridging static environmental priors and physical-layer signal processing. This framework enables joint pilot design, interference mitigation, and integrated sensing and communication. By integrating geospatial data, time-varying channel modeling, and machine learning–driven graph construction, the approach achieves co-design of CKG and signal processing, significantly reducing channel acquisition overhead while enhancing both communication efficiency and sensing performance, thereby offering a novel paradigm for 6G systems.

Accurate and computationally tractable SINR coverage analysis in Poisson cellular networks remains challenging due to inherent trade-offs between precision and analytical feasibility. Method: This paper proposes a hybrid approximation framework: Monte Carlo sampling for dominant near-field interferers, and Laplace functional modeling for the residual far-field interference. Contribution/Results: The approach eliminates reliance on nested integrals and special functions in classical stochastic geometry models, while avoiding failure modes of probabilistic interference models under missing interference moments or restrictive parameter assumptions. Its modular design ensures numerical stability and path-loss independence, and—uniquely—provides a theoretically derived error bound that converges as the number of dominant interferers increases. Experiments demonstrate high accuracy and low computational overhead under both noise-limited and interference-limited regimes, with strong robustness and consistency across diverse channel conditions and network deployment parameters.

This work addresses the challenges posed by model mismatch, data scarcity, adversarial perturbations, and distribution shifts in wireless sensing and communication systems by proposing a unified robust signal processing framework. Integrating techniques from robust statistics, distributionally robust optimization, and adversarial training, the framework systematically characterizes the trade-off between performance and robustness. The proposed approach is evaluated across several critical tasks—including robust ranging and localization, multimodal sensing, receive combining, and waveform design—demonstrating significant improvements in system reliability under non-ideal conditions. Experimental results validate the effectiveness and broad applicability of the framework in enhancing resilience against diverse sources of uncertainty inherent in practical wireless environments.

This work proposes a novel Bayesian-optimal iterative signal recovery algorithm for multiuser linear Gaussian communication systems with randomly right unitarily invariant precoding. Built upon the Orthogonal Approximate Message Passing (OAMP/VAMP) framework, the method achieves efficient iterative updates through interpolation between Expectation Propagation (EP) and OAMP, enabling, for the first time, Bayesian-optimal reconstruction of signals with non-separable priors. The authors innovatively introduce a disorder-averaging technique combined with the replica-symmetric ansatz to establish a finite-sample high-dimensional analysis of the algorithm. Theoretical analysis demonstrates that the proposed algorithm attains Bayesian optimality in the large-system limit and aligns precisely with replica-symmetric predictions, exhibiting superior performance in multiuser communication scenarios.

In millimeter-wave (mmWave) MIMO downlink systems, sporadic, high-power impulsive interference—arising from hardware imperfections or external sources—severely degrades conventional channel estimation performance. To address this, we propose a novel variational inference framework jointly modeling the channel and impulsive interference via sparse Bayesian learning (SBL). Leveraging angular-domain channel sparsity and the intermittent nature of impulsive interference, our method performs coupled Bayesian inference of channel parameters and interference statistics under a mean-field approximation, enabling effective interference separation and enhanced estimation robustness. Simulation results demonstrate that the proposed approach consistently outperforms existing baselines across interference intensities, reducing average channel estimation error by 35%–52%. Notably, it maintains high-accuracy channel recovery even under strong interference. This work establishes an interpretable, scalable statistical inference paradigm for robust mmWave communications.