Time-varying Wireless Channel Tracking with Online Parameter Learning via the Birth-Death-Drift Model This work was supported by the Discovery Grants Program of the Natural Sciences and Engineering Research Council of Canada (NSERC), and the startup funding of University of Tennessee.

Tiancheng Gao, Mohamed Akrout, Faouzi Bellili, Amine Mezghani · 2026

Read Time-varying Wireless Channel Tracking with Online Parameter Learning via the Birth-Death-Drift Model This work was supported by the Discovery Grants Program of the Natural Sciences and Engineering Research Council of Canada (NSERC), and the startup funding of University of Tennessee. by Tiancheng Gao, Mohamed Akrout, Faouzi Bellili, Amine Mezghani (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-24727.txt.

Introduction

Tiancheng Gao, Mohamed Akrout, Faouzi Bellili, and Amine Mezghani improve channel tracking for massive MIMO wireless systems. Such systems need accurate channel state information, yet pilots cost overhead. Using temporal correlation over time can cut that cost. An earlier method pairs approximate message passing with a birth death drift channel model. Three flaws block its deployment, though. It assumes ideal Gaussian sensing matrices, perfect knowledge of model parameters, and handles time information only approximately.

The team's algorithm, BDD VAMP EM, answers all three limits at once. It unifies the birth death drift model with vector approximate message passing. An expectation maximisation loop learns the model parameters online, removing the need for prior knowledge. In simulations, the fully automated tracker consistently outperforms existing benchmarks. The gain is clearest when model parameters are mismatched, the case that matters most in practice. Gao and colleagues present this as evidence of real world viability.

Catalog record

CREATORTiancheng Gao, Mohamed Akrout, Faouzi Bellili, Amine Mezghani
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.24727

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Time-varying Wireless Channel Tracking with Online Parameter Learning via the Birth-Death-Drift Model This work was supported by the Discovery Grants Program of the Natural Sciences and Engineering Research Council of Canada (NSERC), and the startup funding of University of Tennessee. Tiancheng Gao, Mohamed Akrout, Faouzi Bellili, Amine Mezghani Abstract Accurate massive MIMO channel state information (CSI) acquisition with low pilot overhead is critical in dynamic propagation environments. Exploiting temporal correlation is key to reducing pilot overhead, yet most existing methods often rely on impractical assumptions. The approximate message passing with side information (AMP-SI) algorithm, built upon a birth-death-drift (BDD) model, represents a significant step in this direction. However, its practical deployment is hindered by three major limitations: reliance on i.i.d. Gaussian sensing matrices, need for perfect BDD parameter knowledge, and a statistically approximate treatment of temporal information. To address these limitations, we introduce BDD-VAMP-EM, a fully automated algorithm that relies on the BDD model, vector AMP (VAMP), and expectation-maximization (EM) in a unified framework. Simulations show that BDD-VAMP-EM consistently outperforms existing benchmarks, particularly under model parameter mismatch, confirming its practical viability. I Introduction In multiple-input multiple-output (MIMO) and its massive-scale extensions, the performance of physical-layer techniques such as high-order spatial multiplexing [ 1 ] and interference suppression [ 2 ] depend critically on the precision and timeliness of channel state information (CSI). In principle, CSI quality can be improved by transmitting longer training/pilot sequences, especially in presence of a large number of unknown channel coefficients in massive MIMO systems. However, excessive pilot transmission consumes significant time-frequency resources, diminishing throughput for actual data and creating an overhead bottleneck [ 3 ] . Conventional solutions that treat each measurement interval independently fail to exploit the inherent temporal correlation [ 4 ] , which manifests in two main drawbacks: an inefficient use of spectral resources [ 5 ] and estimation lag and inaccuracy in tracking fast variations [ 6 ] . To harness the temporal correlation, a rich body of literature has been developed. Classical methods, such as those based on Kalman filters (KF) [ 7 ] or recursive least squares (RLS) [ 8 ] , are effective only when channel dynamics follow simple, pre-defined models. Their performance deteriorates with non-stationary birth-death dynamics of realistic multi-path components because they lack an inherent mechanism to model structured sparsity evolution [ 9 ] . To address this, modern compressed sensing (CS) techniques, including those based on the approximate message passing (AMP) [ 10 ] , have proven powerful by leveraging the spatial sparsity in angular domain. In this line of work, the AMP with side information (AMP-SI) algorithm represents a notable leap forward, which integrates the birth-death-drift (BDD) model to exploit temporal correlations across consecutive estimation intervals [ 11 ] . To derive the posterior for the current channel state, AMP-SI uses a Bernoulli–Gaussian (BG) prior and forms a Gaussian likelihood by merging the AMP extrinsic message and the latest posterior mean and variance, termed “side information” (SI). Despite its conceptual appeal, AMP-SI also faces certain constraints. Inherited from AMP, the requirement on independent and identically distributed (i.i.d.) Gaussian sensing matrices is often violated in practical systems and potentially leads to algorithmic instability [ 12 ] . Also, the performance is contingent upon precise prior knowledge of the BDD model parameters: the absence of an online learning mechanism confines its use to idealized settings [ 13 ] . Moreover, the treatment of temporal information rests on a Gaussian approximation of the entire posterior parameterized solely by the mean and variance, which discards richer statistical structures, especially in low signal-to-noise ratio (SNR) regimes or when the sparsity pattern is uncertain [ 14 ] . This paper proposes a fully automated algorithm for joint channel tracking and parameter learning, coined BDD-VAMP-EM, to address the three core limitations of AMP-SI: • Robustness to practical sensing matrices : we build upon vector AMP (VAMP) [ 15 ] to provide reliable convergence under broader conditions on sensing matrices than AMP. • Autonomy in dynamic environments : we seamlessly integrate an expectation-maximization (EM) [ 16 ] procedure within the message-passing iteration loop to enable online, joint estimation of the BDD parameters and the channel state. • Statistically faithful temporal fusion : we assume the posterior is a Bernoulli Gaussian-mixture (BGM) distribution and propagate the hyper-parameters as SI to allow for a richer statistical structure than moment-based approximations. The remainder of this paper is structured as follows. Section II details the system model and formulates the channel tracking problem. Section III presents the derivation of the proposed BDD-VAMP-EM algorithm. Section IV provides a comprehensive performance evaluation against state-of-the-art benchmarks. Finally, Section V concludes the paper. Notations : We use Sans Serif fonts (e.g., \mathsf{x} ) for random variables and Serif fonts (e.g., x ) for realizations. We use boldface lowercase and uppercase letters for vectors (e.g., \bm{\mathsf{x}} and \bm{x} ) and matrices (e.g., \bm{\mathsf{X}} and \bm{X} ), respectively. Vectors are column-wise by default. \bm{X}^{\textsf{T}} and \bm{X}^{\textsf{H}} refer to the transpose and hermitian of matrix \bm{X} , respectively. \mathbf{I}_{N} denotes the N\times N identity matrix and \mathrm{vec}(\bm{X}) denotes the vectorization operation that stacks the columns of \bm{X} into a vector. Moreover, \mathrm{Pr}\{\cdot\} denotes the probability of an event, while p_{\mathsf{x}}(x;\bm{\theta}) , p_{\bm{\mathsf{x}}}(\bm{x};\bm{\theta}) , and p_{\bm{\mathsf{X}}}(\bm{X};\bm{\theta}) denote the probability density function (PDF) of random variables/vectors/matrices, as being parameterized by \bm{\theta} . \mathcal{CN}(\bm{x};\widehat{\bm{x}},\boldsymbol{\Sigma}) refers to the multivariate Gaussian distribution of a complex-valued vector \bm{x}\in\mathbb{C}^{N} and \delta(\boldsymbol{x}) refers to the multi-dimensional Dirac delta distribution. In addition, the short-hand notation \sim stands for “distributed according to”. The symbol \otimes denotes the Kronecker product operator. We also use \int as a shorthand for \int_{-\infty}^{\infty} . II System Model and Problem Formulation Consider a point-to-point down-link communication system comprising a base station (BS) and a mobile user equipment (UE), equipped with N_{\textsf{Tx}} and N_{\textsf{Rx}} antenna elements, respectively, each forming a one-dimensional uniform linear array (ULA). We consider block fading scenario, where the channel matrix, denoted by \bm{H}_{t}\in\mathbb{C}^{N_{\textsf{Rx}}\times{N}_{\textsf{Tx}}} at each time step t\in\{1,2,\ldots,T\} , remains constant over each t^{\text{th}} coherence block but may change in subsequent blocks. The physical channel \bm{H}_{t} can be represented in angular domain through the virtual channel model [ 17 ] as \bm{H}_{t}=\bm{\mathcal{A}}_{\textsf{Rx}}\,\bm{H}_{\textsf{v},t}\,\bm{\mathcal{A}}_{\textsf{Tx}}^{\textsf{H}}, (1) where \bm{H}_{\textsf{v},t} is the virtual channel matrix and \bm{\mathcal{A}}_{\textsf{Rx}}\in\mathbb{C}^{N_{\textsf{Rx}}\times{N}_{\textsf{Rx}}} and \bm{\mathcal{A}}_{\textsf{Tx}}\in\mathbb{C}^{N_{\textsf{Tx}}\times{N}_{\textsf{Tx}}} are the normalized discrete Fourier transform (DFT) matrices. The virtual channel \bm{H}_{\textsf{v},t} in massive MIMO systems often exhibits a sparse structure due to the limited number of dominant scattering clusters in the propagation environment. To capitalize on this, we model the dynamics of each element in \bm{H}_{\textsf{v},t} using a BDD process, which tracks the appearance, disappearance, and gradual variation of significant channel components over time. Let h_{\textsf{v},t} denote a generic entry in \bm{H}_{\textsf{v},t} . The evolution of h_{\textsf{v},t} in time given \mathsf{h}_{\textsf{v},t-1} involves the following four cases in the BDD model: i) Birth Case : zero entry becomes nonzero, i.e., h_{\textsf{v},t-1}=0 and \mathsf{h}_{\textsf{v},t}\sim\mathcal{CN}(h_{\textsf{v},t};0,\eta_{t}^{-1}) , characterized by the conditional probability \mathrm{Pr}\{\mathsf{h}_{\textsf{v},t}\neq 0|\mathsf{h}_{\textsf{v},t-1}=0\}\triangleq\beta_{\textsf{B},t} . ii) Death Case : nonzero entry becomes zero, i.e., \mathsf{h}_{\textsf{v},t-1}\sim\mathcal{CN}(h_{\textsf{v},t-1};0,\eta_{t-1}^{-1}) and h_{\textsf{v},t}=0 , characterized by the conditional probability \mathrm{Pr}\{\mathsf{h}_{\textsf{v},t}=0|\mathsf{h}_{\textsf{v},t-1}\neq 0\}\triangleq\beta_{\textsf{D},t} . iii) Drift Case : nonzero entry remains nonzero, i.e., \mathsf{h}_{\textsf{v},t-1}\sim\mathcal{CN}(h_{\textsf{v},t-1};0,\eta_{t-1}^{-1}) and h_{\textsf{v},t}=\alpha_{t}\,h_{\textsf{v},t-1}+\sqrt{1-\alpha_{t}^{2}}\,n_{t} with \mathsf{n}_{t}\sim\mathcal{CN}(n_{t};0,\eta_{t}^{-1}) , characterized by the conditional probability \mathrm{Pr}\{\mathsf{h}_{\textsf{v},t}\neq 0|\mathsf{h}_{\textsf{v},t-1}\neq 0\}=1-\beta_{\textsf{D},t} . Here, we assume \eta_{t}\approx\eta_{t-1}

Read the complete text