Distributed parameter estimation with Gaussian observation noises in time-varying digraphs
adaptive controlcooperative persistent excitationdiffusion strategiesdistributed parameter estimationnetworksensorsstochastic linear regression modelsystem identification
Asymptotic properties of parametric estimators (62F12) Parametric inference under constraints (62F30) Estimation in multivariate analysis (62H12) Networked control (93B70) Adaptive control/observation systems (93C40) Discrete event control/observation systems (93C65) Estimation and detection in stochastic control theory (93E10) Identification in stochastic control theory (93E12)
In this paper, the authors deal with the problem of distributed parameter estimation in a network of \(n\) sensors. At each time \(k\) every sensor \(i\in \{1, \ldots, n\}\) outputs a noisy measurement \(y_{i}(k)\in\mathbb{R}\) determined by the stochastic linear regression model\N\[\Ny_{i}(k)=\theta^\prime \phi_i(k)+w_{i}(k), \, k \geq 0,\N\]\Nwhere \(\theta\in \mathbb{R}^{d}\) is the parameter to be estimated, \(w_{i}(k)\in\mathbb{R}\) is the independent and identically distributed (i.i.d.) Gaussian random noise with zero mean and covariance \(R_i\geq0\), \(\phi_i(k)\) denotes a vector that is either a linear or nonlinear function of \((u_i(0),u_i(1),\ldots,u_i(k-1))\) and \((y_i(0),y_i(1),\ldots,y_i(k-1))\), where \(u_i(t)\) and \(y_i(t)\) are respectively the input and output signals of the \(i\)-th subsystem at time \(t\). The formulated problem of estimating \(\theta\) is fundamental in many applications such as system identification and adaptive control (see, for example, [\textit{K. J. Åström} and \textit{B. Wittenmark}, Adaptive control. Revised and corrected reprint of the 1995 2nd ed. Mineola, NY: Dover Publications (2008; Zbl 1163.93020)]. The authors of this paper assumed that the sensors communicate over a time-varying weighted digraph \(\mathcal{G}(k)=(\mathcal{V},\mathcal{E}(k),A(k))\), where \(\mathcal{V} = \{1, \ldots, n\}\) represents the set of sensors, \(\mathcal{E}(k) \subseteq \mathcal{V} \times \mathcal{V}\) defines the set of edges, \(A(k) = \left\{a_{ij}(k)\right\}\) is the weighted adjacency matrix of \(\mathcal{G}(k)\). The entries in \(A(k)\) are non-negative, and \(a_{ij}(k) > 0\) if and only if sensor \(i\) directly receives information from sensor \(j\), i.e., when \((j, i) \in \mathcal{E}(k).\) The authors approach to the estimation problem is based on the stochastic dynamic regressor extension and mixing (DREM) algorithm introduced in [\textit{S. Aranovskiy} et al., IEEE Trans. Autom. Control 62, No. 7, 3546--3550 (2017; Zbl 1370.93250)]. This algorithm transforms the problem of estimating a \(d\)-dimensional vector parameter to that of \(d\) scalar ones. By combining it with the classical least-mean square (LMS) scheme, the authors propose the distributed estimation algorithm. It is shown that, under weak conditions on network topology and excitation of regressors, the proposed estimators guarantee that each sensor infers the true parameter, even if any individual of them cannot by itself. The efficiency of the established results is demonstrated with the help of numerical examples. Comparison with existing distributed parameter estimation algorithms is proposed. For more details see [\textit{R. Abdolee} and \textit{B. Champagne}, ``Diffusion LMS strategies in sensor networks with noisy input data, IEEE/ACM Trans. Networking 24, No. 1, 3--14 (2014); \textit{R. Abdolee} et al., ``Diffusion adaptation over multi-agent networks with wireless link impairments, Preprint, \url{arXiv:1507.05228}]; \textit{S. Xie} and \textit{L. Guo}, IEEE Trans. Autom. Control 63, No. 11, 3643--3658 (2018; Zbl 1423.93390); \textit{A. S. Matveev} et al., IEEE Trans. Autom. Control 67, No. 8, 4256--4263 (2022; Zbl 1537.93678); \textit{B. Yi} and \textit{R. Ortega}, IEEE Trans. Autom. Control 68, No. 2, 1253--1258 (2023; Zbl 1541.93348); \textit{J. Yan} and \textit{H. Ishii}, ``Distributed parameter estimation with Gaussian observation noises in time-varying digraphs, Preprint, \url{arXiv:2311.03911}].
- A Distributed Implementation of Steady-State Kalman Filter
- Analysis of Distributed Adaptive Filters Based on Diffusion Strategies Over Sensor Networks
- Conditions for Convergence of Dynamic Regressor Extension and Mixing Parameter Estimators Using LTI Filters
- Constrained Consensus and Optimization in Multi-Agent Networks
- Convergence of a Distributed Least Squares
- Diffusion Strategies Outperform Consensus Strategies for Distributed Estimation Over Adaptive Networks
- Diffusion-Based Distributed Parameter Estimation Through Directed Graphs With Switching Topology: Application of Dynamic Regressor Extension and Mixing
- Distributed Cooperative Adaptive Identification and Control for a Group of Continuous-Time Systems With a Cooperative PE Condition via Consensus
- Distributed LMS for Consensus-Based In-Network Adaptive Processing
- Distributed Parameter Estimation in Sensor Networks: Nonlinear Observation Models and Imperfect Communication
- scientific article; zbMATH DE number 4015993 (Why is no real title available?)
- Performance Enhancement of Parameter Estimators via Dynamic Regressor Extension and Mixing
- Resilient output regulation in heterogeneous networked systems under Byzantine agents
- Robust Distributed Diffusion Recursive Least Squares Algorithms With Side Information for Adaptive Networks
This page was built for publication: Distributed parameter estimation with Gaussian observation noises in time-varying digraphs
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6993282)
