Model-Free Learning for Massive MIMO Systems: Stochastic Approximation Adjoint Iterative Learning Control

Leontine Aarnoudse; Tom Oomen

doi:10.23919/ACC50511.2021.9482967

Model-Free Learning for Massive MIMO Systems: Stochastic Approximation Adjoint Iterative Learning Control

Leontine Aarnoudse, Tom Oomen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

Learning can substantially increase the performance of control systems that perform repeating tasks. The aim of this paper is to develop an efficient iterative learning control algorithm for MIMO systems with a large number of inputs and outputs that does not require model knowledge. The gradient of the control criterion is obtained through dedicated experiments on the system. Using a judiciously selected randomization technique, an unbiased estimate of the gradient is obtained from a single dedicated experiment, resulting in fast convergence of a Robbins-Monro type stochastic gradient descent algorithm. Analysis shows that the approach is superior to earlier deterministic approaches and to related SPSA-type algorithms. The approach is illustrated on a multivariable example.

Original language	English
Title of host publication	2021 American Control Conference (ACC)
Publisher	Institute of Electrical and Electronics Engineers
Pages	2181-2186
Number of pages	6
ISBN (Electronic)	9781665441971
DOIs	https://doi.org/10.23919/ACC50511.2021.9482967
Publication status	Published - 28 Jul 2021
Event	2021 American Control Conference, ACC 2021 - Virtual, Virtual, New Orleans, United States Duration: 25 May 2021 → 28 May 2021 http://acc2021.a2c2.org/

Conference

Conference	2021 American Control Conference, ACC 2021
Abbreviated title	ACC 2021
Country/Territory	United States
City	Virtual, New Orleans
Period	25/05/21 → 28/05/21
Internet address	http://acc2021.a2c2.org/

Bibliographical note

Funding Information:
*This work is part of the research programme VIDI with project number 15698, which is (partly) financed by the NWO.

Access to Document

10.23919/ACC50511.2021.9482967

Cite this

@inproceedings{81b8a6446b5546949fcab1787ca858d4,

title = "Model-Free Learning for Massive MIMO Systems: Stochastic Approximation Adjoint Iterative Learning Control",

abstract = "Learning can substantially increase the performance of control systems that perform repeating tasks. The aim of this paper is to develop an efficient iterative learning control algorithm for MIMO systems with a large number of inputs and outputs that does not require model knowledge. The gradient of the control criterion is obtained through dedicated experiments on the system. Using a judiciously selected randomization technique, an unbiased estimate of the gradient is obtained from a single dedicated experiment, resulting in fast convergence of a Robbins-Monro type stochastic gradient descent algorithm. Analysis shows that the approach is superior to earlier deterministic approaches and to related SPSA-type algorithms. The approach is illustrated on a multivariable example.",

author = "Leontine Aarnoudse and Tom Oomen",

note = "Funding Information: *This work is part of the research programme VIDI with project number 15698, which is (partly) financed by the NWO. ; 2021 American Control Conference, ACC 2021, ACC 2021 ; Conference date: 25-05-2021 Through 28-05-2021",

year = "2021",

month = jul,

day = "28",

doi = "10.23919/ACC50511.2021.9482967",

language = "English",

pages = "2181--2186",

booktitle = "2021 American Control Conference (ACC)",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

url = "http://acc2021.a2c2.org/",

}

Aarnoudse, L & Oomen, T 2021, Model-Free Learning for Massive MIMO Systems: Stochastic Approximation Adjoint Iterative Learning Control. in 2021 American Control Conference (ACC)., 9482967, Institute of Electrical and Electronics Engineers, pp. 2181-2186, 2021 American Control Conference, ACC 2021, Virtual, New Orleans, United States, 25/05/21. https://doi.org/10.23919/ACC50511.2021.9482967

Model-Free Learning for Massive MIMO Systems: Stochastic Approximation Adjoint Iterative Learning Control. / Aarnoudse, Leontine; Oomen, Tom.
2021 American Control Conference (ACC). Institute of Electrical and Electronics Engineers, 2021. p. 2181-2186 9482967.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Model-Free Learning for Massive MIMO Systems

T2 - 2021 American Control Conference, ACC 2021

AU - Aarnoudse, Leontine

AU - Oomen, Tom

N1 - Funding Information: *This work is part of the research programme VIDI with project number 15698, which is (partly) financed by the NWO.

PY - 2021/7/28

Y1 - 2021/7/28

N2 - Learning can substantially increase the performance of control systems that perform repeating tasks. The aim of this paper is to develop an efficient iterative learning control algorithm for MIMO systems with a large number of inputs and outputs that does not require model knowledge. The gradient of the control criterion is obtained through dedicated experiments on the system. Using a judiciously selected randomization technique, an unbiased estimate of the gradient is obtained from a single dedicated experiment, resulting in fast convergence of a Robbins-Monro type stochastic gradient descent algorithm. Analysis shows that the approach is superior to earlier deterministic approaches and to related SPSA-type algorithms. The approach is illustrated on a multivariable example.

AB - Learning can substantially increase the performance of control systems that perform repeating tasks. The aim of this paper is to develop an efficient iterative learning control algorithm for MIMO systems with a large number of inputs and outputs that does not require model knowledge. The gradient of the control criterion is obtained through dedicated experiments on the system. Using a judiciously selected randomization technique, an unbiased estimate of the gradient is obtained from a single dedicated experiment, resulting in fast convergence of a Robbins-Monro type stochastic gradient descent algorithm. Analysis shows that the approach is superior to earlier deterministic approaches and to related SPSA-type algorithms. The approach is illustrated on a multivariable example.

UR - http://www.scopus.com/inward/record.url?scp=85111933189&partnerID=8YFLogxK

U2 - 10.23919/ACC50511.2021.9482967

DO - 10.23919/ACC50511.2021.9482967

M3 - Conference contribution

AN - SCOPUS:85111933189

SP - 2181

EP - 2186

BT - 2021 American Control Conference (ACC)

PB - Institute of Electrical and Electronics Engineers

Y2 - 25 May 2021 through 28 May 2021

ER -

Model-Free Learning for Massive MIMO Systems: Stochastic Approximation Adjoint Iterative Learning Control

Abstract

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this