Resource Allocation for Multi-target Radar Tracking via Constrained Deep Reinforcement Learning

Ziyang Lu, M. Cenk Gursoy

Research output: Chapter in Book/Entry/PoemConference contribution

Abstract

In this work, we propose a constrained deep reinforcement learning (CDRL) based approach to address resource allocation for multi-target tracking in a radar system. In the proposed CDRL algorithm, both the parameters of the deep Q-network (DQN) and the dual variable are learned simultaneously. The proposed CDRL framework consists of two components, namely online CDRL and offline CDRL. Training a DQN in the deep reinforcement learning algorithm usually requires a large amount of data, which may not be available in a target tracking task due to the scarcity of measurements. We address this challenge by proposing an offline CDRL framework, in which the algorithm evolves in a virtual environment generated based on the current observations and prior knowledge of the environment. Simulation results show that both offline CDRL and online CDRL are critical. Offline CDRL provides more training data to stabilize the learning process and the online component can sense the change in the environment and make the corresponding adaptation.

Original languageEnglish (US)
Title of host publication2023 IEEE 34th Annual International Symposium on Personal, Indoor and Mobile Radio Communications
Subtitle of host publication6G The Next Horizon - From Connected People and Things to Connected Intelligence, PIMRC 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665464833
DOIs
StatePublished - 2023
Externally publishedYes
Event34th IEEE Annual International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC 2023 - Toronto, Canada
Duration: Sep 5 2023Sep 8 2023

Publication series

NameIEEE International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC

Conference

Conference34th IEEE Annual International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC 2023
Country/TerritoryCanada
CityToronto
Period9/5/239/8/23

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Resource Allocation for Multi-target Radar Tracking via Constrained Deep Reinforcement Learning'. Together they form a unique fingerprint.

Cite this