Green Wireless Transmissions for Advanced Internet of Things: Session-Specific Analysis and Design with Deep Reinforcement Learning
| dc.contributor.author | Xu, Fang | |
| dc.contributor.supervisor | Yang, Hong-Chuan | |
| dc.date.accessioned | 2022-07-08T23:22:41Z | |
| dc.date.available | 2022-07-08T23:22:41Z | |
| dc.date.copyright | 2022 | en_US |
| dc.date.issued | 2022-07-08 | |
| dc.degree.department | Department of Electrical and Computer Engineering | |
| dc.degree.level | Doctor of Philosophy Ph.D. | en_US |
| dc.description.abstract | Reliable and energy-efficient wireless data transmission is of great importance to the success of future Internet of Things (IoT). In this thesis, we analyze and minimize the energy consumption of wireless data transmissions from an individual transmission perspective. In particular, we investigate three fundamental transmission scenarios, namely point-to-point transmission, data collection from wirelessly-powered sensor, and cooperative relaying transmission with wireless power transfer. For point-to-point transmission, we analyze and optimize ideal continuous rate adaptation and continuous power adaptation transmission schemes, where the closed-form expressions for all optimal parameters are derived. These results establish an energy consumption lower limit for wireless data transmissions over fading channels. In the case of data collection from wirelessly-powered sensor scenario, we derive closed-form expressions for optimal transmission parameters for ideal rate adaptive transmission with linear energy harvesting setting. Under more practical assumptions of finite block-length transmission with nonlinear energy harvesting, we propose a deep reinforcement learning (DRL) based approach to arrive at a deep policy network for determining the near-optimal transmission parameters in real time. An online tuning method is also proposed to adjust the policy network using online experience to cater for model inaccuracy and environment variation. Finally, for the case of cooperative relaying transmission with wireless power transfer, under ideal rate adaptive transmission with piecewise linear energy harvesting, we derive the closed-form expressions for all optimal transmission parameters. Then, the optimal design problem is again generalized to the finite block-length transmission with nonlinear energy harvesting setting, where we again apply the DRL-based method to train and tune a deep policy network for determining the near-optimal transmission parameters in real time. For all cases, we illustrate various design tradeoffs through selected numerical examples. Besides improving the energy efficiency of wireless transmissions for future IoT applications, our proposed DRL-based method can serve as a general solution for real-time optimal design problems in wireless communications. | en_US |
| dc.description.scholarlevel | Graduate | en_US |
| dc.identifier.uri | http://hdl.handle.net/1828/14054 | |
| dc.language | English | eng |
| dc.language.iso | en | en_US |
| dc.rights | Available to the World Wide Web | en_US |
| dc.title | Green Wireless Transmissions for Advanced Internet of Things: Session-Specific Analysis and Design with Deep Reinforcement Learning | en_US |
| dc.type | Thesis | en_US |