Bayat, Ali
(2020).
*Designing dynamic resource allocation mechanisms for wireless powered communication networks.*
Thèse.
Québec, Doctorat en télécommunications, Université du Québec, Institut national de la recherche scientifique, 223 p.

Prévisualisation |
PDF
Télécharger (3MB) | Prévisualisation |

## Résumé

In this dissertation, we focus on designing resource allocation algorithms and scheduling schemes for wireless information and power transfer systems. In contrast to wireless information transfer (WIT) systems, the end-to-end channel from the wireless power transmitter to the energy harvesting output in wireless power (WPT) systems is non-linear. This non-linearity should be taken into account while designing optimal transmit signals and scheduling schemes in WPT systems. As the first milestone in this PhD dissertation, novel closed-form formulas relating the output DC current of a rectifier to the shape of transmit signals in WPT systems is derived. Then, using these formulas, two scheduling schemes, i.e. time sharing and spatial multiplexing, are studied, revealing by the end, the importance of considering the nonlinearity effect in designing such schemes. As the second milestone of the thesis, we focus on designing resource allocation mechanisms for WPT, WIT, and simultaneous wireless and information transfer (SWIPT) systems through applying auction theory. In particular, for WIT and SWIPT systems, we design deep learning-based auction mechanisms to solve the NP-hard resource allocation problems for real-time applications in 5G and beyond. The first part of this four-part thesis, is about finding a tractable formula for energy harvesting devices which can be applied by researchers for designing transmit signals for WPT systems. Recently, multi-tone transmit signals are shown to be more efficient for WPT systems. However, there is a lack of a tractable mathematical representation of the output harvested energy to the transmit signal while considering the nonlinearity of the rectifiers. The proposed formula avoids most of the tractability issues already in the literature. As in multi-user WIT systems, the harvesting devices far from the power transmitter in multiuser WPT systems require much more wireless power shares compared with the near devices—known as the near-far problem—when fairness is the objective of the system. One main fairness metric is the min-max criterion. One interesting challenge is how the nonlinearity effect of the energy receivers in WPT systems could affect choosing the optimal scheduling schemes considering the max-min fairness. In part II, using the Bessel-based formulas derived in part I, we mathematically prove that the time sharing scheduling scheme outperforms the spatial multiplexing scheme under the max-min criterion. The results of this part are novel and highlight the importance of considering the nonlinearity effect while designing scheduling schemes for WPT systems. In the next generations of the wireless communications, the economical behavior of devices is going to be considered so that the users have the option to choose between different service providers while considering their own monetary budgets. Auctions provide a proper competition framework for the devices who behave selfishly in order to maximize their own payoff. Game theory is the appropriate tool for analyzing the auctions and finding the equilibrium points of the game. Consisting part III of the thesis, chapters 6 and 7 present distributed algorithms which converge efficiently to the Nash equilibrium points of the game, played by the energy harvesting devices in a WPT system. The problem of finding the optimal allocation strategy considering bidding behavior of the energy harvesting devices, where the bids are dynamically generated by the users, is tackled in this part. A joint game theory and queuing theory analysis approach is applied which illustrates how these two theories can contribute the solution. While in Part III, the allocation algorithms are analyzed with game theory for a WPT system, in the last part, i.e. Part IV, of this dissertation, we are in pursuit of designing optimal auction mechanisms for the task of resource allocation for simultaneous information and power transfer systems. In this part, we face a complex NP-hard resource allocation problem which has exponential time complexity and cannot be solved applying conventional iterative beamforming algorithms. One big challenge with the emergence of 5G and its service heterogeneity is the complexity of resource allocation in such complex systems. Recently, machine learning has successfully entered the wireless domain. One promising practical solution to carry out such complex resource allocation tasks in real-time is through applying deep learning.

Type de document: | Thèse Thèse |
---|---|

Directeur de mémoire/thèse: | Aïssa, Sonia |

Mots-clés libres: | wireless; réseau sans fil; télécommunications; |

Centre: | Centre Énergie Matériaux Télécommunications |

Date de dépôt: | 21 avr. 2021 15:05 |

Dernière modification: | 21 avr. 2021 15:05 |

URI: | https://espace.inrs.ca/id/eprint/11367 |

## Gestion Actions (Identification requise)

Modifier la notice |