Deep Learning–Based Coverage and Capacity Optimization

Marinescu, Andrei; Jiang, Zhiyuan; Zhou, Sheng; DaSilva, Luiz A.; Niu, Zhisheng

doi:10.1002/9781119562306.ch4

Cited by 5 publications

(1 citation statement)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [2], the authors successfully utilize such MDT data for automated tuning of the antenna downtilt based on a RL approach. Similarly, the use of RL variants has further been validated in [10]- [12], with multi-agent systems gaining more interest to address the high dimensional action space required for joint optimization over large areas [13]- [16]. Despite these advances, the required training phase of such model-free approaches still poses a severe practical limitation, most notably in terms of sample-efficiency and risk-aversion, which are ongoing research topics [5], [24].…”

Section: B Related Workmentioning

confidence: 99%

A Differentiable Throughput Model for Load-Aware Cellular Network Optimization Through Gradient Descent

Eller,

Svoboda,

Rupp

2024

IEEE Access

View full text Add to dashboard Cite

The efficient operation of cellular networks requires careful tuning of configuration parameters, such as the transmit power or antenna tilts, to adequately balance interference while providing the necessary capacity to the connected UEs. As manual tuning of these parameters is typically unfeasible, several automated coverage and capacity optimization methods have been proposed. However, most existing solutions are either based on poorly scalable black-box optimization methods or solely consider interference management, while omitting the potential of congested cells. In this work, we instead propose a differentiable framework for cellular network optimization, centered around the end-user throughput, that enables load-aware tuning of network parameters through gradient descent. Hereby, we approach the problem from a data-driven perspective, and include dedicated model subcomponents derived from monitoring data, which enable the calibration to site-specific traffic patterns and KPI measurements. We validate our approach for joint transmit power optimization in a real-world network layout with ≈ 150 cells in two frequency bands. In our evaluation, the gradient descent-based optimization reliably reduces the outage ratio for different levels of demand, while the black-box baseline struggles to explore the large search space. Our results further reveal substantial differences between the proposed load-aware and commonly used SINR-based objectives, for which we repeatably obtain unbalanced network configurations with severely congested cells. In contrast, the proposed end-user throughput objective promotes a balanced network configuration, providing adequate resources to the connected UEs while also limiting interference.

show abstract

Section: B Related Workmentioning

confidence: 99%

A Differentiable Throughput Model for Load-Aware Cellular Network Optimization Through Gradient Descent

Eller,

Svoboda,

Rupp

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

Holistic learning and optimization scheme for wireless networks

Husen,

Chaudary,

Ahmad

et al. 2024

Trans Emerging Tel Tech

View full text Add to dashboard Cite

Future networks are expected to use Artificial Intelligence and machine learning applications intensely. The primary focus of machine learning techniques will be on intelligent automation and decision‐making based on information learned from network functions, resources, and users. The research community has studied several schemes for using machine learning in communications and networks in the last two decades. However, the schemes mainly focus on learning the state information of individual network functions and resources and performing their optimization in a standalone fashion. Recently, a study group from the International Telecommunication Union has outlined machine learning pipeline requirements. It addresses several aspects of machine learning operations, such as model selection, data sources, and action points. However, the mechanisms to define the state of network functions and resources for different network layers and domains are yet to be explored. Therefore, in addition to the above, specific studies are required for cross‐layer and administrative domain knowledge sharing and holistic optimization. Without considering the holistic view, applying machine learning to optimizing individual functions and resources may result in nonoptimal and unexpected behaviors instead of having any positive effect. The need for global and deep holistic learning has recently been identified in the literature; however, no implementation of the above scheme has been studied or evaluated for networks. This article proposes a novel approach, implementing global and deep holistic learning in wireless networks toward fulfilling the requirements of future intelligent networks. It also proposes an objective function‐based feature engineering for wireless networks. Experimental results of the proposed scheme show significant improvements in the network performance and accuracy of the machine learning models. Moreover, applications of transfer learning to base learners for network functions and resources can significantly expedite decision‐making with reduced computational costs.

show abstract