Deep Learning Cellular Network Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optimizing cellular networks is challenging due to their increasing size and complexity, leading to slow and costly optimization processes, inaccurate solutions, and the need for iterative adjustments that can disrupt network performance and lack transferability between networks.
Innovation Solution
The use of deep learning techniques, specifically constructing multi-dimensional state tensors to input data into deep neural networks, which identifies underperforming cells and their neighbors, allowing for automated optimization without requiring on-site verification or retraining for different locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual modeling is used to optimize cellular networks, then optimization can be performed without disrupting the real network, but the optimization process becomes slow and requires days or weeks to complete
Solution Approach 1:
The patent pre-trains deep neural networks using virtual models and synthetic data before they are needed for actual optimization. This preliminary training phase allows the system to learn optimal parameter adjustments in advance, so when real optimization is needed, the pre-trained models can provide recommendations quickly without requiring lengthy simulation runs or iterative real-world testing.
2Measurement precision
If iterative optimization is performed on the real network, then accurate feedback can be gathered, but the network may lose coverage and performance deteriorates during the process
Solution Approach 1:
The patent creates virtual copies of the cellular network that replicate real-world conditions using synthetic data and simulated environments. These virtual models serve as safe testing grounds where optimization algorithms can be trained and validated without risking actual network coverage. The virtual models are designed to accurately reflect interference patterns, propagation conditions, and network topology, providing realistic feedback without compromising service.
3Manufacturing precision
If conventional optimization methods are used, then solutions can be found for specific networks, but the solutions cannot be transferred to different cellular networks
Solution Approach 1:
The patent develops deep neural networks with universal architectures that can be trained on data from different network configurations and then applied to various cellular networks. The models learn fundamental optimization patterns and interference management strategies that are transferable across different topologies, frequencies, and deployment scenarios, eliminating the need to retrain from scratch for each new network.
4Area of stationary object
If the number of cells is increased to provide ubiquitous coverage, then coverage area improves, but the complexity of optimization increases exponentially
Solution Approach 1:
The patent divides the large-scale cellular network into smaller, manageable units such as clusters of neighboring cells or individual cells with their immediate neighbors. The deep neural networks are trained to optimize these segmented units independently, learning local optimization strategies that can be applied throughout the larger network. This segmentation reduces the computational complexity from exponential to polynomial scaling with network size.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present technology provides a new approach to optimizing wireless networks, including the coverage and capacity of cellular networks, using deep learning. The proposed method involves generating a group of cells comprising a cell identified as underperforming and one or more neighboring cells, ranking the one or more neighboring cells based on one or more relationship parameters between the underperforming cell and the one or more neighboring cells, and generating a multi-dimensional multi-channel state tensor for the group of cells based on the ranking of the one or more neighboring cells. This approach to cellular network optimization improves the coverage and capacity of cellular networks using a process that is faster, more accurate, less costly, and more robust.