Photon Mapping and ML for Real-Time Radio Propagation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for computing radio propagation in cellular network planning, such as ray tracing, are inefficient and time-consuming, especially in urban areas with many obstacles, leading to low-quality results and high computational costs.
Innovation Solution
Combining photon mapping with machine learning to simulate radio propagation, allowing for near-real-time computation of cellular coverage by transforming low-resolution photon mappings into high-resolution representations using GPUs, thereby reducing computation time and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ray tracing is used to compute radio propagation, then measurement precision is improved, but computation time increases significantly
Solution Approach 1:
The patent creates a virtual 3D model copy of the physical environment including buildings, terrain, and obstacles. Photon mapping algorithms operate on this digital copy rather than performing physical measurements or full-scale ray tracing, enabling fast computation while maintaining accuracy through realistic electromagnetic wave propagation simulation in the virtual model.
Solution Approach 2:
The patent replaces traditional ray tracing mechanical computation with photon mapping algorithms that use different computational mechanics. Instead of tracing individual rays through complex geometric intersections, the system uses photon packet tracking with spatial hashing and k-d trees, substituting the computational mechanism to achieve both speed and accuracy.
2Manufacturing precision
If more photons are traced to improve coverage map quality, then manufacturing precision is improved, but computation time increases
Solution Approach 1:
The patent performs preliminary photon mapping to generate a photon map data structure that stores photon positions, directions, and energies in an optimized spatial format. This preliminary action creates a reusable database that can be queried multiple times without re-computing photon trajectories, enabling fast generation of high-resolution coverage maps through repeated sampling of the pre-computed photon map.
Solution Approach 2:
The patent changes the computational parameters by using photon packet methods with varying photon counts and energy distributions. By adjusting parameters such as photon packet size, spatial hashing resolution, and k-d tree depth, the system can dynamically balance between coverage map resolution and computation speed, achieving high precision when needed while maintaining productivity for routine operations.
Data Source
AI summary
Aspects of the subject disclosure may include, for example, network deployment or radio-propagation computation based on a combination of photon mapping and machine learning including supporting near-real-time computation of the radio transmissions for different layouts of antennas and allowing examination of a large variety of antenna locations and layouts, changing configuration details, e.g., tilting antennas or optimally selecting the sector that each antenna covers, and so on. Other embodiments are disclosed.