3D Spectrum Situation Completion Using GANs for UAV Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D spectrum situation completion methods face challenges such as reliance on prior information, inability to achieve 3D completion, and low precision due to insufficient data and complex UAV trajectories, which are difficult to implement in practice.
Innovation Solution
A 3D spectrum situation completion method using a generative adversarial network (GAN) that processes incomplete historical or empirical data to generate complete 3D spectrum situations without requiring prior information, utilizing graying and coloring preprocessing, a trained generator network, and a discriminator network to learn the mapping relationship between incomplete and complete spectrum situations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If traditional two-dimensional spectrum situation completion algorithms are extended to three dimensions, then the completion can be performed in 3D space, but the precision is low because the algorithms assume spectrum data is only distance-dependent which is not consistent with the physical model
Solution Approach 1:
The patent transforms the single-channel spectrum data into three-channel spectrum situation data by introducing spatial direction parameters (azimuth and elevation angles). This parameter expansion allows the model to capture the directional characteristics of electromagnetic wave propagation, making the completion process consistent with the physical model and significantly improving precision in 3D space.
Solution Approach 2:
The patent extends the completion from two-dimensional horizontal plane to three-dimensional space by adding the vertical dimension and directional information. The graying and coloring preprocessing converts scalar spectrum values into vector field representations with spatial orientation, enabling accurate 3D completion that respects electromagnetic wave propagation physics.
2Measurement precision
If complete historical or empirical spectrum data is used for training, then the completion precision can be improved, but it is difficult to obtain complete 3D spectrum situations in practice
Solution Approach 1:
The patent inverts the traditional approach by training the model to map from incomplete to complete spectrum situations rather than requiring complete data for training. The generator network learns to reconstruct the full three-channel spectrum situation from partial observations, and the discriminator validates the realism of the completed data, enabling effective training without access to complete ground truth.
Solution Approach 2:
The patent introduces a generative adversarial network as an intermediary mapping system between incomplete and complete spectrum situations. The GAN framework acts as a mediator that learns the underlying distribution and relationships, allowing the model to generalize from incomplete training data and produce accurate completions even when complete reference data is unavailable.
3Ease of operation
If discrete and incomplete spectrum data is used through sampling, then the data collection becomes feasible, but the data volume is insufficient for effective completion
Solution Approach 1:
The patent enables the system to self-enhance the insufficient training data through the generative adversarial process. The generator creates synthetic complete spectrum situations from incomplete samples, and the discriminator evaluates their realism. This self-service mechanism allows the model to learn from limited sampled data while generating additional training examples, effectively overcoming the data volume limitation.
Solution Approach 2:
The patent performs graying and coloring preprocessing as a preliminary action to transform the incomplete sampled data into a structured three-channel format before feeding it to the GAN. This preprocessing step organizes the sparse data into a form that preserves spatial and directional information, making the subsequent generative process more effective despite the limited data volume.
4Reliability
If graying and coloring preprocessing is performed to create three-channel incomplete 3D spectrum situation maps, then the data structure is improved for GAN training, but the processing complexity increases
Solution Approach 1:
The patent segments the spectrum situation data into three distinct channels through graying and coloring preprocessing, where each channel represents different spatial or spectral characteristics. This segmentation transforms the complex incomplete 3D data into structured multi-channel input that the GAN can process effectively, improving training reliability while maintaining manageable complexity through systematic organization.
Data Source
AI summary
A three-dimensional (3D) spectrum situation completion method and device based on a generative adversarial network includes performing graying and coloring preprocessing based on incomplete 3D spectrum situations from historical or empirical spectrum data obtained by a UAV through sampling a target region, obtaining three-channel incomplete 3D spectrum situation maps displayed in colors, forming a training set based on the incomplete 3D spectrum situation maps; training the generative adversarial network based on the training set and obtaining a trained generator network in the generative adversarial network, performing graying and coloring preprocessing based on a measured incomplete 3D spectrum situation obtained by the UAV through sampling a specified measurement region, obtaining a three-channel measured incomplete 3D spectrum situation map displayed in colors, and using the measured incomplete 3D spectrum situation map as input data to the generator network to obtain a three-channel measured complete 3D spectrum situation map displayed in colors.