Model Complexity and Inference Time for a Wideband Signal Processing Framework

A 5.5 KB dataset by Lulu Liu, last updated on 2026-04-27, containing experimental results for a proposed wideband signal processing framework. The framework uses a RepViT backbone and semantic segmentation of signal spectrograms for detection, recognition, and parameter extraction. Results include a maximum signal recognition rate of 82.43% and normalized root mean squared error metrics for time and frequency parameters.

Use Cases

• Benchmarking deep learning models for automatic modulation recognition based on reported recognition rates.
• Evaluating the performance of semantic segmentation techniques for signal detection in spectrograms.
• Analyzing the trade-off between model complexity and inference time for real-time signal processing applications.
• Comparing parameter extraction accuracy for time and frequency features of wireless signals.

Strengths

• Provides concrete performance metrics, including a maximum recognition rate of 82.43% and average rate of 65.16%.
• Reports specific error ranges for parameter extraction, such as NRMSE for time parameters between 0.3% and -2.8%.
• Released under a permissive CC-BY-4.0 license for broad reuse.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small at 5.5 KB, indicating limited scope, likely containing summary results rather than raw data.

Provenance

Source

Lulu Liu via figshare

Collection Method

Experimental results on a large-scale synthetic dataset, as described.

Time Range

The dataset itself has no specified temporal coverage; it contains model performance results.

Freshness

Last updated 2026-04-27 17:39:29; freshness should be verified.

Geography

null