Table 3 Computational Resource Consumption and Real-World Feasibility of the Proposed Model on NVIDIA Jetson TX2.
Sl No | Category | Details | Feasibility |
|---|---|---|---|
1 | Hardware | NVIDIA Jetson TX2 (256-core Pascal GPU, 6-core ARM CPU, 8Â GB LPDDR4 RAM) | Compact and powerful edge AI hardware, suitable for medical use |
2 | Power Consumption | 7.5–15 W (under moderate-to-high inference load) | Low enough for mobile clinics or battery-powered diagnostic tools |
3 | Energy Efficiency |  ~ 0.5 W/image (based on 15W max power and 30 images/sec throughput) | Energy-efficient for real-time steganography in portable and embedded setups |
4 | Model Size |  ~ 25–30 MB | Lightweight for TX2’s onboard storage |
5 | Disk Usage |  ~ 3.0 GB (model, dataset, dependencies) | Easily accommodated on TX2’s internal or external storage |
6 | RAM Usage |  ~ 1.8–2.2 GB during inference | Fits comfortably within 8 GB RAM |
7 | Processing Time (Latency) |  ~ 25–35 ms per image (encoder + decoder on Jetson TX2 with CUDA acceleration) | Real-time embedding and decoding achievable |
8 | Processing Throughput |  ~ 25–30 images/sec | Suitable for continuous or batch secure image transmission |
9 | Environmental Constraints | 0–50 °C | Suitable for mobile labs, rural health units, or static clinical environments |
10 | Deployment feasibility | Plug-and-play deployment | High-no external dependencies beyond standard CUDA stack |
