SWEET: Simulation Results for Edge Inference Serving System

Simulation results from 2026 demonstrate the SWEET inference serving system. The data, authored by Xiangchen Li and shared under CC-BY-4.0, likely contains metrics on time, power consumption, communication payload, and accuracy degradation from experiments co-optimizing model quantization and workload partitioning for edge devices.

Use Cases

• Benchmarking edge inference latency based on the described simulation of end-to-end time reduction.
• Evaluating energy efficiency for edge AI based on the reported power consumption metrics.
• Analyzing communication overhead in split computing based on the described greater than 80% reduction in payload.
• Studying accuracy-performance trade-offs in model quantization based on the described accuracy degradation metric.

Strengths

• Includes specific performance results, such as a greater than 80% reduction in communication payload.
• Explicitly accounts for accuracy degradation, keeping it below 1% in simulations.
• Shared under a permissive CC-BY-4.0 license for reuse.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

figshare

Collection Method

Likely generated from simulations of the SWEET system.

Freshness

Last updated 2026-06-03 22:01:15; freshness should be verified.