Baymac’s performance gain stems from its ability to balance local computation and communication in real time. However, the Bayesian model adds computational overhead (≈5% CPU on an ARM Cortex-A53), which may be nontrivial for ultra-low-power devices. Future work includes hardware acceleration for the inference engine and exploring multi-hop Baymac coordination.
Disclaimer: This content is for informational purposes only. Specific coverage limits, eligibility, and services vary based on the current policies of the organization and the membership tier selected. baymac
| Scheduler | Latency (ms) | Energy (mJ/pkt) | Accuracy (%) | |--------------------|--------------|----------------|--------------| | Round-robin | 142 ± 21 | 3.4 ± 0.3 | 78.2 | | Threshold-based | 118 ± 18 | 2.9 ± 0.4 | 82.1 | | | 97 ± 12 | 2.1 ± 0.2 | 88.6 | Baymac’s performance gain stems from its ability to