A Self-Calibrating Agentic AI Framework for Autonomous Edge Resource Allocation
The paper reports a self-calibrating LLM-agent framework that beat baseline agents on edge workload resource prediction by 91.7%.
The framework uses an ARIMA-based calibration mechanism to limit drift where reliable ground truth is not continuously available. The authors test it on profiling resource use for zero-knowledge workloads in edge computing networks. They also report a 71.7% speed improvement over pure profiling, and say their ARIMA “leaping” ground-truth generation is 52% faster than standard ARIMA at the same accuracy. The work has been submitted to IEEE Transactions on Network and Service Management. ArXiv · AI/CL/LG's note
The framework uses an ARIMA-based calibration mechanism to limit drift where reliable ground truth is not continuously available. The authors test it on profiling resource use for zero-knowledge workloads in edge computing networks. They also report a 71.7% speed improvement over pure profiling, and say their ARIMA “leaping” ground-truth generation is 52% faster than standard ARIMA at the same accuracy. The work has been submitted to IEEE Transactions on Network and Service Management. ArXiv · AI/CL/LG's note
score 4