A statistical framework using parametric estimation and quantile regression for execution times, combined with communication latency models, selects optimal GPU clock frequencies to achieve 95% probability of meeting 500 ms end-to-end deadlines while reducing energy consumption by over 50%.
and Smith, Alan Jay , title =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Statistical Analysis for Energy-Efficient Satellite Edge Computing with Latency Guarantees
A statistical framework using parametric estimation and quantile regression for execution times, combined with communication latency models, selects optimal GPU clock frequencies to achieve 95% probability of meeting 500 ms end-to-end deadlines while reducing energy consumption by over 50%.