Conflict-Resilient Incremental Offloading of Deep Neural Networks to the Edge of Smart Environment

Table 2

Time spent in each step with different models (seconds).

DNN models	Execution strategy	Planning	Offloading	Classifying	Total
Alex-Net	All-at-once	—	5.0584	0.1440	5.2024
	Enhanced IONN	5.191E − 05	3.0320	0.1839	3.2160
	PerDNN	4.107E − 05	2.8233	0.1961	3.0194
	CIODE	1.302E − 05	2.2688	0.1882	2.4570
Res-Net	All-at-once	—	2.0166	0.0359	2.0525
	Enhanced IONN	7.995E − 05	1.3463	0.0905	1.4369
	PerDNN	7.024E − 05	1.4251	0.0976	1.5228
	CIODE	2.211E − 05	0.6755	0.1041	0.7796
VGG-16	All-at-once	—	1.5674	0.1294	1.6968
	Enhanced IONN	6.396E − 05	1.1425	0.1499	1.2925
	PerDNN	6.508E − 05	1.0207	0.1502	1.1710
	CIODE	1.591E − 05	0.7865	0.1417	0.9282