| Abstract—In order to reduced training time and
achieve internet of things (IoT) real-time response.
This paper presented an architecture and a dynamic
distribution, reducing the training time and transmission
time by distributing training data in fog computing
platform and cloud computing platform. This paper
presented an architecture which process the advantages
of both edge processing and cloud computing for AI
application on sensor data and image recognition. This
paper combined the AI and IoT together. First,the paper
designed the AIoT architecture. End device transferred
sensor data and image to fog computing platform
by LoRa, XBee, Wi-Fi. Configure end-devices to
monitor environmental data. The proposed advanced
dynamic distribution approach is an integrated strategy
of edge/fog computing and cloud computing. Fog computing
platform received the partial training data and
testing data, transmitted from cloud drive. After the advanced
dynamic distributing, fog computing platform
starts training model by Node-RED. Through the advanced
dynamic distribution, cloud passed the partial
data to fog computing platform then start tensorflow
training model simultaneously. The main advantage of
the proposed strategy is aims to reduced training time.
When fog computing platform and cloud computing
platform completed model training, cloud transfer
trained model to fog computing platform. Node-RED
received trained model from cloud computing platform.
Node-RED combined trained model from cloud and
local. Node-RED transferred the combined model to
NVIDIA JETSON TX2 and provide TX2 to execute
data inference.
|
