US 11,916,755 B2
Method and device for execution of deep neural network in IoT edge network
Jyotirmoy Karjee, Bangalore (IN); Kartik Anand, Bangalore (IN); Vanamala Narasimha Bhargav, Bangalore (IN); Praveen Naik S, Bangalore (IN); Ramesh Babu Venkat Dabbiru, Bangalore (IN); Srinidhi N, Bangalore (IN); Anshuman Nigam, Bangalore (IN); and Rishabh Raj Jha, Bangalore (IN)
Assigned to Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed by Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed on Mar. 24, 2022, as Appl. No. 17/703,296.
Application 17/703,296 is a continuation of application No. PCT/KR2022/004085, filed on Mar. 23, 2022.
Claims priority of application No. 202141012716 (IN), filed on Mar. 24, 2021; and application No. 2021 41012716 (IN), filed on Feb. 28, 2022.
Prior Publication US 2022/0311678 A1, Sep. 29, 2022
Int. Cl. G06F 15/16 (2006.01); H04L 41/16 (2022.01); H04L 43/0888 (2022.01); H04L 67/10 (2022.01)
CPC H04L 41/16 (2013.01) [H04L 43/0888 (2013.01); H04L 67/10 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method for execution of a deep neural network (DNN) in an internet of things (IoT)-edge network, the method comprising:
selecting, by an IoT device, at least one edge device from a plurality of edge devices within communication range of the IoT device to offload computations related to at least one DNN layer of a plurality of DNN layers, wherein the plurality of edge devices are devices having a higher computing capability than the IoT device;
identifying, by the IoT device, a network for connecting the IoT device with the selected at least one edge device;
determining, by the IoT device, a split ratio for splitting a computation task based on at least one of an inference time of the DNN and a transmission time required for transmitting output of each layer of the DNN from the IoT device to the selected at least one edge device;
splitting, by the IoT device, the plurality of DNN layers into a first part and a second part based on the determined split ratio; and
transmitting the second part to the selected at least one edge device through the network,
wherein computation task of the first part is executed on the IoT device and computation task of the second part is executed on the selected at least one edge device.