Target distributed fusion identification method based on edge computing

Abstract

The invention discloses a target distributed fusion identification method based on edge computing, which comprises the following steps: step 1, designing a neural network for each edge device and an edge domain fusion center, wherein the network located on the edge device must compress a sensing signal, and the network located in the edge domain fusion center can receive and fuse input from a plurality of edge device neural networks and further output a final identification result, and the neural networks are connected according to communication network topology of the edge devices and the edge domain fusion center to form a distributed execution overall depth neural network model; the invention provides a target fusion identification method based on edge calculation and a deep neural network to solve the problems of high transmission communication cost and delay in a target identification process, and the method can quickly complete fusion identification of a target on an edge side. The transmission communication traffic between the edge equipment and the edge domain fusion center can be reduced through the neural network.

Description

technical field [0001] The invention belongs to the field of edge computing, the sub-field of deep neural network in the field of artificial intelligence, and the field of distributed systems, and specifically relates to a target distributed fusion recognition method based on edge computing. Background technique [0002] Edge computing refers to a new type of computing model that can perform computing at the edge of the network. The edge refers to any computing and network model between the path from the data source to the cloud computing center, while the downlink data at the edge represents cloud services, and the uplink data represents Internet of Everything Service. In addition, the actual deployment of edge computing is naturally distributed, supports distributed computing and storage, has dynamic scheduling and unified management of distributed resources, supports distributed intelligence, and has capabilities such as distributed security. At present, edge computing c...

AI Technical Summary

Problems solved by technology

[0003] At present, there have been some studies on the application of neural networks for target recognition. For example, Yang Weichao used deep learning methods to modulate and recognize signals in the literature "Research on Communication Signal Modulation Recognition under Alpha Stable Distributed Noise", and achieved good recognition results; such as Li Jiachen In the literature "Research on the recognition method of digital modulation signal based on deep learning", the research on the recognition method of digital modulation signal based on deep learning is proposed, and two different deep learning models are applied respectively, and the parameters required in the algorithm are selected through simulation to achieve the best results. However, none of the above researches considered the problem of applying neural network on the edge side for target recognition. Therefore, we propose a target distributed fusion recognition method based on edge computing.

Images