Use of sockets technology in the network system of a virtual laboratory based on machine learning algorithms

Abstract

The use of virtual laboratory systems has become a modern trend in many areas of human activity. Modern networked virtual laboratories use virtual reality technology to implement laboratory simulations and interactive environment technology to control simulations. The work examines the advantages, principles of building a virtual laboratory and existing problems in the communication mechanism. In particular, when exchanging data between node machines in a network virtual laboratory system, the network layer cannot ensure the same transmission of packet data. The article reviews modern methods of machine learning: supervised, unsupervised, semi-supervised and reinforcement learning. It was determined that the main characteristics of the virtual laboratory system architecture are openness, modularity and real-time communication. The communication mechanism of the virtual laboratory system does not meet the requirements of high bandwidth and real-time two-way communication. To solve the problem of technical requirements of high throughput and ultra-low delay of the virtual laboratory system, a communication planning technology based on a deep machine learning algorithm is proposed. The work uses the Socket mechanism to implement communication based on the TCP/IP protocol. In accordance with the technical requirements of high throughput and ultra-low latency of the network system of the virtual experiment, a technology of self-tuning and optimal planning of the communication network based on the deep machine learning algorithm is proposed. The proposed technology makes it possible to make the optimal decision regarding resource planning and implement optimization of the indirect selection of system parameters.