Гибридная модель эффективного представления данных в беспроводных транспортных сетях с динамической топологией VANET

Abstract

At the moment, traffic information systems require the aggregation of big data to provide recommendations to vehicles in the current conditions, which leads to an increase in user comfort. The main tool for improving the level of safety was timely informing traffic participants about the current situation on the road, weather conditions, etc. In this case, if the network object is subjected to an attack and the data is replaced during transmission, then the disclosure of confidential information, the creation of emergency si¬tuations, etc. is possible throughout the visibility zone of the VANET segment. In this regard, the most urgent issue is ensuring security, including when transmitting traffic, and conducting an additional analysis of big data about anomalies and ongoing unauthorized actions. Aim. To develop a hybrid model for the efficient placement of source and intermediate data in wireless transport networks with a dynamic VANET topology, which represents a structural representation of a software-configurable network and edge computing tools, with the ability to optimally analyze data from network nodes and identify anomalies. Methods. The considered Edge computing approach consists in locating computing capacities in geogra¬phically distributed computing devices closer to end users. Software-configurable SDN networks transfer part of the control and physical transmission functions from routing and switches, reducing the load. Within the framework of this study, an RD algorithm has been developed – a protocol for transmitting and processing intermediate data. To carry out clustering of vehicles on a network segment, the DBSCAN unsupervised learning method was used. Preliminary analysis of abnormal traffic was carried out on the basis of RNN neural network models with short-term memory. Results. The developed hybrid model of efficient placement of initial and intermediate data makes it possible to react faster to unauthorized actions. Conclusion. The results obtained in the course of the study confirm the need to implement and scale a hybrid model with boundary calculations in practice.