9004 Problems of information security. Computer systems Проблемы информационной безопасности. Компьютерные системы 2071-8217 11 An approach to detecting botnet attacks in the Internet of Things networks Подход к обнаружению ботнет-атак в сетях Интернета вещей 0000-0002-6419-0072 Tatarnikova Tatiana Tm-tatarn@yandex.ru 0000-0001-6289-3295 57200960264 Sikarev Igor sikarev@yandex.ru Bogdanov Pavel Timochkina Tatyana St. Petersburg State University of Aerospace Instrumentation Russian State Hydrometeorological University Saint-Petersburg State University of Aerospace Instrumentation 12 11 2021 3 108 117

An approach to detecting network attacks based on deep learning methods - autoencoders is proposed. It is shown that training examples can be obtained when connecting IoT devices to the network, as long as the traffic does not carry malicious code. Statistical values and functions extracted from traffic are proposed, on which patterns of behavior of IoT devices are built.

 Internet of Things Network Attack Attack Detection System Autoencoder Principal Component Method Unsupervised Learning