Аннотация:
The paper is dedicated to the development and study of models and methods of authorization
processes and high load modes for IoT components in the operation queue environment.
An essential part of authorization is the authentication of the transmitted data. In the trusted
nodes, there's a risk of an unauthorized modification of payloads. It poses a threat because
they may be infected with various viruses or botnets. Such nodes include trusted networks,
load balancers, and proxy servers, where the data is transmitted in an unencrypted form. In
the worst case, the attacker may even steal the private key of the TLS certificate. It enables
him not only to analyze all the traffic but also to modify the payload. Situations of data
spoofing at vulnerable points are poorly researched. A new "chunking" method was
previously proposed and compared to the existing ones. Analytical studies have shown an
increase in the performance of the new "chunking" method due to the more efficient usage of
resources.
For efficient resource management in the multithreaded environment, it is necessary to load
all the processor cores equally. A decent solution to take advantage of the multithreading
environment is the operation queues. However, if an operation queue can't process
accumulated operations efficiently, the system goes into high load mode. Such behavior may
cause the system to run out of memory. The patterns of occurrence of high load modes have
been researched previously. To prevent the high load modes, the load balancing of the server
cluster method has been further developed, which can also be applied to a single server.
The new model of authentication and request processing in high load modes for IoT
components has been created. It differs from the existing ones in the ability to prevent high
load modes during the authentication of large payloads. It uses the limitation in the total
bandwidth as well as the limitation of the number of requests. The information system for the
IoT components, which implements the new model has been developed. In the test, several
experiments were conducted, which show: high load mode, the effect of limitation of the
bandwidth, effect of the limitation of the number of requests.