EVALUATING THE COMPLEXITY OF IMPLEMENTING DATA PROCESSING ON IOT-DEVICES

Abstract

A method for implementing data processing in the Internet of Things systems, based on the end device, is considered. While existing approaches are based on the Cloud or Edge paradigm, processing on the end device of the IoT system allows you to reduce the amount of data transmitted at the initial stage. Correlation processing is an effective way to detect signals, however, practical implementations with a long pulse response duration are not suitable for low-power devices. The paper compares a number of implementations with an estimate of the number of computational operations, as well as an improved approach that reduces not only the number of operations, but also the processing delay. In addition, the implementation study is carried out when implementing on the basis of field programmable gate arrays (FPGA). The directions related to the research of signal processing directly on intermediate devices and, especially, on end devices (on-sensor processing) are represented to a lesser extent. This fact is due to the fundamental limitations of the end devices and systems of the Internet of Things, as well as the contradictory requirements. First of all, the devices should be as cheap as possible, autonomous, compact and at the same time have low power consumption. These requirements limit the performance of end devices. The network, in turn, must provide the required quality of service (QoS) and the speed and reliability of data transmission. The implementation of data processing on end devices in IoT systems is of great scientific and practical interest. This article will consider an approach based on correlation processing (consistent filtering). The traditional approach with large orders of filters on low-power, from a computational point of view, devices is redundant and not always feasible.