THE ROLE OF EDGE COMPUTING IN REDUCING LATENCY DURING DATA PROCESSING IN AUTONOMOUS SYSTEMS

Abstract

The relevance of this study is determined by the rapid development of autonomous systems in transportation, industry, robotics, and cyber-physical environments, which is accompanied by a steady increase in the volume of sensor and control data and by stricter requirements for real-time data processing. Traditional centralized cloud-based approaches do not always provide the required level of performance and control determinism due to network latency, connection instability, and dependence on remote computational resources. In this context, the need for scientifically grounded application of edge computing as a tool for latency reduction and for improving the robustness of autonomous systems becomes increasingly important. The purpose of the article is the substantiation of the possibilities of using edge computing to reduce data processing latency in autonomous systems and to improve the performance, reliability, and security of their operation. The research methods are based on a system analysis of architectures of autonomous and cyber-physical systems, generalization of contemporary approaches to distributed data processing, comparative analysis of centralized and edge computing models, and logical and analytical assessment of the impact of latency and its variability on the effectiveness of closed-loop control systems. The research results indicate that the localization of time-critical computational processes at the edge computing level creates prerequisites for reducing overall data processing latency, decreasing its variability, and increasing the determinism of control decisions. The generalization of empirical data demonstrates that edge-oriented architectures contribute to the stabilization of closed-loop control systems, improvement of control accuracy, and enhancement of system autonomy under dynamic operating conditions. The highest effectiveness of this approach is achieved in hierarchical architectures in which operational functions are implemented at the edge level, while time-noncritical tasks are moved outside the core control loops. The conclusions confirm the expediency of applying edge computing as a system-forming element of modern autonomous systems, provided that an architecturally balanced distribution of computational functions is ensured. At the same time, the effectiveness of this approach is constrained by challenges related to the scalability of heterogeneous infrastructures, consistency of distributed decisions, and information security, which necessitates comprehensive interdisciplinary solutions. Prospects for further research are associated with the formalization of models for latency evaluation in multilevel edge architectures, the development of coordinated decision-making methods in distributed autonomous systems, and the creation of adaptive security mechanisms for edge computing environments.