ALGORITHM FOR STABLE DATA AGGREGATION IN DISTRIBUTED ENVIRONMENTAL MONITORING NETWORKS BASED ON THE TRUSTED WEIGHTING METHOD

Abstract

This paper focuses on the development and comprehensive analysis of a stable data aggregation algorithm for decentralized IoT networks used in environmental monitoring. The collection and processing of multidimensional information in such large-scale distributed networks are significantly complicated by the presence of stochastic environmental noise and Byzantine nodes—malicious or hardware-faulty sensors that systematically or chaotically generate anomalous readings. The primary aim of this study is to create and empirically evaluate a mathematically grounded algorithm capable of maintaining system homeostasis and ensuring high computational accuracy without relying on a single central coordinator, thereby eliminating the Single Point of Failure problem. The research encompasses an in-depth vulnerability analysis of the classic Mean-gossip algorithm and a detailed justification of the proposed hybrid method, TWTMOM (Trust-Weighted Trimmed Median-of-Means). This approach innovatively combines

multidimensional statistical anomaly trimming using the Mahalanobis distance, aggregation based on the Median-of-Means method, and an adaptive Markovian node reputation weighting system. To practically evaluate the algorithm's efficiency and scalability, a highly concurrent software simulation was developed using the Golang programming language, which accurately mimics the operation of a high-load network consisting of 1000 sensors. A comparative analysis was conducted based on the criteria of resilience to massive data poisoning and stealth gradient drift attacks, measuring the Root Mean Square Error (RMSE). The findings revealed that the classic method exhibits a critical vulnerability even with 5-10% compromised nodes. In stark contrast, the newly developed TWTMOM algorithm demonstrated unprecedented reliability, maintaining a minimal level of mathematical error even under the simultaneous destructive influence of up to 40% anomalous sensors in the network. The study results firmly confirm that the TWTMOM algorithm is an optimal and highly scalable solution for implementation in mission-critical decentralized environmental and industrial monitoring systems, where unquestionable cybersecurity, Byzantine fault tolerance, and data integrity are top priorities.

Keywords: IoT, environmental monitoring, data aggregation, Byzantine fault tolerance, Median-of-Means, trust weighting, cybersecurity.