Methods for optimizing data storage and transmission in distributed information systems, taking into account the bandwidth limitations of communication channels

Abstract

Distributed information systems used in the construction industry and architectural design are experiencing an exponential increase in the volume of data related to building information modeling, including BIM models in IFC and RVT formats, geospatial information, point clouds, executive documentation, and engineering calculations. However, limited communication bandwidth, particularly in remote construction sites using satellite and mobile connections, can lead to significant synchronization delays, version conflicts, excessive network traffic, and data loss risks. Analysis of the structural and semantic organization of such data reveals their high hierarchical structure, the repeatability of typical elements, and the presence of spatial relationships, which creates the preconditions for the development of specialized compression methods that significantly outperform universal compression algorithms such as GZIP, LZ4, or Zstandard, especially for point clouds and IFC models, where compression ratios can increase by 3 or more times due to the consideration of entropy characteristics and structural patterns. The integration of delta encoding, incremental replication, and content-addressable storage methods minimizes the amount of information transmitted by focusing on changes relative to base versions, while the use of machine learning algorithms to predict user access patterns enables proactive caching, intelligent data distribution across nodes, and dynamic adaptation of replication strategies. Comparative modeling demonstrates that adaptive hybrid replication strategies that switch between synchronous and asynchronous modes based on the channel status and the semantic significance of updates reduce network load by up to 9 times, reduce synchronization delays by up to 4,8 times, and increase resilience to connection failures while simultaneously reducing the likelihood of version conflicts. Similarly, the transition to an adaptive distributed storage architecture using machine learning-based forecasting and traffic prioritization based on the impact on design elements results in a 85,7% reduction in traffic compared to fully replicated solutions, maintaining data availability above 99,7% and optimizing response times within acceptable limits for collaboration. The cumulative application of these approaches in a single platform that takes into account mathematical models of information theory, queuing theory, and the correlation between channel characteristics and optimization effects, provides a comprehensive increase in the efficiency of system operation, even in the context of extremely limited bandwidth, opening up opportunities for the full-fledged deployment of BIM technologies on remote infrastructure facilities, accelerating the coordination processes, and significantly reducing the costs of telecommunications infrastructure in the construction industry.