Model for storing structured data in heterogeneous distributed programmable mobile systems

Abstract

The rapid development of mobile ad hoc networks and edge computing paradigms necessitates the design of resilient storage architectures capable of operating efficiently under stochastic topology changes. Existing decentralized storage approaches exhibit prohibitive signaling overhead and integrity degradation during critical link failures, making adaptive information management highly relevant. The purpose of the article is to ensure high availability and integrity of information in distributed programmable mobile systems in environments with dynamic network topologies and limited resources of mobile nodes. To achieve this, the following tasks were formulated: to propose a hybrid structured data storage model based on semantic standards; to develop an adaptive replication algorithm predicting device trajectories; and to experimentally evaluate the effectiveness of the developed solutions. The following methods were applied: graph theory for network topology modeling, the software-defined networking and software-defined storage paradigms for resource orchestration, and mathematical modeling techniques for calculating link expiration time. The scientific novelty of the work lies in the improvement of the structured data storage model through the integration of software-defined network paradigms and heuristic adaptive replication with competitive aging mechanisms, resolving the severe compromise between consistency and availability. The practical significance of the obtained results lies in their direct applicability to intelligent transportation systems, smart cities, and emergency response networks during disaster recovery. The most significant results demonstrate that the proposed model maintains a request delivery success rate of eighty-three point nine percent at movement speeds up to one hundred kilometers per hour while simultaneously reducing the average data lookup delay from two hundred and seven milliseconds to eighty-seven milliseconds. Conclusions. Developing a hybrid model based on intelligent orchestration and a two-level consistency policy successfully enables bounded inconsistency during network partitioning, enabling continuous application operability within isolated segments with a guaranteed delayed restoration of global semantic data integrity.