A METHOD FOR SECURE PROGRAMMING OF DECENTRALIZED APPLICATIONS THROUGH COMPOSITIONAL VALIDATION OF SIGNED OPERATIONS

Abstract

This article studies authorization of signed operations in decentralized applications based on smart contracts. A valid signature authenticates a message but does not establish whether it remains executable after changes in state, privileges or context. The objective is to formalize a compositional admission method and assess attack blocking, false rejections and computational costs. The study combines a literature review, state transition modeling and descriptive analysis of aggregate simulation data. The specification integrates signature, domain, nonce, expiry, current role, application limits and invariant checks. Conditions for at most one successful execution are established within one domain and a consistent state history. Validation and state updates are ordered before external interactions. The model dataset includes seven attack classes and legitimate requests; component contributions are examined through ablation. The full configuration blocks 99.26% of attacks, compared with 25.57% for basic signature verification. False rejections increase from 0.70% to 1.25%, while mean modeled costs rise by 31.02%. Removing the domain, nonce or current role check is associated with pronounced losses for the corresponding attack classes. Aggregate effectiveness depends on the class distribution, and a high average rate does not eliminate the need to investigate missed attacks. These results are descriptive: missing generator code and trial records prevent independent reproduction and limit causal interpretation and significance testing. Modeled costs are not measurements of a deployed EVM contract. Practical value lies in a coherent authorization specification and test scenarios linking each check to a transition property. Further work should implement the method in EVM, investigate residual errors and publish reproducible trials, including contract wallets and cross-function reentrancy.