Vol 13 Issue 2 October 2026-March 2027
D. A. Onuma, I. M. Ossai, A. H. Mini, B. M. Kpobe
Abstract: Decentralized artificial intelligence enables collaborative learning without centralized data collection but remains vulnerable to model poisoning, malicious participants, and insecure execution environments. This study developed a Decentralized Artificial Intelligence Framework with Confidential Computing and Smart Contract-Based Trust Verification to strengthen privacy, integrity, and scalability. Design Science Research Methodology (DSRM) guided framework development and evaluation, while Object-Oriented Analysis and Design Methodology (OOADM) supported modular system design. The framework integrated peer-to-peer learning, Trusted Execution Environments, remote attestation, blockchain-anchored model updates, smart contracts, and adaptive reputation management. Experimental evaluation used seven benchmark datasets: CICIDS2017, CSE-CIC-IDS2018, UNSW-NB15, CICIoT2023, TON_IoT, Bot-IoT, and NSL-KDD. Assessment covered classification effectiveness, trust verification, privacy, attack resilience, computational overhead, latency, throughput, and scalability. Accuracy and F1-score generally exceeded 97%, while integrity verification exceeded 99%. Compared with the strongest evaluated baseline, blockchain throughput increased by 67.1% and maximum supported nodes by 60.0%, while confidential-execution overhead, false-trust rate, and data-leakage rate decreased by 54.4%, 52.3%, and 51.7%, respectively. Detection of model-poisoning, Byzantine, Sybil, and malicious-node attacks also improved, although residual attack success remained. These findings demonstrate the potential of integrating confidential execution and programmable trust governance to secure collaborative AI in distributed cloud and IoT environments. Further operational evaluation is required to establish real-world effectiveness.
Keywords: Decentralized artificial intelligence; confidential computing; smart contracts; trust verification; DSRM; OOADM.
Title: A Decentralized Artificial Intelligence with Confidential Computing and Smart Contract-Based Trust Verification
Author: D. A. Onuma, I. M. Ossai, A. H. Mini, B. M. Kpobe
International Journal of Recent Research in Mathematics Computer Science and Information Technology
ISSN 2350-1022
Vol. 13, Issue 2, October 2026 - March 2027
Page No: 1-28
Paper Publications
Website: www.paperpublications.org
Published Date: 10-November-2026
