A comprehensive systematic literature review on artificial intelligence for error correction and modulation schemes in next-generation satellite communications

• Published in: The Artificial intelligence review Vol. 58; no. 10; p. 321
• Main Authors: Sharma, Ekta, Davey, Christopher P., Deo, Ravinesh C., Carter, Brad D., Salcedo-Sanz, Sancho
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 21.07.2025; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Artificial neural networks; Codes; Coding theory; Communications systems; Computer Science; Convolutional codes; Efficiency; Error correcting codes; Error correction; Error correction & detection; Fault tolerance; Hamming codes; Literature reviews; Low earth orbits; Machine learning; Modulation; Performance measurement; Power efficiency; Satellite communications; Wireless communication; Coded modulation; Error correction; Power efficiency; Artificial intelligence; LEO satellite communication
• ISSN: 1573-7462; 0269-2821; 1573-7462
• DOI: 10.1007/s10462-025-11317-4

Communication systems continue to embrace the potential of Artificial Intelligence (AI) in error correction codes (ECC) with coded modulation schemes (CMS). Despite this, there remains a substantial performance gap in AI methods in terrestrial and satellite communication systems. Additionally, AI and power efficiency for Low Earth Orbit (LEO) satellites have shown a critical gap. To the best of the author’s knowledge, this is the first Systematic literature review attempting to bridge this vital gap to boost efficiency and add fault tolerance. From 389 articles published between 1993 and 2023, the construction and performance of 33 AI algorithms have been comprehensively reviewed for 16 ECC, seven higher-order CMS, and LEO satellites. Based on four key parameters: error correction, modulation, power, and energy efficiency, the PRISMA strategy with a 27-item checklist was adopted and 63 studies were selected to investigate the AI-based performance of terrestrial (40-studies) and LEO satellites (23-studies). Analysing nine performance metrics, Convolutional Neural Network was the most popular choice (20.6%) with an accuracy of 99% and SNR from 6-20dB, followed by Deep Neural Network (19.04%). The least used algorithm was Reinforcement learning (9.52%). Modified Reed Solomon codes showed the best measurement of power consumption and error rate. Adaptive LDPC codes provided a 45% increase in energy efficiency with an 11% computation decrease. Considering appropriate merits and challenges, the review identifies, discusses, and synthesises AI results to create a summary of current evidence for terrestrial and LEO satellites contributing to evidence-based practice for future researchers.