META-COMPUTING

Abstract

This monograph presents a conceptual and architectural framework for Meta-Computing as a mechanism for adaptive optimization of AI workloads in distributed computing systems. The work proposes a Meta-Computing Architecture (MCA) that integrates runtime observation, meta-level analysis, adaptive control, and continuous feedback to enable computational systems to evaluate and modify their execution strategies during runtime. The monograph introduces the Adaptive Meta-Optimization Algorithm (AMOA), Workload Complexity Index (WCI), Meta-Efficiency Score (MES), and Meta-Computing Feedback Loop (MCFL) as the principal components of the proposed framework. A mathematical model is developed to represent the relationship between computational state, workload complexity, optimization decisions, and feedback-driven adaptation. The proposed framework is evaluated through a simulation-based experimental study using distributed AI workload scenarios of varying complexity. The evaluation compares static execution with the proposed Meta-Computing approach using execution time and meta-efficiency as principal measures. The results indicate improved execution performance under the evaluated simulation conditions, with greater reductions in execution time observed as workload complexity increases. The work is presented as a foundational framework for further research into adaptive, self-optimizing, and increasingly autonomous computing systems. The monograph also identifies the limitations of simulation-based evaluation and discusses future directions including real-world distributed deployment, cloud and edge integration, learning-based meta-optimization, autonomous resource management, and advanced self-aware computing systems.