A Closed-Loop Predictive Intelligence Framework for Adaptive Resource Allocation, Routing, and Data Management in Cloud Computing

Cloud communication platforms are running under a variety of workloads that are heterogeneous, bursty and latency sensitive. Static provisioning and reactive rule-based control can only react after workload changes have already taken place, resulting in unnecessary latency, resource wastage and administrative burden. This study suggests an intelligent cloud communication framework consisting of workload forecasting, adaptive resource allocation, routing optimization, edge, assisted processing, anomaly detection, and automatic data administration in a closed feedback loop. The methodology allows for the formalization of workload prediction and multi-objective decision making with measurable variables for latency, throughput, resource utilization, scalability and energy consumption. The four configurations (Traditional Cloud, Optimized Cloud, ML Based Cloud, and Predictive ML Cloud) and five workload/deployment conditions are used for a comparative evaluation. The data-processing efficiency of the Predictive ML configuration is the highest reported (93–96%) and the latency (60–150 ms under the specified load) is the lowest reported, its scalability performance is the highest reported (93–95%). The above values are to be taken as the comparative values as reported by the manuscript and are not to be reproduced independently. The updated paper provides a better problem formulation, research questions, mathematical notation, reproducible experimental setup, literature synthesis of 2021-2026, explicit research gaps, and a discussion oriented towards validation.