Energy Optimization in Smart Networks using Machine Learning-Driven Fog Computing to Reduce Unnecessary Cloud Data Transmission
Nandini Gowda Puttaswamy, Anitha Narasimha Murthy · Engineering Technology & Applied Science Research · 2025
Smart network power management is obstructed by the increased data transmission complexity and inefficiency in conventional Cloud Computing (CC) methods. Centralized processing in CC leads to increased latency, increased energy consumption, and bandwidth constraints, making it not suitable for real-time systems. To address such constraints, this paper proposes a Machine Learning-driven Fog Computing (ML-FC) framework, which integrates Machine Learning (ML) with Fog Computing (FC) to improve data processing and energy efficiency. ML-FC is proposed to operate in structured steps. First, IoT devices send real-time data, which are routed to the fog layer for preliminary processing. Second, an ML-based model filters and prioritizes the data to process only significant information, reducing computational overhead. Third, data are processed in priority in the fog layer, reducing bandwidth usage and response times. Fourth, only significant processed data are forwarded to the cloud for storage and advanced analytics, significantly reducing unnecessary transmissions. Experimental results indicate that the ML-FC framework achieves a 0.35% reduction in energy consumption, a 0.28% reduction in latency, and a 0.22% improvement in network throughput compared to conventional methods. The framework provides improved scalability, real-time decision-making, and network efficiency. It is highly useful in healthcare monitoring, smart cities, industrial automation, and intelligent traffic management. The proposed technique enables a more efficient, adaptive, and energy-sensitive system for next-generation smart networks.