An Energy-Efficient Edge–AI Framework for Real-Time IoT Analytics in Smart Communication Networks

Abstract

The fast growth of Internet of Things (IoT) applications has heightened the need of real-time edge intelligence in intelligent communication networks. Even though Edge -AI lessens cloud reliance and enhances responsiveness, it adds a large power overhead since it has linked both communication and computation operations. Existing solutions are usually more efficient in terms of resources utilised in transmission or processing on an independent basis thus, result in inefficient tradeoffs on energy versus latency and little cross layer coordination. This paper introduces a cross-layer Edge-AI architecture that would optimise both wireless transmission power and edge-CPU frequency to attain real-time IoT analytics, which are energy-efficient. Coherent characteristic analytical framework is created to describe the rate of communication, computation load, total power usage, and latency. The ensuing multi-objective optimization problem reduces a weighted power constrained energy-latency cost constraint with frequency and delay constraints. An adaptive resource allocation algorithm is a dynamically changing program which changes the parameters of transmissions and computation based on the network conditions and workload intensity. Simulation analysis shows that the proposed structure can save the overall energy use up to 42 percent and also cut the end-to-end latency by about 28 percent compared to the traditional cloud-based and base edge configurations, without impacting the accuracy of inference. Such findings indicate the efficiency of cross-layer optimization that is integrated with the aim of deploying sustainable and scalable Edges -AI deployment in future IoT communication networks.