The rapid evolution of smart grids has necessitated advanced communication-aware demand response (DR) strategies to handle the uncertainties introduced by renewable energy sources and heterogeneous consumer behaviors. This paper proposes a novel IoT-enabled communication-aware residential demand response framework that tightly integrates MQTT-based real-time communication, Fuzzy C-Means behavioral clustering, hybrid ARIMA-LSTM load forecasting, and NSGA-II multi-objective optimization. Unlike conventional DR approaches that treat communication performance as a passive constraint, the proposed framework jointly optimizes energy objectives (electricity cost, Peak-to-Average Ratio, Composite Load Index) and communication metrics (latency, Packet Delivery Ratio, reliability) in a unified manner. Using real-world smart meter data from 80 households, simulation results demonstrate superior performance with 28% peak load reduction, 18–22% electricity cost savings, average latency reduction to 70 ms, and PDR improvement to 99.2% compared to conventional methods. The main contributions include the development of a fully integrated cyber-physical architecture, incorporation of behavioral clustering for consumer heterogeneity, and explicit communication-aware multi-objective optimization. This work provides a practical pathway toward more efficient, reliable, and consumer-centric smart grid operation.