Abstract— Wireless Sensor Networks (WSNs) are used in smart and mission-critical applications because their wireless technology enables remote monitoring. The distributed design of WSNs combined with their limited computing power creates a situation where they become highly susceptible to advanced cyber threats. The research presents an SDN-oriented hybrid intrusion detection system which uses Fuzzy CNN-LSTM architecture at network border points to solve these technical problems. The proposed model uses a convolutional neural network (CNN) to extract spatial features and a long short-term memory (LSTM) network to model temporal behavior while its fuzzy inference layer improves prediction accuracy by handling softmax output uncertainty to decrease false alarm rates. Edge nodes conduct sensor node traffic data processing and analysis, while SDN controllers provide centralized network monitoring and defense capabilities through flow control and traffic redirection and distribution balance operations. The framework uses the WSN-DS and WSNBFSF datasets to test its performance under binary classification between normal and attack modes. The proposed method achieved a detection accuracy of 99.67% and a precision value of 98.33% and a recall rate of 98.16% and an MCC value of 98.07% when tested with WSN-DS. Our model achieved 99.2% accuracy and 97.75% precision and 96.9% recall and 96.85% MCC on the WSNBFSF dataset while outperforming standalone CNN and LSTM models which maintained low inference latency for real-time deployment across both datasets. The results demonstrate that deep learning combined with fuzzy uncertainty reasoning and SDN control creates a sustainable solution to detect intrusions across resource-limited WSN environments.