An attacker is said to be engaging in an attack using a man-in-the-middle strategy when they pretend to be a legitimate party involved in a discussion. The most important things are a customer and a program. This attack has as one of its goals getting either eavesdrop on the conversation or to make it appear as though the interaction is proceeding without any problems. Materials and Methods: In terms of correlation values, AlexNet is at 66.77%, while Novel Convolutional Neural Network is at 96.76%. The fact that the P value is lower than 0.05 is an indication that these two methods will have a significant impact on the statistical analysis. This is because of the fact that the P number exists. The statistical significance between the two approaches can be inferred from the fact that the P value is less than 0.05. With the goal of discovering occurrences of man-in-the-middle attacks on IoT devices, Novel Convolutional Neural Networks and AlexNet are trained and tested in different ways. In order to determine approximately 85% of the power, a G Power test is utilized, 0.05 is the value that has been assigned to α, and 0.85 is the value that has been assigned to power. Result: According to the data, the Novel CNN achieves a higher accuracy rate (93.966% vs. AlexNet, p < 0.05), and the significance value is 0.001. Conclusion: Compared to AlexNet, Novel CNN achieves much higher accuracy.