**8. Conclusion**

In this work, we propose a probabilistic hierarchical hidden Markov model (PHHMM) applied for IoT intrusion detection which is more efficient than the existing HHMM without compromising classification accuracy. The main idea of our model is to reduce the huge problem state space of IoT traffic through dimensionality reduction by PCA and SVD. The proposed model is tested on the CICISD2019 dataset to detect and predict DDoS attacks. We evaluated our model on major performance metrics including Accuracy, Precision, Sensitivity, and False Negative Rate, Specificity and shows that our scheme has better detection accuracy and low error rates compared to Naive Bayes and neural network classification algorithms. It shows that PHHMM achieves a comparable accuracy as HHMM, better than NB and NN, and better efficiency than HHMM.
