PROJECT TITLE :
Improving ZigBee Device Network Authentication Using Ensemble Decision Tree Classifiers With Radio Frequency Distinct Native Attribute Fingerprinting
The popularity of ZigBee devices continues to grow in home automation, transportation, traffic management, and Industrial Management System (ICS) applications given their low-price and low-power. However, the decentralized design of ZigBee ad-hoc networks creates distinctive security challenges for network intrusion detection and prevention. In the past, ZigBee device authentication reliability was enhanced by Radio Frequency-Distinct Native Attribute (RF-DNA) fingerprinting employing a Fisher-based Multiple Discriminant Analysis and Most Probability (MDA-ML) classification method to distinguish between devices in low Signal-to-Noise Ratio (SNR) environments. But, MDA-ML performance inherently degrades when RF-DNA options do not satisfy Gaussian normality conditions, that usually happens in real-world scenarios where radio frequency (RF) multipath and interference from alternative devices is gift. We introduce non-parametric Random Forest (RndF) and Multi-Category AdaBoost (MCA) ensemble classifiers into the RF-DNA fingerprinting arena, and demonstrate improved ZigBee device authentication. Results are compared with parametric MDA-ML and Generalized Relevance Learning Vector Quantization-Improved (GRLVQI) classifier results using identical input feature sets. Fingerprint dimensional reduction is examined using three methods, specifically a pre-classification Kolmogorov-Smirnoff Take a look at (KS-Take a look at), a post-classification RndF feature relevance ranking, and a GRLVQI feature relevance ranking. Using the ensemble strategies, an $rm SNR=18.0$ dB improvement over MDA-ML processing is realized at an arbitrary correct classification rate $(hbox%C)$ benchmark of $hbox%C=90hbox%$ ; for all $rm SNRin [0, thirty]$ dB consid- red, $hbox%C$ improvement over MDA-ML ranged from 9p.c to twenty fourp.c. Relative to GRLVQI processing, ensemble ways once more provided improvement for all SNR, with a best improvement of $hbox%C=10hbox%$ achieved at rock bottom tested $rm SNR=0.0$ dB. Network penetration, measured using rogue ZigBee devices, show that at the $rm SNR=12.zero$ dB $(hbox%C=90hbox%)$ the ensemble methods properly reject 31 of thirty six rogue access makes an attempt based on Receiver Operating Characteristic (ROC) curve analysis and an arbitrary Rogue Settle for Rate of $rm RAR < 10hbox%$. This performance is healthier than MDA-ML, and GRLVQI that rejected twenty five/thirty six, and twenty eight/thirty six rogue access makes an attempt respectively. The key benefit of ensemble method processing is improved rogue rejection in noisier environments; gains of 6.0 dB, and 18.zero dB are realized over GRLVQI, and MDA-ML, respectively. Collectively considering the demonstrated $hbox%C$ and rogue rejection capability, the use of ensemble strategies improves ZigBee network authentication, and enhances anti-spoofing protection afforded by RF-DNA fingerprinting.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here