4. PERFORMANCEOFBIOMETRICSKEY GENERATION Inthissection,weevaluatethesecuritymeasurementofreliability and min-entropy of the keys generated from ECG signals. To investigate the reliability of the ECG recognition systems for personal authentication with smart door locks, we extract the keys at different times from the same person to represent how the keys are robust. Higher reliability corresponds to better agreement of the keysandsmall intra-class variation. Toprove thatthe keys arerandom we considered min-entropy. Large entropy indicates excellent distinction among the keys generated by different people and how the keys are robust against an attacker. If the min-entropy value is close to 1 that means it has good quality as a key. We present the results in Table 1. Iris recognition is another biometric modality for its high identification accuracy. We apply the same algorithm with the same parameters on iris features as well to compare its performance with ECG. The iris results are shown in the last column of Table 1. Comparing the two biometrics modalities on average, ECG provides both the longest key, reliability, and highest 1-bit entropy while iris-based generation exhibits less effective results. As shown in Table 1, we achieved 727 as the average key length for normal ECG signals: roughly 200 bits more than producedbytheirismodality. Wehaveusedfreelyavailabledatabases fortwobiometricmodalities,thePTBdiagnosticdatabase[16]and iris database [8] to test the authentication process. 4.1 BiometricDataRandomness 4.1.1 Distributed Uniqueness Analysis For authentication purposes, biometric keys should be as unique