What are the seven characteristics of biometrics that make it useful for security?

A biometric trait must be reliable to be a useful security tool. This article begins with an overview of the seven characteristics used to compare the reliability of biometric traits. It then looks at how well some common biometric traits satisfy those requirements. It considers how multimodal authentication and sensor fusion can enhance security and concludes with a glance at behavioral biometrics.

The seven characteristics of biometrics that make it useful include:

Universality: every person must possess the trait.
Distinctiveness: the trait must be sufficiently different between individuals.
Permanence: the trait must be sufficiently invariant over time with minimal changes.
Measurability: the trait must be quantitively described.
Acceptability: users must be willing to provide the required biometric data.
Circumvention: the trait must be difficult to imitate or forge.
Performance: the trait must have a high level of accuracy in distinguishing between individuals.

Individual biometric traits support varying levels of performance for the seven characteristics.

Examples of traits

Just because a biometric trait scores high on most or all the characteristics doesn’t necessarily make it suitable for a specific application, for example, DNA is highly accurate at identifying individuals and nearly impossible to circumvent, but it’s not easily implemented in a security system.

The rankings in Table 1 below are not absolute; they are qualitative, subjective, and can vary according to the specific implementation. Fingerprints and facial recognition have good overall scores and are widely used. Hand recognition can be problematic if a person has been in an accident and loses a digit. Iris recognition is more difficult to implement but is increasingly deployed in security applications. Voice recognition scores are relatively lower, and some artificial intelligence (AI) techniques have been developed to mimic a person’s voice and circumvent or confuse voice recognition systems.

Multimodal biometrics and sensor fusion

Multimodal biometric security systems use sensor fusion to combine traits with different performance levels for key characteristics like facial recognition and fingerprints. However, simultaneously using diverse sensor sources like facial recognition and fingerprints can increase system complexity. Using data from a similar source, such as combining a fingerprint with a finder vein analysis, can simplify the implementation of multimodal identification.

A multimodal identification system will have two or more decision channels. Data and analysis can be fused at several levels of analysis, including the sensor level, feature level, matching score analysis, and decision classifiers (Figure 1). If not properly designed and implemented, multimodal biometric recognition can significantly delay obtaining the final classification, limiting its utility.

Behavioral biometrics

Rather than focusing on physical traits, behavioral biometrics measures and distinguishes patterns in device users’ behavior. Behavioral biometrics have been developed to increase online transaction security and detect potential fraud.

Behavioral biometrics tends to be inherently multimodal. Behavioral biometric verification methods include keystroke dynamics, mouse usage patterns, and swipe and touch analysis. Device indicators like IP addresses and geolocation are also employed in some systems.

Summary

The seven characteristics of biometric traits that make them useful are universality, distinctiveness, permanence, measurability, acceptability, circumvention, and performance. Various biometric traits have different performance levels for those characteristics, sometimes limiting their applicability in a specific security scenario. Sensor fusion can overcome the limitations of individual traits and provide more robust security. Behavioral biometrics has been developed for online security to rely on behavioral patterns instead of physical traits to identify trusted individuals.