Behavioral features
With many persons, biometric features based on behavior are not stable over time. One example is a signature, which can change considerably during the course of a person’s life. However, it is rare for these changes to occur suddenly, and they are usually quite gradual and slow. Many systems that use biometric features therefore use adaptive methods that accept any changes in the feature that are detected during a correct identification as a new reference pattern, which is then stored in the smart card.

Typing rhythm
It has been determined that there are large differences in the manners in which different individuals type characters on a keyboard. These primarily relate to the pauses between individual letters. This can naturally be used as a biometric feature for identification. The procedureworks by having the person to be identified type a prescribed character string (which is different for each test) on a keyboard. The computer to which the keyboard is attached evaluates the typing rhythm as the character string is typed. A text chosen by the user can also be used to evaluate the typing rhythm, but this requires more characters to be typed than with a prescribed text. The primary advantage of this method is that it does not need any additional hardware, since in most cases a keyboard and computer are already available. Unfortunately, between 100 and 150 alphanumeric characters are needed for the test, and they must be typed using the 10-finger system. This is the main drawback of this method.

Vocal features
Like the face, a person’s voice is characteristic of the person, so it can also be used for identification purposes. The person to be identified speaks one or more sentences into a microphone. These must be different for each session, since otherwise the system could be attacked very easily by playing back a previous identification session, which for example may have been recorded on magnetic tape. The waveforms of the spoken text are subjected to a Fourier analysis, which yields the characteristic frequency spectrum of the speaker. This is then compared with a reference value to determine whether the speaker’s identity is genuine. The entire gamut of modern computational wizardry, such as fuzzy logic, neural networks and the like, is also employed with this method. Of course, this method also has its shortcomings.Aperson’s voice is very strongly influenced by his or her current bodily condition. Furthermore, all background noises must be reliably filtered out to make unambiguous spectral analysis possible in the first place. A different sentence must be spoken for each test to prevent recorded speech from being played back, which very much complicates the procedure and makes recognition more difficult. However, these technical difficulties are offset by good user acceptance, which makes this a very attractive biometric identification method.

Dynamic signature
The only identification method that is commonly used in everyday life is writing a signature. Due to its very individual character, a signature can also be used as a biometric feature.With a static method, the signature is evaluated after it has been written. With a dynamic method, by contrast, measurements are made while the signature is being written. The static method is only of theoretical interest, since it cannot distinguish a photocopied signature from a genuine one. The parameters measured in the dynamic method may for example be the general form of the signature, the speed, acceleration and pressure of the pen on the writing surface, and the time required to write the signature. A special pen, or a special pad that can sense the parameters to be measured, can be used to make the measurements. Figure 8.10 shows an example of a possible arrangement in which an ordinary pen is used on a special pad, and Figures 8.11 through 8.14 show examples of measured signals that can be used as the basis for a biometric identification process. Pressure sensors are located at the intersections of the grid wires, and their signal amplitudes are transmitted to the computer via conditioning logic. The computer can then use various algorithms to process the measured data into a standardized format and compare the results with a stored reference pattern. Using a dynamic signature for purposes of identification has the highest degree of acceptance of all personal identification methods, since signatures are used daily by everybody in almost the same fashion. However, here the technical solutions are not simple, since signatures change over time and are never fully identical. You need only consider the difference in your signature if you write it while sitting or standing to appreciate the truth of this.