Biometrics is defined as the measurement of life signs. One of the main aims of current security research is to acquire biometric data of sufficient detail and reliability for verification or identification of individuals. Technologies used for data acquisition include fingerprint patterns,1 iris scans, facial images, and voice recognition2 using methods such as the pressure and timing of keystrokes.3 We demonstrate the feasibility of a new class of biometric measurement based on electric-field sensors and highlight its use in physiological-signal acquisition.
Our newly developed technique is sensitive to both electrical and movement components of physiological signals. However, it is a passive approach without the need for excitation signals. Alternative methods for remote physiological-data acquisition, such as optical vibrocardiography4 and microwave Doppler radar,5 suffer from being active: they require irradiation of the subject by laser- or microwave-excitation signals. They detect chest-wall movement—as opposed to electrophysiological activity—and therefore only operate reliably from the front of the subject.
The electric-field (or potential) sensors (EPS) we developed are most easily understood by assuming that they are perfect voltmeters.6 We can approach the ideal infinite-input impedance more closely than previously achieved. The electric field is measured simply by placing the sensor at the point of interest. The technology has been demonstrated in a number of areas including body electrophysiology (through clothing), nondestructive testing of composite materials, imaging of integrated circuits, and charge distributions.
EPS uses a combination of techniques currently found in conventional laboratory electrometers, such as guarding, bootstrapping, and neutralization. However, these on their own are not sufficient. An additional mechanism is required to provide a stable input bias current that does not compromise sensor performance. The latter is vital if the ultrahigh impedance is to be maintained when the sensor is weakly coupled to a signal by a small capacitance through either a dielectric spacer or an air gap. This combination of features, integrated into an active sensor, makes EPS unique. Signal-discrimination performance in a noisy environment was recently enhanced by a factor of up to 50dB. This was achieved by including frequency-selective switched-capacitor filters in the sensor's feedback loop to reject the main external-noise components. These filters are externally programmable (using a clock frequency) to define their central frequency. It is usually sufficient to attenuate the fundamental and first three harmonics of the interference signal. This method is particularly relevant for remote detection of physiological signals in the presence of ambient electrical noise due to the proximity of operational computer equipment or other electrical appliances.7
Figure 1 shows a block diagram of a generic EPS typically achieving an input impedance of ~1018Ω, input capacitance of ~10−15F, and a frequency range from quasi-DC (<1mHz) to ~200MHz. In reality these specifications are not all attainable in the same sensor but each is achievable in some combination. For example, we have verified that stable operation is possible with an effective input capacitance as low as 10−16F in an unshielded environment, albeit with a restricted bandwidth of ~10kHz.
Figure 1. Block diagram of a generic electric-potential sensor.
In Figure 2 data from a seated subject is shown, using a chair-mounted sensor at 10cm from the back of the subject with a sensor electrode 2cm in diameter. The data differs from the usual electrocardiogram response due to the presence of an additional movement signal. The coupling capacitance between the sensor and the skin surface is estimated at ~10−13F. Figure 3 shows the corresponding measurement from the front at a distance of 40cm from the chest. All data was obtained in a noisy laboratory close to active line-operated (50Hz) equipment and computers.
Figure 2. Signal from sensor at 10cm from the back of a seated subject.
Figure 3. Signal from sensor at 40cm from the chest of a seated subject.
These preliminary results illustrate that our novel electric-field sensor can monitor physiological signals remotely in a laboratory affected by significant levels of electrical noise. Development of large-scale 2D sensor arrays in the near future will significantly enhance their data-collection capability. Further work is required to establish to what extent this technique fits the criteria for a biometric system.
Department of Engineering and Design
University of Sussex
Robert J. Prance is a reader in electronic engineering and head of the Centre for Physical Electronics and Quantum Technology. He has published over 150 journal papers spanning low-noise electric and magnetic sensors, nuclear magnetic resonance, nondestructive testing, and nonlinear and quantum systems, and was a joint recipient of an Institute of Physics 2002 measurement science and technology ‘Best Paper Award.’