© 1999 Robert A. Freitas Jr. All Rights Reserved.

Robert A. Freitas Jr., Nanomedicine, Volume I: Basic Capabilities, Landes Bioscience, Georgetown, TX, 1999

7.4.2.1 Mechanical Inmessaging

Perhaps the simplest example of user inmessaging is finger-drumming on a hard surface such as a tabletop, a method which we will now examine in some detail. A ~15 gm finger may be drummed at a maximum of ~5 Hz^920,921 over a 2 cm path length at a maximum velocity of ~0.2 m/sec; if the impacting finger is brought to a halt within 1 mm of the ~1 cm² fingertip and ~50% of the kinetic energy is converted into acoustic energy, the phalangeal impact produces a maximum compression wave of ~0.3 atm (~300 watts/m²). The lightest audible tapping produces a pressure wave of ~0.01 atm (~0.3 watts/m²). Either pulse is easily detectable by ~10^-6 atm-sensitive nanorobot acoustic sensors (Section 4.5.1) located in the patient's hand or arm.

Consider five sensor-equipped nanorobots each positioned ~2 mm below the epidermis in the soft tissue just beyond the distal phalanx of five different fingers, and each with access to a local communications network (Section 7.3) spanning the hand. An impact pulse applied to a given finger reaches the sensor stationed in that finger in ~1 microsec by conduction through soft tissue at ~1540 m/sec. The acoustic pulse then travels down the phalanges and metacarpals, crosses the distal row of carpals, then travels back up through the metacarpals and phalanges of the other fingers mostly via bone conduction at ~3550 m/sec (Table 6.7), reaching the adjacent sensors 90-110 microsec after the initial impact. The pulse does not travel exclusively in bone as it must pass through 8-11 diarthrodial articulations (see Figs. 7.2, 8.25). Approximating each as a bone/soft-tissue interface with R = 66% reflection loss across each interface (Section 6.4.1), the 0.01-0.3 atm pressure pulse is reduced by at most 10^-4, still detectable even at the lightest tapping pressure.

Assuming nanorobots may possess synchronized internal clocks with at least 1-10 microsec accuracy (Section 10.1), then data sharing of pulse arrival times among the five nanorobots via the network allows the reconstruction of five unique temporal signatures of an impact event at any particular finger (Table 7.1). Conduction also occurs through soft tissue pathways which may be ~15 cm shorter than the typical osteal path length (X_path) used in Table 7.1, and may also produce much less signal attenuation. These soft-tissue pulses arrive ~120-160 microsec post-impact because of the slower conduction speed and thus should not be confused with the osteal signal. Still other pathways with widely varying conduction velocities will increase noise levels, but a brief training session should produce acceptable signal/noise ratios for event recognition especially since maximum pulse extinction time (~200 microsec) << pulse repetition time at ~5 Hz (~0.2 sec).

Interphalangeal amplitude attenuation provides an additional check on the identity of the tapping finger. An impact deceleration in ~5 x 10^-3 sec produces a single ~5 millisec pulse (~200 Hz) which from Eqn. 4.52 attenuates from 10,000 x 10^-6 atm to at most 9997 x 10^-6 atm over a ~20 cm path length through soft tissue, or to at most ~9900 x 10^-6 atm over a ~35 cm path length through bone -- either of which is detectable by nanorobot acoustic sensors sensitive to changes of ~1 x 10^-6 atm.

With single-finger coding, each tap represents log₂(5) = 2.32 bits/tap; a maximum tapping rate (~5 Hz) for each of the five fingers simultaneously (e.g., playing Rimsky-Korsakov's "Flight of the Bumblebee" on the piano) gives a maximum capacity of ~58 bits/sec. Multifinger coding is slower, since each tap becomes a 5-digit binary word representing log₂(31) = 4.95 bits/tap, and a maximum multi-finger tap rate of ~5 Hz gives a maximum capacity of ~25 bits/sec, about the same as Quastler's result.¹²⁶⁵ The practical limit for this modality is probably 150 bits/sec, roughly 2-100 English words/min or 10-500 decimal digits/min. Experiments by the MIT Wearable Computers Group with one-handed chorded keyboards (such as The Twiddler, manufactured by HandyKey¹⁵⁹⁵) suggest a requirement of ~5 minutes of training to learn the alphabet, ~1 hour to learn to touch type, and ~3 days to reach ~10 words/minute.

Many conceptually similar methods of mechanical inmessaging may be employed by a patient whose in vivo nanorobots have been appropriately programmed and positioned, including clapping/slapping of hands/feet (~3 bits/sec), mandibular clicking (~2 bits/sec), tongue-tapping on palate (~2 bits/sec), eyelid or eyebrow flapping (~2 bits/sec), dental/dermal drumming using a rigid object (~1 bit/sec), epiglottal pulsing (~1 bit/sec), thoracoabdominal diaphragmatic contractions (~1 bit/sec), or toe tapping (~0.5 bit/sec).

Other mechanical stimuli at the skin -- including pressure, touch, vibration, and tickle (see Schmidt¹⁶³³) -- may also be eavesdropped by medical nanorobots. Interestingly, the frequency response of these receptors may range as high as 800 Hz for Pacinian corpuscles (sensory organs located at the skeletal joints, tendons, and in the muscles;¹⁶³⁴ Table 7.3). Thus an eavesdropping subdermal nanorobot could receive >~100 Hz modulated signals through this channel that would only be perceived as "flat stimulus" by the conscious human patient.

Last updated on 19 February 2003