Article on Personal Area Networks

[ This article: = HTML, PDF, ASCII, PostScript ]

Personal Area Networks: Near-field intrabody communication

by = T. G. Zimmerman

Reprint Order No. G321-5627.

As electronic devices become smaller, lower in power requirements, and= less expensive, we have begun to adorn our bodies with personal informat= ion and communication appliances. Such devices include cellular phones, p= ersonal digital assistants (PDAs), pocket video games, and pagers. Curren= tly there is no method for these devices to share data. Networking these = devices can reduce functional I/O redundancies and allow new conveniences= and services. The concept of Personal Area Networks (PANs) is presented = to demonstrate how electronic devices on and near the human body can exch= ange digital information by capacitively coupling picoamp currents throug= h the body. A low-frequency carrier (less than 1 megahertz) is used so no= energy is propagated, minimizing remote eavesdropping and interference b= y neighboring PANs. A prototype PAN system allows users to exchange elect= ronic business cards by shaking hands.

We are heading toward an electronic future where information will be acce= ssible at our fingertips, whenever and wherever needed. Some of the compu= ting and communication equipment required to provide this intimate and im= mediate access to information will be incorporated into our attire. Just = as a glance at today's wristwatch saves a trip to the nearest clock, a gl= ance at tomorrow's wristwatch will replace finding a terminal to check e-= mail.

A person who carries a watch, pager, cellular phone, personal stereo, per= sonal digital assistant (PDA), and notebook computer is carrying five dis= plays, three keyboards, two speakers, two microphones, and three communic= ation devices. [1] The duplication of = I/O components is in part a result of the inability of the devices to exc= hange data. With proper networking these devices can share I/O, storage, = and computational resources.

The ability to share data increases the usefulness of personal informatio= n devices, providing features not possible with independent isolated devi= ces. Imagine the following scenario: I am at home preparing for the day a= nd want to find the time of my first meeting. I call out "When is my firs= t meeting?" The microphone in my watch transmits my voice through a serie= s of transponders distributed throughout my house to a voice recognition = computer that searches my calendar and sends back a response to a speaker= or visual display in my watch. When I leave my house the door senses my = departure and sends a message to my colleagues. When I approach my office= building, the door acknowledges me by opening, sends a message of my arr= ival to my colleagues, and uploads any new messages.

The scenario requires the user to wear a device that periodically transmi= ts a unique user code to allow a nearby stationary transceiver to identif= y, locate, and exchange messages with the user's device. Clearly, privacy= is a big issue in such scenarios. Privacy is both a right and a commodit= y. To maintain privacy control, a wearer must determine when the identifi= cation beacon is activated and what type of information can be transmitte= d. Retail stores could encourage shoppers to transmit needed demographic = information by providing perhaps a 5-percent discount to shoppers who lea= ve their user profile beacons on.

The capabilities of autonomous yet interconnected devices may transform t= he notion of ubiquitous computing [2] = to the concept of ubiquitous I/O. As wireless network ports become more c= ommon, there will be less need to carry around power-hungry processors an= d bulky mass storage. As the growth of wireless services (e.g., cellular = phones, pagers, and radio frequency local area networks [RF LANs]) fills = up the limited RF spectrum, near-field communication offers an alternativ= e to congesting the airwaves with data.

Previous work on electric field sensing

The development of the Personal Area Network (PAN) grew out of a meeting = between Professor Mike Hawley's Personal Information Architecture Group a= nd Professor Neil Gershenfeld's Physics and Media Group, both at the MIT = Media Laboratory. Professor Hawley's group needed a means to interconnect= body-borne information appliances, [1] and Professor Gershenfeld's group had been applying electric field sens= ing to position measurement. [3] Profe= ssor Gershenfeld and I realized that by modulating the electric field we = were using for position measurement, we could send data through the body.=

Alternatives to near-field communication

Infrared (IR) communication is used in television remote control units an= d is becoming a popular means to network portable computers but is not pr= actical for devices located inside wallets, purses, and pockets. Focused = IR relies on line-of-sight transmission, which is difficult to maintain w= ith body-based devices that are constantly in motion. Diffused IR relies = on a wide-angle beam of high optical power requiring hundreds of milliwat= ts. Electrostatically coupled PAN devices use the body as a "wet wire" an= d can operate on several milliwatts of power.

Far-field (radio) communication is more susceptible to eavesdropping and = interference than near-field communication due to the former's propagatio= n properties: an isotropic radio transmitter propagates energy with a sig= nal strength that decreases with distance squared, whereas near-field str= ength decreases with distance cubed. In addition, the earth shunts the el= ectric field, further attenuating the signal and making near-field commun= ication more difficult to intercept. Some of the data communicated among = PAN devices will be of a sensitive nature, for example, credit card and t= elephone numbers, client notes, diary entries, business communications, a= nd computer passwords. Radio systems can be made secure with cryptographi= c strategies, requiring additional overhead, but the best security is a m= essage that is never intercepted. Signal attenuation also reduces inter-P= AN interference, an important consideration for meeting rooms and auditor= iums filled with people wearing PANs. Frequency, time, and code division = techniques may also be used to further prevent neighboring PAN systems fr= om interfering with one another.

The fundamental difference between near-field and far-field communication= is clearly demonstrated in antenna design. Near-field electrostatic coup= ling is proportional to electrode surface area. Far-field transmission ef= ficiency is maximized by matching the impedance of the transmitter to fre= e space, typically by using a half-wavelength antenna. PAN devices 25 to = 80 millimeters long (e.g., a watch or credit card) would require a carrie= r of several gigahertz for efficient transmission. Since the energy consu= med by electronic components increases with frequency, any increase in th= e frequency of the carrier beyond that required to contain the informatio= n increases energy consumption. Near-field communication can operate at v= ery low frequencies (0.1 to 1 megahertz) that can be generated directly f= rom inexpensive microcontrollers. For example, the prototype PAN transmit= ter operates at 330 kilohertz (KHz) at 30 volts with a 10-picofarad elect= rode capacitance, consuming 1.5 milliwatts discharging the electrode capa= citance. A majority of this energy is conserved (recycled) by using a res= onant inductance-capacitance (LC) tank circuit.

Far-field transmission is also subject to regulations and licensing that = vary from country to country. Near-field communication avoids these compl= ications; for example, the PAN prototype (about the size of a thick credi= t card) has a field strength of 350 picovolts per meter at 300 meters, 86= decibels (dB) below the field strength allowed by the Federal Communicat= ions Commission.

The PAN is based on the seven-layer ISO 7498 network standard [4] of the International Organization for Standar= dization. The work presented here concerns the physical layer, examining = the electrical properties of the communication channel, and the second la= yer, establishing a reliable information link. The third layer would conn= ect PAN devices to specific applications.

Basic concept of a PAN communication channel

Figure 1 shows a PAN transmitter communicating= with a PAN receiver. Both devices are battery powered, electrically isol= ated, and have a pair of electrodes. The PAN transmitter capacitively cou= ples a modulating picoamp displacement current through the human body to = the receiver. The return path is provided by the "earth ground," which in= cludes all conductors and dielectrics in the environment that are in clos= e proximity to the PAN devices. The earth ground needs to be electrically= isolated from the body to prevent shorting of the communication circuit.=

Figure 1

Lumped model of communication channel: Symmetry breaking

In Figure 2 the PAN transmitter is modeled as = an oscillator, and the receiver is modeled as a differential amplifier. T= he basic principle of a PAN communication channel is to break the impedan= ce symmetry between the transmitter electrodes and receiver electrodes. T= he transmitter's and receiver's intraelectrode impedances are ignored sin= ce the former is a load on an ideal voltage source and the latter is mode= led as an open circuit. The four remaining impedances are labeled A, B, C= , and D.

Figure 2

The circuit is rearranged to show how PAN device communication works by b= reaking the symmetry between the four electrodes. The circuit is a Wheats= tone bridge where any imbalance of the relationship A/B =3D C/D will caus= e a potential across the receiver. Since the ratios must be exactly equal= in order to null the circuit, and body-based PAN devices are constantly = in motion, there will nearly always be an electrical communication path, = as long as the receiver is sensitive enough to detect the imbalance.

PAN device electric field model

A more detailed electrical model is derived by identifying all the electr= ic field paths in the system. Electric fields exist between bodies at dif= ferent potentials. = Figure 3 illustrates an electric field model o= f a PAN transmitter T communicating with a PAN receiver R. A small portio= n of the electric field G reaches the receiver R.

Figure 3

The transmitter T electrode closest to the body tb has a lower impedance = to the body than the electrode facing toward the environment te. This ena= bles the transmitter T to impose an oscillating potential on the body, re= lative to the earth ground, causing electric fields A, B, C, D, and E.

= Similarly the impedance asymmetry of the receiver electrodes (rb and re) = to the body and environment allow the displacement current from electric = fields F and G to be detected. Since the impedance between the receiver e= lectrodes is nonzero, a small electric field H exists between them.

The electric fields model is used to produce the electric circuit shown i= n = Figure 4. Some typical component values are sh= own for watch-based PAN devices. The transmitter T capacitively couples t= o receiver R through the body (modeled as a perfect conductor). The earth= ground provides the return signal. The circuit reveals that body capacit= ance to the environment E degrades PAN communication by grounding the pot= ential that the transmitter T is trying to impose on the body. For exampl= e, in one experiment, standing barefoot reduced communication between wri= st-mounted devices by 12 dB.

Figure 4

The circuit model also suggests that feet are the best location for PAN d= evices, providing large electrodes in close proximity to the body and env= ironment, respectively. This is particularly true for the environment ele= ctrode (te or re), which is the weakest link (largest impedance) in the c= ircuit. The location also suggests a novel power source: PAN devices embe= dded in shoe inserts that extract power from walking. An adult dissipates= several hundred milliwatts while walking. A piezoceramic pile charging a= capacitor at an efficiency as low as 10 percent can provide enough power= for a PAN device.

PAN locations and applications

PAN devices can take the shape of commonly worn objects: watches, credit = cards, eyeglasses, identification badges, belts, waist packs, and shoe in= serts, as shown in = Figure 5. Head-mounted PAN devices can include= headphones, hearing aids, microphones, and head-mounted displays. Shirt = pocket PAN devices may serve as identification badges. The wristwatch is = a natural location for a display, microphone, camera, and speaker. A wais= t pouch can carry a PDA, cellular phone, keypad, or other devices that ar= e large and heavy. PAN devices incorporating sensors can provide medical = monitoring for such bodily functions as heartbeat, blood pressure, and re= spiratory rate. Pants pockets are a natural location for wallet-based PAN= devices to store information and identify the possessor. Shoe inserts ca= n be self-powered and provide a data link to remote PAN devices located i= n the environment, such as workstations and floor transponders that detec= t the location and identity of people.

Figure 5

Channel capacity

A communication network is judged primarily by channel capacity, with a t= heoretical limit defined by the Hartley-Shannon law [5] C =3D Blog(1 + S/N), where C is channel capacity in bit= s/second, B is bandwidth, S is signal, and N is noise. The -3 dB bandwidt= h of the prototype PAN receiver is 400 KHz (100 KHz to 500 KHz), resultin= g in a maximum channel capacity of 417 kilobits (Kbits) per second, assum= ing a robust signal-to-noise ratio of 10. The PAN transceiver prototype i= mplements a modest 2400 bits-per-second modem.

Development of a PAN prototype

A PAN prototype has been developed to demonstrate the digital exchange of= data through a human body using battery-powered low-cost electronic circ= uitry. The detector is a current amp (gain =3D 106) followed by an analog= bipolar chopper controlled by a digital microcontroller, as shown in Figure 6. The detector synchronously integrates t= he tiny received displacement current (e.g., 50 picoamperes, 330 KHz) int= o a voltage that can be measured by a slow, low-resolution analog-to-digi= tal converter (50 KHz, 8 bits) provided by the microcontroller. The PAN t= ransceiver uses five "off-the-shelf" components costing less than $10 in = large volumes. Ultimately the analog components and microcontroller can b= e combined into a single CMOS (complementary metal-oxide semiconductor) i= ntegrated circuit to produce a low-cost integrated PAN transceiver.

Figure 6

The transmitter is an LC tank (Q =3D 6) made from a surface-mount inducto= r and the inherent electrode capacitance. The resonant tank circuit produ= ces a clean sine wave output from a square wave input, minimizing RF harm= onics, and boosts the output voltage in proportion to the Q of the tank. = The transmit voltage can also be digitally programmed by varying the puls= e width of the driving square wave. The integrator is discharged after ev= ery message bit (integrate-and-dump filtering) to minimize intersymbol in= terference. [5]

Modulation strategies

Two modulation strategies were examined for PAN communication: on-off key= ing and direct sequence spread spectrum. On-off keying turns the carrier = on to represent a message bit one and turns the carrier off for a message= bit zero. The signal-to-noise performance is improved by increasing the = transmit voltage. Direct sequence spread spectrum modulates the carrier w= ith a pseudonoise (PN) sequence, producing a broadband transmission much = greater than the message bandwidth. Symbol-synchronous PN modulation is u= sed where a message bit one is represented by transmitting the entire PN = sequence, and a message bit zero is represented by transmitting the inver= ted PN sequence. The signal-to-noise performance increases with the lengt= h of the PN sequence.

The prototype hardware is capable of detecting either on-off keying or di= rect sequence spread spectrum, determined by microcontroller coding. For = on-off keying the bipolar chopper switches are driven at the carrier freq= uency, and the integrated result is compared to a fixed threshold to dete= rmine the value of the message bit. Quadrature detection is implemented b= y performing two sequential integrations, at 0 and 90 degrees phase, for = each message bit. For spread spectrum the switches are driven by the PN s= equence, and the integrated result, which is the correlation, is compared= to two thresholds. If the correlation is greater than a positive thresho= ld, the message bit is one. If the correlation is less than a negative th= reshold, the message bit is zero. If the correlation is between these thr= esholds (the dead zone), no message bit is received.

Once the message has been successfully received and demodulated, the micr= ocontroller transmits the message to a host computer over an optical link= (not shown), which electrically isolates the transceiver allowing evalua= tion and debugging independent of an electrical ground reference.

The business card handshake

The demonstration prototype of the PAN system, shown in Figure 7, consists of a battery-powered transmitter and recei= ver, and a host computer running a terminal program. The PAN prototypes m= easure 8 × 5 × 1 centimeters, about the size of a thick credit = card. The transmitter contains a microcontroller that continuously transm= its stored ASCII characters representing an electronic business card. The= devices are located near the feet, simulating PAN shoe inserts.

Figure 7

When the woman and man depicted in the figure are in close proximity, par= ticularly when they shake hands, an electric circuit is completed, allowi= ng picoamp signals to pass from the transmitter through her body, to his = body, to the receiver by his foot, and back through the earth ground. ASC= II characters are sent to the receiver, demodulated, and sent via serial = link to the host computer where they are displayed. Thus, when they shake= hands, the woman downloads her electronic business card to the man.

Conclusion

This research has uncovered a novel means to perform local communication = using electric fields. The trade-offs among cost, speed, size, power, and= operating range must be further studied and quantified in order to engin= eer practical PAN devices. Many of the design and engineering techniques = of radio and digital communication can be applied to PAN devices.

Telephone modems have pushed modulation and digital signal processing tec= hniques to their practical limits. The application of modem telephony tec= hniques to PAN devices may deliver channel capacities of 100 Kbits per se= cond. Data compression will also increase the effective capacity of a PAN= communication channel.

The concept of power sneakers is intriguing. Shoes are something that we = almost always have with us, are unlikely to lose, and are difficult to st= eal. Networked shoes could keep track of whom we meet, where we have been= , and what we have done. When we walk by a store, advertisements could up= load to our shoes, sticking like digital chewing gum. Homework assignment= s, grades, shopping lists, errands, and reminders could be automatically = exchanged among family members at the dinner table.

The research has explored the first two layers of a network. The upper la= yers must be designed to make a practical PAN. Applications need to be de= eply considered. Privacy issues are raised when personal data can effortl= essly be exchanged. As with other communication and data systems, there i= s a trade-off among access, convenience, and security.

The sensitivity and bit rate must be increased in order to realize a watc= h-sized PAN transceiver. The low-frequency carrier of near-field communic= ation makes direct sequence spread spectrum both practical and desirable.= Ultimately PAN devices will be judged a success when they appear as comm= on useful objects that perform magic--from picture telephone "Dick Tracy"= watches to self-powered smart sneakers that seamlessly interconnect us t= o a worldwide information and communication network.

Acknowledgments

This research was sponsored in part by Hewlett-Packard Corporation and th= e Festo Didactic Corporation and was conducted at the Physics and Media G= roup of the MIT Media Laboratory.

Cited references

General references

B. H. Brown, D. C. Barber, and A. D. Seager, "Applied Potential Tomograph= y, Possible Clinical Applications," Clinical Physics and Physiological= Measurement 6, No. 2, 109-122 (1985).

T. H. Bullock, "Electroreception," Annual Review of Neuroscience = 5, 121-170 (1982).

Electrical Impedance Tomography, J. G. Webster, Editor, Adam Hilge= r (publisher), New York (1989).

J. Fraden, AIP Handbook of Modern Sensors, American Institute of P= hysics, New York (1993), p. 421.

N. Gershenfeld, Method and Apparatus for Electromagnetic Non-Contact P= osition Measurement with Respect to One or More Axes, U.S. Patent No.= 5,247,261 (September 21, 1993).

N. Gershenfeld, T. G. Zimmerman, and D. Allport, Non-Contact System fo= r Sensing and Signaling by Externally Induced Intra-Body Currents, U.= S. Patent Application (May 8, 1995).

P. Kearey and M. Brooks, An Introduction to Geophysical Exploration, Second Edition, Blackwell Scientific Publications, Oxford, UK (1991), = p. 216.

J. P. Mills, Electromagnetic Interference, Prentice-Hall, Inc., En= glewood Cliffs, NJ (1993), p. 143.

W. I. Orr, Radio Handbook, Twenty-Third Edition, Howard W. Sams &a= mp; Company, Indianapolis, IN (1987), ch. 22, p. 6.

K. Pahlavan, T. H. Probert, and M. E. Chase, "Trends in Local Wireless Ne= tworks," IEEE Communications Magazine 33, No. 3, 88-95 (Mar= ch 1995).

Part 15 Radio Frequency Devices, Federal Communications Commission= , Washington, DC (1993).

S. Ramo, J. R. Whinnery, and T. Van Duzer, Fields and Waves in Communi= cation Electronics, Third Edition, John Wiley & Sons, Inc., New Y= ork (1994), pp. 254-256.

J. R. Smith, Towards Electric Field Imaging, M.S. thesis, MIT Medi= a Laboratory, Cambridge, MA (September 1995).

D. A. Weston, Electromagnetic Compatibility, Marcel Dekker Publish= ers, New York (1991), p. 36.

T. G. Zimmerman, Personal Area Networks (PAN): Near-Field Intra-Body C= ommunication, M.S. thesis, MIT Media Laboratory, Cambridge, MA (Septe= mber 1995).

Accepted for publication April 8, 1996.

Reprint Order No. G321-5627.