Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Workshop Summary
Session ~ Technology for Owner-Authorized Handguns Speaker Presentations Dr. Lance Davis, executive officer of the National Academy of En- gineering (NAE), opened the Workshop on Owner-Authorized Handguns with some welcoming comments. He explained that NAE's mission is to promote the technological welfare of the nation by marshalling the talents of eminent members of the engineering profession to study issues at the intersection of technology and society, such as the focus of this workshop, owner-authorized handguns. The people attending the workshop, who are involved in the social, legal, or political aspects of gun safety, are well aware that a so-called smart handgun could have an impact on crime and public health and that several technology-related ques- tions are central to the issue. How mature is the technology for owner- authorized weapons? How reliable will these guns be? How lone would it take to put them into production? What would it cost? Dr. Davis noted that the complex issues surrounding smart handguns could not be addressed in a single day and the workshop would probably not provide definitive answers to those questions. The workshop would help NAE frame a future study of the state of research and technology for smart guns. FIRST KEYNOTE SPEAKER The first of the two keynote speakers, Don Sebastian of the New Jersey Institute of Technology (NJIT), has been involved with the issue of 9
10 OWNER-AUTHORIZED HANDGUNS smart-gun technology for about three years, since the New lersey Senate began considering legislation to mandate the development of smart-gun technology. (The legislation was enacted in December 2002.) In 2001, WIT received a $1 million appropriation from the New lersey legislature to study smart-gun technology in 1999. The legislature appropriated an additional $500,000 in both 2000 and 2002 to supplement the first award. The legislature mandated that the study focus on smart guns that could address the issue of child safety in the home (legislators used the terms "child-safe" guns, "owner-only" guns, and "smart" guns interchangeably). The legislators wanted to address three concerns: accidental shootings, de- liberate crimes by children, and teen suicides. Dr. Sebastian said that WIT willingly pursued the study under those terms not for the money but because the institution has always been willing to tackle difficult public- policy questions. Such work, he said, is part of the institution's responsibil- ity as a technological research university. With the appropriation, the WIT study focused on three critical ques- tions: Does smart-gun technology exist? If it doesn't but is possible, what would be needed to bring it to the commercial marketplace? Finally, what can NJIT do to bring smart guns closer to market? One of the first discoveries the WIT researchers made, Dr. Sebastian said, was that "smart gun" is a term of art. People talked about smart guns as if one existed; there was not an engineering definition of it much less an actual product. The researchers also observed that most people believe that the technologies to create a smart gun already exist. It was, people thought, just a matter of desire for all of the pieces to be put together. The team also learned that a single smart-gun design was not likely to solve all handgun safety problems. As background for the WIT study, the team looked into a study (Weiss, 1996) conducted by Sandia National Laboratories in 1995 with funding from the National Institute of Justice (NIl). That study had focused on the problem of law-enforcement "take-away" when a police officer's gun is taken away during a crime. The New Jersey legislature, in contrast, was concerned with guns owned by private individuals. Whereas police weapons must be protected from an array of potential abusers, household weapons must be protected most of all from family members. The WIT researchers concluded that a typical gun owner could not be depended on to adhere to rigorous standards of gun safety. Owners often do not use locking devices or maintain the integrity of the ones they do use. As an example, Dr. Sebastian cited some extant devices that depend on a personal
SESSION 1: TECHNOLOGYFOR OWNER-AUTHORIZED HANDGUNS 11 identification number. The research team found that the codes became common knowledge in a family, and that most codes could be easily discovered by hackers. In short, private gun owners need a technological solution that performs well despite neglect and abuse. Police officers need a technology that will guard against illegal transfer and perform well while . . . . t. le owner maintains strict, rigorous proce( lures. Gun hobbyists and sports enthusiasts have entirely different needs, according to Dr. Sebastian. The main concern of these gun users is very likely accidental discharge. Some firms are already working on smart-gun technologies to deal with that issue. The WIT researchers developed a list of requirements for home-owner handguns. When not in use, the gun would have to be in a protected mode yet be ready to fire. The gun would have to be able to identify its rightful owner without time-consuming, cross-modal actions. People often assume that a single technology will solve all of these problems. Popular expectations tend to focus on technology based on fin- gerprints or voice recognition. Dr. Sebastian emphasized, however, that solving all of these problems will require a systems approach. A gun that turns on when it recognizes a legitimate user will have to be based on a system that integrates several technologies. During the study, WIT worked with the New lersey-based Joint Ser- vices of the Small Arms Program aSSAP), which designs sidearms for the armed services. ISSAP tested 18 commercially available lock-on, bolt-on handgun safety devices and concluded that all of the products manufac- tured for home use could be compromised with relative ease. In fact, they concluded, none was any better than a traditional lock box. NIT is also funding research on electronic analogues of a lock and key. Designed for law enforcement, most of these technologies require a device worn on the person, such as a ring or watch, that communicates with the gun by radio frequency (RF) or ultrasonics. In a home environment, how- ever, the "key" component of these systems is likely to be stored near the gun or in an easily discovered hiding place. And so these technologies, although important from a research standpoint, are not well suited to home use. To create a true smart gun, according to Dr. Sebastian, it will be necessary to develop a system that uses an attribute of the user or owner- a biometric as a nontransferable token, a key that cannot be counter- feited or shared. Moreover, of the technologies that might be used in smart guns, biometrics seems to be the most mature. Biometrics are being
12 OWNER-AUTHORIZED HANDGUNS investigated for an array of other uses and, therefore, already have a fairly diverse market demand. WIT researchers studied commercially available fingerprint identifica- tion devices in the laboratory. Although these devices have not been embedded in guns, they are already in common use, typically to provide computer-related security. The team studied four fingerprint-recognition technologies one that used optical scanning, two that used capacitative measurements, and one that used infrared. The hardware was tested with commercially available software and with a more sophisticated pattern- recognition software developed by BES, a New Jersey company. They also tested BioMouse, an accessory for a computer mouse used in the log-in process in e-commerce to verify purchaser identity. But Biomouse, the size of a traditional optical scanner, cannot be miniaturized to fit inside a gun. The mechanical failure rate of traditional handguns is around 1 in 20,000. With fingerprint devices failure rates were at best 1 in 100; at times, the failure rate was as high as 1 in 4. The best result with these devices was a 99 percent recognition rate, and that rate could only be achieved under pristine laboratory conditions, with the sensor stringently cleaned between every use. And that rate is about two orders of magnitude higher than the mechanical failure rate. Trying to capture a fingerprint with a fixed sensor, the researchers concluded, would be difficult in real- world situations. In addition, there was no apparent technological growth path for any of the sensor technologies that would bring the failure rate down to the ideal rate. This does not mean that fingerprint technology should be discarded, but it does suggest fingerprint technology might be more effective if it were integrated into a multisensory approach to user . . ,~ . 1( ~entl~lcatlon. The WIT team next looked into using the way each person grabs a gun handle to differentiate among gun users. The researchers captured hand placements and sizes from several hundred people to see if they could be used to identify individuals. Hand size was fairly reliable they found, although reproducibility was better for practiced handgun users, such as law-enforcement professionals, than for non-users. When the researchers added sensors to the grip that measured grip pressure during the period of trigger squeeze, the trace of the grip over time could be used to differentiate among hundreds of different users. Satisfactory results were achieved with as few as nine pressure sensors. The NJIT laboratory is now experimenting with micropatterning inside the hand grip of the guns used by the New lersey State Police.
SESSION 1: TECHNOLOGYFOR OWNER-AUTHORIZED HANDGUNS 13 Dr. Sebastian said he foresees a system approach based on multiple modalities. If one technology has a 1 in 100 failure rate, and another has its own 1 in 100 failure rate, when they are used together, they would have a 1 in 10,000 failure rate. That rate would approach the rate of mechanical failure. Dr. Sebastian noted that many people think gun manufacturers should be able to develop a reliable smart gun using their own R&D capacity. He estimated that total annual revenues for the gun industry are about $500 million and a reasonable investment in research of 1 percent or $5 mil- lion per year would not be enough to bring a smart handgun to the mar- ketplace. In fact, he said, neither public entities nor private firms will be able to create smart guns on their own. Gun manufacturers do not have the necessary expertise in software development and microsensor technologies. Public entities simply do not have enough funding to carry out the research on their own. Some sort of public-private partnership will be necessary. WIT is now working on a small project with the Picatinny Arsenal to evaluate hand-grip technology in live-fire situations using advanced simu- lation environments. WIT has also submitted a proposal to NIT to create grip-dynamic technology and validate it in live-fire situations. But this is an enormous undertaking, and as the technology gets closer to real-world test- ing, the costs of development will increase exponentially. Training a weapon to recognize its authorized user is a real concern. At present, NJIT has developed a grip template that uses 30 "trigger" points to recognize the user 10 points as a baseline and the remaining 20 to estab- lish its robustness. One important question that has arisen is whether a user's profile changes in stressful situations. Handguns in private hands present different problems. If a gun comes with a simple home-use kit if the process is very open children might be able to reprogram the gun themselves. Would that invalidate the authorization? Different states already have different policies regarding handgun safety, and states may also demand that they be allowed to train and validate weapons in their own ways. Dr. Sebastian said there is an acute need for solid, platform- independent, technology-independent standards for legislation regarding smart guns. The standards must be appropriate for the application environ- ment and must specify the levels of recognition and reliability a smart gun would be expected to provide. Only such neutral, objective rules would provide technology developers with standards for judging the suitability of a particular technology or system. Eventually nontechnology issues, such as product liability, will also have to be addressed. Dr. Sebastian emphasized that even though gun
14 OWNER-AUTHORIZED HANDGUNS manufacturers are working on the problem of smart guns, unless the matter of liability is resolved, smart guns will never reach the marketplace. An- other important issue is the question of proprietary rights. If smart guns are developed through public-private partnerships, who will produce the new weapons? Who will have the right to distribute them and to train people to use them? When smart guns are ready for sale, who will administer and authenticate the technology? Moreover, the entire issue of deployment and enforcement could stray into the sensitive area of gun control. These problems will have to be ad- dressed early on, before they become explosive and expensive to address. SECOND KEYNOTE SPEAKER John Wirshinski of Sandia National Laboratories, the next keynote speaker, talked about a very different user group for smart guns: law- enforcement personnel. In 1994, he said, NIT asked Sandia to take a sys- tems engineering look at the problem of law-enforcement officers being killed in the line of duty when their own weapons were used against them. Sandia's 1996 report was based on that review. More recently, NI! asked Sandia to take another look at the gun take-away issue to determine how the technology has evolved (Wirshinski, 20011. Both reports were driven by an assessment of user requirements and focused on systems engineering. The first report used a key analogy: a key, a lock, and a discriminator. In the second report, keys and discriminators were examined from the perspective r ot access control. The term smart gun has become such a catchall phrase that the Sandia researchers felt it essential to establish categories to distinguish the types of smart-gun technology. Just as there are several ways to get into a building, there are several possible ways to limit access to a gun. The first way is with mechanical safeties. These are not really smart-gun technologies because they cannot keep a gun from being used by anyone who knows how to operate the mechanical device. However, Mr. Wirsbinski said that he had heard of at least two cases in which a mechanical device on a police officer's gun had delayed an adversary from firing long enough for the officer to regain control of the weapon. Some types of mechanical devices are not very helpful, however. Trigger safeties, for instance, cannot prevent the de- liberate firing of the weapon by an unauthorized user with the safety off, the gun will discharge no matter who pulls the trigger. Mechanical thumb safeties, however, could delay adversaries for a brief period.
SESSION 1: TECHNOLOGYFOR OWNER-AUTHORIZED HANDGUNS 15 Another type of access device is a gun lock. Gun locks are commer- cially available now, both as retrofits and integrated in new firearms. Many are token based, requiring a key. One problem is that the keys tend to be universal, so anybody could have one. But the mere act of locking a gun does provide some security, although it probably would not be very useful in a law-enforcement scenario. It would be hard to imagine a police officer carrying around a locked weapon, and having to get out a key to use it. Police officers could use gun locks when it was necessary to relinquish a gun, if it had to be stored for a period of time, for instance. Most current technology development is focused on self-locking weap- ons that revert to a secured state when they are released from a firing grip. Right now there is a great deal of discussion about technologies to realize such weapons, such as magnetic rings and biometric approaches, including grip scans and fingerprint readers. One company is experimenting with thermal scans of subcutaneous structures. In a law-enforcement scenario, the latter approach could pose problems. For instance, wearing gloves might affect the scan. Another company is producing a fingerprint reader that uses sound waves that can actually penetrate gloves, which might resolve some of the problems associated with fingerprint scans. The ideal solution for law enforcement is a truly personalized weapon, which will almost certainly be an evolution of a self-locking weapon. But a self-locking weapon, by definition, can only offer a partial solution because it cannot satisfy all of the requirements established in Sandia's 1996 report. But no self-locking weapon is available now; and none will be available in the near future. Producing such a weapon will require systems integration, which is always a troublesome engineering problem. In addition, miniatur- ization issues will have to be addressed. Most likely, the solution will emerge from one of two technologies: biometrics or a system using an RF transmitter. One promising type of RF transmitter would be a weapon activated by a chip embedded in the user's hand. Mr. Wirsbinski told the audience that when he first heard this tech- nology proposed as an answer for a self-locking weapon, he did not take the idea seriously. But such chips are already being used in other fields. They are surprisingly popular in health care, for example, where they are used to embed vital medical information that patients always carry with them. Embedded chips have also been used for quite some time to identify pure- bred animals. Using this technology for a user-authorized weapon would require a small coil inside the gun to generate an induction field with enough power to retrieve information from a chip embedded in the
16 OWNER-AUTHORIZED HANDGUNS authorized user's hand. The embedded chip concept would resolve a number of problems with locking technology, such as the possibility of losing a magnetic ring. Mr. Wirshinski said he has been approached by a gun manufacturer in a country whose military was interested in developing such a weapon for its special forces. In the United States, Mr. Wirshinski said, developing such a technology for use by police would introduce troublesome privacy issues. Before biometric user-authorized weapons become common, Mr. Wirshinski said, many hurdles will have to be overcome, including user acceptance. A few years back, he said, he had worked at an Air Force facility that had just put in place a huge new access-control system that included retinal scans, card swipes, and pin codes. Soon after the system was imple- mented, a general came through and flatly refused to submit to the retinal scan. He said, "You are not scanning my eyeball with a laser; I don't trust it." The retinal scan was turned off. Another serious issue in the law- enforcement community is reliability. It will be essential that any new tech- nology be proved as reliable in everyday use as the technolo~v the police have now. -A ~ r Another very complex issue is training handguns to recognize their authorized users. Access control is based on the idea that there is a central location where the access-control database is stored. It is analogous to using a badge reader to get into a building. When you swipe the card, data on the card are compared to data in the computer database, a relatively simple matter. But with a biometric weapon, things would be more complicated. Ideally, the weapon itselfwould carry the database. When the weapon is in enrollment mode, it would build a template of the authorized user. Then, when it is in use, it could recognize the user, either by comparing his or her template to all of the templates in the database and looking for a match or by directing the system to compare the biometric reading to the specific template of the user to confirm his or her identity. But for a handgun to carry the entire database of stored templates, enormous technical challenges would have to be overcome. Another potential problem in training a biometric weapon is that a user who is properly enrolled onto a biometric weapon may not be recog- nized because the template can be thrown off by user stress. When the body goes into fight-or-flight mode, many things change, such as blood flow, muscle tension, and even physical structures. At these times, things that were learned in an unstressed situation become inaccessible. How can bio- metric gun technology cope with this reality?
SESSION 1: TECHNOLOGY FOR OWNER-AUTHORIZED HANDGUNS 17 Another problem with biometric weapons is false rejections. Over the years, researchers have learned that for the first two to Six weeks after bio- metric and access control devices are deployed, they have a higher false rejection rate. Once the user is well trained, the devices work far more reliably. Whether this will happen with firearms is an open question. In addition, when a biometric device is not used for an extended period of time six months, perhaps the false rejection rate recurs. The biometrics community is not sure whether this is a function of an aging template- that a person's biometrics change over time or whether it is a function of user training. In either case, this could be a serious problem. Many officers leave their guns holstered for extended periods of time, and many never fire their weapons at all in the line of duty. The two biggest changes in smart-gun technology since the Sandia group's 1996 report have been evolutionary rather than revolutionary. First, there is now a commercialized electronic primer technology, which is used only in long arms. Second, some manufacturers are looking into totally electronic weapons. In fact, a few prototype "proofs of concept" have been created. In these guns, the bullets are stacked in the barrel, and the priming compound is stacked in the barrel; there are no mechanical operations. No self-contained cartridge is chambered and ejected. Everything is electronic. But, this also raises some serious issues. Mr. Wirshinski asked the audience to think of how personal computers work and decide if they would be willing to bet their lives on the reliability of electronics. In addition, electronic guns would require the development of a totally new manual r ot arms. These technologies may one day lead to the development of a smart gun. Many of them are promising and some of them suggest answers that might work in certain situations. But, so far, none of them meets the needs of law enforcement.
Panel Presentations _ he panel session opened with remarks by Ken Green, who repre- sented the National Shooting and Sports Foundation and the _ ~ Sporting Arms and Ammunition Manufacturers Institute (SAAMI). Established in the early 1900s, SAAMI's purpose is to set stan- dards for the manufacture of firearms and ammunition. Currently, SAAMI has 25 member organizations, but the standards apply to all commercial manufacturers of firearms and ammunition. SAAMI has set some 700 standards, covering the dimensions of cartridges and chambers, the velocity of projectiles, and chamber pressure for rimPire pistol, revolver, shotgun, and rifle cartridges. There is also a standard for the abuse and mishandling of weapons. Many of the standards are listed with the American National Standards Institute (ANSI), and all are volun- tary, although most manufacturers do follow them. ANSI standards docu- ments may be seen on its website, ansi.org. When the technologies for user- authorized guns become available, standards will be set for them as well. Kevin Foley, a representative of Smith & Wesson, said that his com- pany has been working in the area of user-authorized weapons for some time. The company acted in an advisory capacity during the 1996 study by Sandia National Laboratory (Weiss, 19961. Once the study was completed, he said, Smith & Wesson became more active in researching the technol- ogy. All of Smith & Wesson's revolvers are now sold with integrated me- chanical locking systems. By next year, the locking systems will also be on 18
SESSION 1: PANEL PRESENTATIONS 19 all of its pistols, thus meeting the legal requirements for several states. The company has provided external gun locks since 1997. Smith & Wesson has experimented with placing electronics and other devices inside a conventionally produced handgun. The handgun would then fire only when a user wore a wristwatch-type transponder. One prob- lem with this approach involves the mechanics of a gun. The inside of a conventional handgun is entirely mechanical, with levers, springs, and pins. To prevent the trigger from firing, the linkage between the trigger and the firing pin must be disconnected or blocked. This might be done by adding a separate, microelectromechanical system (MEMS). But the Smith & Wesson researchers found it would be impossible to make the MEMS reli- able. Because handguns are designed to be field stripped, disassembled, and cleaned, a MEMS device could be easily disabled. Besides, it would be very easy to bypass a sophisticated electronic security system inside a handgun. Transponders, which are essentially electronic keys, also turned out to be unsatisfactory. When Smith & Wesson researchers compared transpon- ders with the key locks the weapons industry had been providing for some time, they concluded that transponders offered no additional benefit. In the officer take-away situation, nothing was gained. Based on these experiences, Mr. Foley said, the research team set some design goals. First, the reliability and durability of the weapon must not be compromised by a security system. Second, if the security system is removed, the gun must stop functioning and become useless. Third, there should be no keys. An authorized-user-only weapon should function, or not, based on the user's identity, not on what the user brings to the gun, he said. The Smith & Wesson team concluded that the design goals could only be met by eliminating the mechanical firing mechanism and designing the weapon so the electronic security system could be integrated into the firing mechanism. The user would be identified by biometric information, and identification could be instantaneous. If the authorized user picks up the gun, it will function, but if anyone else picks it up, it will not. At this point, the research team examined Remington's EtronX primer, which is used in rifles that fire cartridges electronically. A rifle stock has a lot more room than the handgun grip. Beginning in 1997, Smith & Wesson has worked to miniaturize the technology for electronic firing. At this point, company researchers have fired about 50,000 rounds of ammunition and are building 50 handguns, a number of which have fired more than 5,000 rounds. But the space is very cramped inside those guns, and they are not
20 OWNER-AUTHORIZED HANDGUNS perfect. Mr. Foley said that attaining a durable and reliable system is be- coming increasingly feasible. Working with funding from the National Institute of Justice (NIT), Smith & Wesson's next step will be to miniaturize the firing electronics to make room for biometrics. The company is working with a biometrics company in New Mexico to develop a small light emitting diode (LED) sensor that conforms to the grip but that does not rely on fingerprint im- age, so that grip placement does not have to be precise. The electronics are small and fast, and the performance of the gun has the potential to be as good as or better than anything else the researchers have seen. The team is now past the proof-of-concept stage. In March 2002, the company built its first solid-state system, which is now being tested. Mr. Foley said Smith & Wesson researchers are about two years away from a handgun with integrated biometrics that can be tested in the field. The next speaker, Peter Sebelius of the Charles Stark Draper Laboratory, noted that since September 11, 2001, research on biometric identification has increased dramatically. Identification by biometrics, he said, is an interesting process, one that has tended to focus on the automation of mechanical systems already in use. Mr. Sebelius argued that biometric identification should take a more expansive approach. lust as people identify each other by drawing on several senses, similarly, biometric identification should use multiple sensors and an automated judgment, or weighting, scheme. Examples of judgment schemes are already in use. Fault-tolerant sys- tems, for example, draw input from multiple sensors, sometimes sensing the same things and sometimes sensing different things. This kind of sys- tem is used in the SSN-21 submarine, where four flight-control-system channels measure the vessel's position and movement, then instantaneously "vote" whether the computer has come to the right conclusion as a result of the sensor input. If one channel or component fails, the other channels recognize that, shut the system off, and then continue to operate in a re- duced mode, while issuing warnings about the failure. Training is vital to using the system, and it will be vital that handgun users be trained to use biometric weapons. Although, as one participant in the conference pointed out, federal law does not set requirements for handgun ownership, many states require that owners be trained to use handguns. At that time, owners could also be trained to use biometric iden- .^ · r tlilcatlon features.
SESSION 1: PANEL PRESENTATIONS 21 It is also important to remember that mechanical safety devices are already available commercially but are often not used. Often in the gun- related tragedies reported in the media, for instance when a child takes his father's gun and shoots a classmate, all of the extant mechanical locks were available to the father. A trigger lock could have been put on the gun or the gun could have been put in a cabinet and the key kept by the father. Mr. Sebelius recalled one workshop participant pointing out that a national survey of private firearms ownership indicated that of households with handguns only 43 percent kept them locked and unloaded (Cook and Ludwig, 19961. Even fewer households kept guns unloaded. Another issue that arises in this context is the reliability of an elec- tronic device. Mr. Sebelius argued that we already trust many electronic devices. We trust electronic watches to tell time. When we fly, we trust computers with our lives. A lot of work is being done to create trustworthy electronic weapons. Nacem Zafar of Viridicom, a spinoff of Bell Labs located in Silicon Valley, noted that his company is working on fingerprint-sensor technology and fingerprint-sensing algorithms. The company has commercialized a silicon sensor and holds some of the original patents on the technology. Three patent applications related to these technologies are in different stages of the patenting process. He said Veridicom's sensor is the most widely used fingerprint sensor in the world. The biggest market for biometrics, he said, is in South Africa, where government payments are made on the basis of fingerprint identification. The same technology is becoming popular in China, where the govern- ment has issued national identity cards to every citizen in the form of smart cards carrying two fingerprint sensors. The technology is also being adopted in Malaysia, Hong Kong, and Italy, he said. Eventually, it will be possible to put biometrics into the gun itself, but many questions will have to be answered first. Biometrics-based handguns will not be accepted unless they are safe and simple to operate. It is fine to investigate exotic technologies, but, ultimately, the police want a gun that is deployable and does not hinder critical actions. Sometimes, he said, po- lice must just pick up a gun and shoot, with no time to think about placing their fingers just so. At Veridicom, researchers are approaching the take-away problem not as a matter of securing a gun, but as a matter of securing gun access. One solution Veridicom is now pursuing is a secure holster. The holster looks
22 OWNER-AUTHORIZED HANDGUNS like a normal holster ancl can carry a normal gun, but it has a fingerprint sensor inside that unlocks the gun once it confirms the print. Recognition takes only a few milliseconcls. The holster, he saicl, is convenient, simple to program ancl use, ancl keeps a record of the last 500 people who tried to access the gun. Right now, as a specialty item, the holster costs around $100. When it enters mass procluction, the cost will drop to about $20. Mr. Zafar's associate, Andrew Eros, president of AcciMetrix, the com- pany working with Vericlicom to assemble the holster technology, said that if the crevice does not recognize the user, the gun remains lockocl in the holster. The holster is powered by a 9-volt battery ancl has a shelf life of one year. When the battery fails, the holster releases the gun. If a catastrophic failure should occur, for instance if the gun is dropped ancl accidentally clischarges, the holster keeps the gun lockocl in place. Mr. Eros said that the holster, which he described as an intermediate solution to the problem of gun take-aways, will enter production when testing is complete, perhaps by the end of 2002. In reply to a query about whether fingerprints are unique enough iclen- tifiers, Mr. Zafar said fingerprints are more individual than many other identifiers now in common use, such as PIN numbers ancl passworcls. Vericlicom has been working on ways to circumvent certain finger- print-sensor problems. For instance, currently the technology will not work when a user is wearing gloves or when fingerprints are coated with various solvents ancl household proclucts. The company has recently introduced software that can screen out "olcl" (latent) fingerprint impressions on the sensor in order to read only the current print. In reply to a question about fingerprint registration as being tanta- mount to handgun registration, Mr. Zafar said that the fingerprints repre- sent the gun owner's relationship to the gun. They are not currently regis- terecl with the government ancl do not pass into a centralized database but remain in localized databases on the weapons themselves. In the last three years, the reliability of fingerprint-iclentiflcation tech- nology has improved enormously. Vericlicom's testing with the 1.2 million registered users in the South Africa database has reduced the failure rate to 1 in 100,000. Fingerprint-recognition technology is being cleployocl more widely than . . .. . ~ . . . ,% . . In lust weapons applications. nor Instance, It IS rlnc sing many uses In secur- ing medical recorcls. Ancl as part of an access-control crevice, Vericlicom's chip is built into PCs made by NEC, Fujitsu, Acer, Panasonic, ancl Hitachi; its competitor's chip is built into Samsung ancl Micron equipment. About
SESSION 1: PANEL PRESENTATIONS 23 45,000 PCs a month are sold with the Veridicom chip built in. Mr. Zafar said he believes everyone will eventually have some kind of small biometric device, on a keychain, perhaps, or an electronic ring, for everyday con- sumer transactions to pay for groceries, open a car door, log onto a PC, or protect valuable digital content, such as music. Wently Howe, a program manager in the NI! Office of Science and Technology, defined the purpose of NIT, the research arm of the U.S. Department of Justice, as improving the criminal justice system, either through the identification and modeling of programs in the social sciences, or through improvements in science and technology. The Office of Science and Technology began supporting research on smart guns in 1994, prompted by the gun take-away problem in law enforcement. Police officers have very specific views about what they need in a smart gun. They do not want weapons that require wearing a secondary unit, such as rings on both hands or watches on both wrists or badges or pagers that carry the technology. The recognition technology and power source needed to be transparent and seamless. They don't want to go through extensive training, and they want new weapons to resemble the weapons they already know. The 1996 Sandia report found that one of the most critical feature to law-enforcement officers was that, if the power source in the smart weapon failed, the weapon would revert to a normal sidearm; in other words, it must "fail live." Thus, the officer could still use the weapon, but in the event of a take-away, so could the assailant. Another important concern is related to multiple authorized users. Most law-enforcement agencies in the United States have fewer than 25 officers. Sometimes, an entire department must have access to one weapon; in certain situations, partners must share a single firearm. Particu- lar units, such as SWAT teams, often share weapons. Everyone wanted the weapon to be cost effective, preferably a technol- ogy that could be retrofitted to existing guns so that police departments would not have to strain their limited budgets to buy new weapons. In addition to the report, Sandia produced several proof-of-principle devices actually boxes containing air pistols and electronics and a power source. Ms. Howe's team then attended conferences of major law- enforcement organizations, such as the International Association of Chiefs of Police, the National Sheriffs Association, and the Fraternal Order of Police, as well as technology fairs in local communities. At each venue, they
24 OWNER-AUTHORIZED HANDGUNS demonstrated the devices and described the research process, addressing technology issues, such as voice recognition and magnetic rings. The boxes were a difficult sell because they were rudimentary and could not demonstrate the full potential of the recognition technology. But these demonstrations did provide opportunities for law-enforcement personnel to discuss their requirements for such a firearm. For instance, some officers pointed out a serious flaw for police in voice-recognition technology an officer could not "talk to" his or her gun without revealing his or her the location to the suspect. The higher up the chain of command, the greater the interest in the proof-of-principle devices. Command staffwas far more interested in the liability issue associated with smart guns than line officers, who typically saw only a cumbersome box containing an air pistol. When the NI! team exhibited a 40-caliber weapon (developed by Colt) that au- thorizes the user through a radio-frequency technology contained in a wrist- watch, many officers were very receptive because the gun was similar to the ones they already use and was activated by the mere presence of a watch. In 1999, in response to a public solicitation for proposals, NIJ awarded nominal grants ($300,000 each) to Smith & Wesson and to FN Manufac- turing for smart-gun technology development. In 2000, after Congress ap- propriated $8 million in new money specifically for smart-gun research, NIT provided significant funding to Smith & Wesson for its electronically fired, biometric recognition firearm, and FN Manufacturing's ultrasonic with embedded microelectrnics identification weapon. NIT also funded Sandia to update the 1996 study, and it released a directed solicitation for the development of smart-gun technology. NIT's goal was not to have one universal smart-gun technology for all firearms but to provide seed money to bring several promising technologies to fruition. Of the 12 proposals that were submitted for independent peer review, Ms. Howe said four were funded. The proposals used various ap- proaches, including additional biometric recognition systems, chemical compounds and radio-frequency technologies. Two of the four projects- one using ultrasonic technology together with a transponder, and one using biometric identification in an electronically fired weapon are moving for- ward very well and have gone through independent peer reviews with law- enforcement agencies to determine the functionality of the firearms in con- trolled operational settings. The grantees include VLe Small Arms, Expo- nent Inc., Technology Next, and Mosermation. Ms. Howe said that it will be at least five years before there is a weapon capable of undergoing rigorous laboratory and field testing. A smart gun for law enforcement will not be ready for many years.
