IoT security tips: 7 biometric security technologies

    The so-called biometric technology is to identify personal identity by closely combining computers with high-tech means such as optics, acoustics, biosensors and biostatistics principles, using the inherent physiological characteristics of the human body (such as fingerprints, facial images, irises, etc.) and behavioral characteristics (such as handwriting, voice, gait, etc.).

　　Biometric technology uses human characteristics such as fingerprints, voice, etc. to identify individuals. Currently, there are many biometric technologies available for identity authentication. Here, we describe how most popular biometric technologies work and briefly comment on their functions of capturing, extracting features, comparing and matching.

    1. Iris recognition technology

　　The iris is a colorful, fabric-like ring inside the pupil of the eye. Each iris contains a unique structure based on features such as the crown, lens, filaments, spots, structures, pits, rays, wrinkles and striations. It is claimed that no two irises are alike. Iris scanning security systems involve an automatic camera that searches for your eyes and begins focusing when it finds your iris. Trying to trick the system by blinking is not possible.

　　advantage

　　Easy to use for users; may be the most reliable biometric, although it has not been tested; only requires the user to be in front of the device without physical contact.

　　shortcoming

　　One of the most important disadvantages is that it has not been tested. The current iris recognition system has only been tested on a small scale using statistical principles, but has not been tested for uniqueness authentication in the real world. It is difficult to miniaturize the image acquisition device. An expensive camera is required due to the need for focusing. The minimum bid for such a camera is $4,000. The lens may distort the image and greatly reduce reliability. It is extremely difficult to read with dark eyes. A better light source is required.

　　2. Retinal recognition technology

　　The retina is also a feature used for biometric identification. Some people believe that the retina is a more unique biometric feature than the iris. Retinal recognition technology requires a laser to illuminate the back of the eyeball to obtain the uniqueness of the retinal feature.

　　advantage

　　The retina is an extremely stable biometric feature because it is "hidden" and therefore cannot wear out, age or be affected by disease; the user does not need to have direct contact with the device;

　　It is the most difficult system to deceive because the retina is invisible and therefore cannot be faked.

　　shortcoming

　　Retinal technology has not been tested. It is clear that retinal technology may cause health damage to users, which requires further research; retinal technology is not attractive to consumers; and it is difficult to further reduce its cost.

     3. Facial Recognition

　　Facial recognition technology recognizes facial features and the relationships between them. Recognition technology is very complex when based on these unique features. This requires artificial intelligence and machine knowledge learning systems. The two technologies used to capture facial images are standard video and thermal imaging. Standard video technology uses a standard camera to capture an image or a series of images of the face. After the face is captured, some core points are recorded, such as

　　The positions of the eyes, nose and mouth and their relative positions are recorded and then form a template; thermal imaging technology generates facial images by analyzing the heat lines generated by the blood in the capillaries of the face. Unlike video cameras, thermal imaging technology does not require good lighting conditions, so it can be used even in dark situations. An algorithm and a neural network system plus a conversion mechanism can turn a fingerprint image into a digital signal, and finally generate a match or non-match signal.

　　advantage

　　Facial recognition is contactless, and users do not need to have direct contact with the device; although desktop video cameras can be used, only more advanced cameras can effectively capture facial images at high speed;

　　shortcoming

　　The position of the user's face and the surrounding light environment may affect the accuracy of the system; most people who study biometrics agree that facial recognition is the least accurate and most easily deceived; improvements in facial recognition technology rely on improvements in feature extraction and comparison technology, and the equipment used to capture images is much more expensive than the technology; changes in the human face due to hair, accessories, aging and other factors may need to be compensated through artificial intelligence, and the machine learning function must constantly compare previously obtained images with current ones to improve core data and make up for minor differences; it is difficult to further reduce its cost, and we must sell high-quality equipment at a high cost.

　　4. Signature recognition

　　Signatures have been used as a means of identity verification for hundreds of years, and we are all familiar with signing our names on bank forms as a mark of our identity. Digitizing a signature is a process that measures the image itself as well as the motion of the entire signature - the speed, sequence and pressure of each letter and between letters. Signature recognition, like voice recognition, is a form of behavioral measurement.

　　advantage

　　Using signature recognition is more easily accepted by the public and is a recognized identity recognition technology.

　　shortcoming

　　Signatures change with experience, temperament and lifestyle; we have to compromise on security to deal with the inevitable changes in signatures; we cannot use it on the Internet because it is slow; the tablet used for signatures is complex and expensive, and it is technically difficult to combine the two because of the huge difference in resolution between the tablet and the touchpad on a laptop; and it is difficult to miniaturize it.

    5. Voice recognition

　　Voice recognition technology is similar to signature recognition. Voice recognition is also a behavior recognition technology. Voice recognition equipment continuously measures and records the waveform and changes of voice. Voice recognition is based on accurately matching the voice collected on site with the registered voice template.

　　advantage

　　Voice recognition is also a contactless recognition technology that users can accept naturally.

　　shortcoming

　　Like other behavioral recognition technologies, sound is difficult to match accurately because of its wide range of variation. Sound will affect the results of collection and comparison as the volume, speed and sound quality change (for example, when you have a cold). With the development of technology, you may be able to detect and reject recorded sounds. However, at present, it is still easy to deceive the sound recognition system with sounds recorded on tapes. High-fidelity microphones are very expensive. Fingerprint recognition system Fingerprint recognition as a recognition technology has a long history and has a solid market backing. According to the general public, fingerprint recognition technology analyzes the global features of fingerprints and the local features of fingerprints, feature points such as ridges, valleys and end points, bifurcation points or divergence points. The feature values ​​extracted from fingerprints can be very detailed so that a person's identity can be reliably confirmed through fingerprints.

    On average, each fingerprint has several unique measurable feature points, each feature point has about seven features, and our ten fingers produce at least 4900 independently measurable features - enough to confirm whether fingerprint recognition is a more reliable identification method. Here, we roughly give the process of acquiring images, extracting features and comparing in fingerprint recognition.

　　6. Optical technology and capacitive technology in image acquisition

　　The two main technologies used to collect fingerprint images are optical technology and capacitive technology. Optical technology requires a light source to be reflected from a prism to a finger pressed on a camera, and the light illuminates the fingerprint to collect the fingerprint.

　　Using semiconductor technology using capacitance technology, the ridges and valleys of the finger pressed against the collection head produce different capacitances between the finger epidermis and the chip. The chip obtains a complete fingerprint by measuring the different capacitance fields in the space.

　　shortcoming

　　Because the chip of capacitance technology is expensive, and the chip is as big as the finger, it is expensive, so several companies try to launch a chip that can be smaller than the fingerprint and only collect part of the fingerprint for verification. With this collection method, the user must place the finger accurately to ensure that it can be read correctly. This will inevitably make the reading head difficult to use. Another disadvantage of using this small chip is that only using part of the fingerprint is bound to be less reliable than collecting the whole fingerprint for comparison.

　　Another disadvantage of the capacitive acquisition head is that it is easily affected by interference, ranging from interference from 60HZ cables to interference from user contact and electrical interference inside the fingerprint collector.

　　The final problem with capacitive sensors is reliability. Static electricity, salt in sweat or other dirt, and finger wear can make it difficult for the sensor to read fingerprints.

　　In fact, so far, optical acquisition heads provide a more reliable solution. By improving the original optical imaging technology, the new generation of optical fingerprint collectors has made the capacitive solution pale in comparison with impeccable performance and relatively low prices.

   7. Advantages of fingerprint recognition

　　Fingerprints are unique features of the human body, and their complexity is sufficient to provide sufficient features for identification; if we want to increase reliability, we only need to register more fingerprints and identify more fingers, up to ten, and each fingerprint is unique; scanning fingerprints is very fast and very convenient to use; when reading fingerprints, the user must touch the finger with the fingerprint collection head, and direct contact with the fingerprint collection head is the most reliable way to read human biometrics. This is also a major reason why fingerprint recognition technology can occupy most of the market. The fingerprint collection head can be more miniaturized and the price will be lower.

　　Disadvantages of fingerprint recognition

　　The fingerprints of some people or groups are difficult to image because they have few fingerprint features; in the past, because fingerprints were used in criminal records, some people were afraid of "having their fingerprints recorded". However, in fact, the current fingerprint identification technology can ensure that no data containing fingerprint images is stored, but only the encrypted fingerprint feature data obtained from the fingerprint is stored. Every time a fingerprint is used, the user's fingerprint will be left on the fingerprint collection head, and these fingerprint traces may be used to copy fingerprints. It can be seen that fingerprint recognition technology is currently the most convenient, reliable, non-invasive and inexpensive biometric technology solution, and has great potential for application in the vast market.

　　The importance of software and firmware in fingerprint recognition technology Although price, size and hardware design are very important for fingerprint recognition systems, this only determines whether the entire system is robust, but the importance of firmware and software in the entire system is no less than that of hardware, especially for products that want to occupy the market.

　　The firmware in the fingerprint acquisition head is responsible for processing images and connecting to the PC. In most systems, the firmware is relatively simple. It keeps sending data to the computer. However, there are many problems with this. The first major problem is that the data transmitted in this way can be easily recorded and reused, which puts the system at potential risk. Another problem is that we must supply power to the fingerprint acquisition head. The computer must constantly capture images to determine whether there is a fingerprint pressed on it, so that the fingerprint image can be captured at the most appropriate time. By designing a good firmware to handle these problems, the performance of the entire system can be improved; using the USB interface is currently a good interface solution that can provide power, bandwidth and plug-and-play functions; the fingerprint transmitted to the computer needs to be encrypted to ensure security; and the firmware should be switched to a low-power state when the fingerprint is not being read.

　　After the host securely obtains the fingerprint image from the computer, the recognition algorithm proceeds to the next step of the verification process. Fingerprints are such reliable biometrics that only very little information is needed for comparison. However, this characteristic of fingerprints is not reflected in most fingerprint recognition algorithms. Most systems require users to press all their fingers, and users must press their fingers very carefully. If the fingerprint is not in the right position or the fingerprint quality is not high, the verification cannot be carried out, and the user must press the finger again, and such products cannot gain a foothold in the market.

　　Therefore, a good system should be easier to use and more reliable. Users do not have to worry about the placement of the fingerprint. The algorithm should support 360-degree rotation and incomplete fingerprints. Users only need to press their fingers lightly without worrying about whether the position is right or only pressing part of the fingerprint. The system should be able to solve the pressure, rotation, quality of the finger, as well as the dust and mist of the collection head.

　　Monitored and unmonitored operation, one-to-one comparison and one-to-many comparison. To use fingerprint recognition technology, we must first understand the difference between monitored and unmonitored comparison. In the traditional criminal fingerprint database, fingerprints are obtained under monitoring and the comparison is based on multiple comparisons in the same fingerprint database. To bring fingerprint recognition algorithms to the market, fingerprint reading must be based on unmonitored conditions. In this way, only light pressure is required without waiting for the fingerprint image to reach the best. In addition, when processing e-commerce, the data will be compared with the registered fingerprints. In most cases, it is a one-to-one comparison rather than a one-to-many comparison. Unmonitored imaging and one-to-one and one-to-many comparison algorithms have good countermeasures in the system. The unmonitored mode must enable the comparison algorithm to handle poor quality fingerprints and the algorithm must be more reliable than the monitored state.

　　A new era of biometrics

　　The increase in security needs has led to the accelerated popularization of biometric technology. We believe that security needs are the core driving force for the development of the biometric technology market. 1) Personal security needs: the explosion of biometric technology applications in the smart terminal market. According to our industry chain research, after the first application of fingerprint recognition technology on the iPhone 5S, the Samsung Note 4 mobile phone released in September may be equipped with an iris scanning recognition function to improve the security of the phone, and other technology manufacturers are also following suit. 2) Public security needs: large and medium-sized application systems will be accelerated under the leadership of the government. With the development of the global economy and the advancement of urbanization, all countries are increasing their investment in public security supervision such as terrorism and violence. Governments in major regions around the world, represented by the United States, the European Union, and India, have successively implemented large-scale system projects related to biometric technology to promote the accelerated popularization of biometric technology in the field of public security. The global biometric market will grow at a compound rate of 35.2% in the next five years, and the market size will reach US$37 billion.

　　According to our model calculation, the biometrics market, which is dominated by fingerprint recognition, face recognition, and iris recognition, is expected to reach $37.02 billion in 2018, with a compound growth rate of 35.2% from 2013 to 2018. In terms of structure, the driving force of fingerprint recognition and iris recognition mainly comes from the explosion of the smart terminal application market, while the driving force of face recognition mainly comes from the increase in demand in the public security market. Face recognition and iris recognition have a small base but strong explosive power, and their share of the entire biometrics market will increase from 11.4% and 5.1% in 2013 to 22.4% and 22.1% in 2018. As a populous country, China is accelerating the opening of the potential market space for biometrics, driven by multiple factors such as information security needs, public security needs, and the promotion of biometrics standardization.

　　Conclusion

　　Biometric recognition technology is not just the fingerprint recognition and face recognition technology mentioned above, but also includes many other technologies such as vein recognition and palm print recognition. However, fingerprint and face recognition are the most widely used technologies at present or in the future. In particular, face recognition technology, if it can make good use of the cameras that have been deployed, will be able to prevent some crimes from happening. Of course, personal privacy issues also need to be further explored. It is worth mentioning that the recognition technology that integrates multiple biometrics will also be one of the hot directions in the future.