Your heartbeat is as unique as your fingerprint

©WIRED

In the US, the cities of San Francisco, Somerville, and Oakland recently banned the use of facial recognition by law enforcement and government agencies, while in Portland the discussion is about banning the use of facial recognition entirely, even by private businesses. Some thirty private organizations representing more than fifteen million members have come together to call for a general ban on facial recognition by US law enforcement agencies.

In the UK, revelations that the London Metropolitan Police secretly provided facial recognition data to the developers of Kings Cross Estate for a secret facial recognition system sparked outrage; an inquiry was demanded. The office of the Information Commissioner has launched an investigation into the legality of the program. However, the scandal happened to coincide with a landmark judgment by the Cardiff High Court holding that the use of facial recognition by the police in South Wales is legal – a decision that is likely to be challenged.

When it comes to state surveillance, facial recognition is only the tip of a huge iceberg, however. If stricter rules for the use of facial recognition are adopted, it is possible that we will simply switch to one or more of the other types of surveillance technologies currently under development. Many are just as invasive as facial recognition, if not more so – and possibly even harder to regulate. Here is a look at some of the technologies looming in our future.

The way you walk?

The rapidly growing field of behavioral biometrics is based on recognizing individuals based on their patterns of movement or behavior. One example is the recognition of a person’s gait, which may be the next big surveillance technology, especially if facial recognition becomes subject to more stringent regulations. This technology is already being tested by police in China, which is often the leader when it comes to finding new ways to monitor people – whether they like it or not.

There are several ways to recognize a person by his or her gait. The method that is being tested by the Chinese police is based on the technology developed by a company called Watrix, and uses video from surveillance cameras to analyze a person's movements while walking. In a recently awarded patent, Watrix describes a method of training an AI system using a convolutional neural network that allows thousands of data points to be analyzed about a person as he moves, from the length of his stride to the angle of his arms. They then use these parameters to identify people by their "gait record." Watrix claims that its systems achieve up to 94 percent accuracy and that it possesses the world's largest database of gait records.

With the visual gait recognition methods developed by Watrix and others, people can be identified from a distance, even in crowds or on the street, in much the same way as facial recognition works – which means it could represent a quick and easy substitute if facial recognition would ever be prohibited. Increasingly, many video surveillance systems are collecting multi-modal biometric data. This means that they may be using both facial recognition and gait recognition simultaneously, which theoretically should increase both the accuracy and the ability to identify people who are facing away from the cameras.

Another way to identify people through their gait is based on sensors embedded in the floor. Researchers at the University of Manchester have used data from 20,000 footsteps taken by 127 different individuals to create a neural network and identify 24 different factors, such as how a person's cadence progresses and the time it takes to move from one heel onto the next. Using this system, they were able to identify people with an accuracy of over 99 percent in three "real" scenarios: at work, in the home, and at security checkpoints at the airport.

According to the researchers, the advantage of this type of identification over visual systems is that it is less invasive and less prone to being disturbed by objects or other people obscuring the camera's view. Or in other words: the people being monitored don’t recognize that they are under surveillance. People do notice that they are being watched by cameras, but they won’t be aware of sensors in the floor.

Surveillance by heartbeat

Your heartbeat and breathing patterns are as unique as your fingerprint. A small but growing number of remote sensing technologies are being developed to detect vital signs from a distance – signals that pass through skin, clothing, and in some cases, even walls.

In June, the Pentagon announced a new laser-based system that can identify people from a distance of up to 200 meters. This technology, dubbed “Jetson”, uses a technique known as laser Doppler vibrometry to detect surface movements caused by your heartbeat.

The goal is to be able to identify an object within five seconds by its cardiac signal or "heartprint". For the moment, however, the Pentagon system still has a number of limitations. For example, the target person must be standing still, must be wearing only light clothing (thicker clothes, such as an overcoat, can disrupt the signal), and above all, there must be a clear line of sight between the laser and the target.

Coats, walls, even stones and debris are no obstacle, though, to a completely new surveillance technology. Researchers are working hard to develop radar-based systems that can track vital signs for a variety of other purposes, from non-invasive patient monitoring, to supporting medical diagnoses, to assisting in the search for survivors in search and rescue operations.

Read the complete article here.