Artificial Intelligence, Real Security

AI has the potential to enhance multiple components of security systems

When considering the challenges of ensuring life safety, improving physical security and increasing employee and contractor accountability, there are more options available than ever before. It is important to develop a comprehensive plan that includes layers of protection and utilizes the appropriate technology to assist, rather than replace, the humans who are responsible for physical security.

Before examining some of the new technologies and how they can be utilized, consider the progress that has been made to get here. Not too many years ago, the security industry was a collection of companies that produced proprietary, closed architecture systems that were not interoperable. Eventually, though, leading industry organizations agreed to unite and support open standards.

There is often resistance to new ideas and approaches for a variety of reasons – some fear change; some want to delay change until they can catch up; and some see change as an existential threat. Fortunately, the security industry accepted change and embraced open standards. That is what has enabled so many of the advancements that are now available to improve life safety and security.

AI: The Myth, the Magic and the Reality

It is important to clarify what AI is and how it affects physical security. While AI is one of the hottest trending topics, it seems to be often misunderstood or mistrusted. At a recent security conference, one industry technology executive stated that AI does not exist. In other words, he saw it as a myth.

Fortunately, he is incorrect. AI is not a myth. It simulates the human intelligence process in that it acquires information and learns. It identifies patterns and it reacts based on those patterns. One of the core components of this learning process is an artificial neural network (ANN). Similar to the human brain, an ANN works with thousands of sensors.

Human sensors are eyes, ears, nose, skin and tongue. They identify features of the environment – people, things, temperature, light or darkness, moisture, sounds and many other things. As those sensors activate, humans learn. Internal variables adjust, patterns are identified and reactions are formulated. For example, the smell of smoke indicates fire, and the brain sends signals to the body to take appropriate action to escape danger.

A more sophisticated example would be how human behavior and, ultimately, a person’s character is assessed. Over time, personality traits are identified and changes in behavior are observed. All of this information leads to the development of a trust or no-trust relationship.

When it comes to technology, the artificial neural network sensors are cameras, sensors, access control equipment, Internet of Things (IoT) devices, big data, social media plug-ins and human input. As those sensors feed internal variables, the ANN learns and identifies patterns of risk, providing realtime situational awareness to human personnel.

AI may seem like magic, but it is real. It can evaluate large amounts of data to identify threats that might otherwise go unnoticed. This fundamentally changes the game in physical security and will affect all traditional components of a layered solution.

Networks: The backbone of integrated security

Reliable, secure and fast networks are fundamental to integrating security technology. Over the past few years, there have been tremendous advancements in cybersecurity. The implementation of AI in this category has produced solutions that can identify a wide range of threats via anomalous event detection. This is particularly important when information is transported between facilities or across cloud-based solutions.

One of the fundamental decisions regarding networks is the choice of cloud or no cloud. There are many variables to consider when making this choice. Are you deploying at a single site, a campus or an entire enterprise? Will you need internet services or interaction with third-party systems or sites? What is the relationship of price versus vulnerability. What IT resources and support will you need? Will the system that you are selecting adhere to your IT policies and work with you as is or will it require specialized equipment and support?

Perimeter defense: Spotting threats at a distance

With the advent of AI-based approaches, perimeter defense solutions can be an integral part of a layered security system. Specialized sensors can be utilized in conjunction with other elements and monitored with software powered by AI to identify threats just outside the walls of an organization. Improvements in multiaxis sensor technology has made this layer more affordable.

Entrance control: Understanding people flow to identify risk

While turnstiles may not have changed dramatically in form, their use within a layered security solution can provide invaluable data to an AI-based solution. Anomalous traffic patterns or unusual access can be identified before individuals reach higher security areas. Randomized spot checks, controlled throughput and directed traffic flows can be initiated and managed via intelligent, risk-based solutions as threats are detected.

Access control: Mitigating risk in real time

No physical security devices have benefitted more from AI than physical access control systems. When integrated with an AI-based solution, an access control system can now quickly react to threats and adjust permissions accordingly. Having the ability to identify anomalous events, insider threats, and hazardous situations and dynamically change permissions is a major breakthrough for the physical security world.

The use cases in this category are wide ranging. For example. when AI is applied to access control, it can identify unusual activity, such as off-hour and abnormal location access and combine it with other threat indicators to quickly identify insider threats. Risk-adaptive access is able to prevent people from entering an area where a traditionally “non-obvious” danger is located, and it can allow a first responder with the right credentials to access an otherwise restricted area when threatening conditions exist.

Video surveillance: The unblinking AI

Video has evolved tremendously with the migration from analog to digital and DVR to NVR, and with the adoption of ONVIF and cloud deployments. There is now a suite of video analytics available and some analytics even exist on the camera itself.

When used as a sensor for an ANN in an AI environment, video is a key component for providing realtime, situational awareness to the security operations center (SOC).

Traditionally, SOC operators have been presented with hundreds, possibly thousands, of camera feeds to monitor. Studies show that human attention spans are limited in their ability to effectively monitor video and identify potential threats. However, when video is part of an AI-based solution, attention span is limitless. AI never gets bored or distracted. It can sift through the mundane and assist humans in identifying threats as they emerge.

Drones: Assessing risk from above

The use of drones is sparking interesting discussions about standards, public policy, safety and privacy. These conversations typically go down one of two paths. Some will look for systems that can use drones to assess threats. Others will look for systems that can detect drones, viewing the technology as a threat.

In some security environments, drones provide tremendous assistance in threat assessment and verification. At critical infrastructure sites and across wide areas and difficult to reach locations, drones with video devices can help operators better understand developing situations. A drone can be quickly dispatched to evaluate anything from a pipeline leak to a lone worker who is nonresponsive. Drones can also be sent into hazardous conditions, such as chemical spills, before humans enter, to assess the situation and gauge the danger level.

When added to an AI-based solution, drones can be automated, and the system can ensure that they work according to all public and private policies.

Robotics: Addressing risk on the ground

While drones can help assess a situation, advancements in robotics can help address it. Robotics options today can be overwhelming. Robots can ride motorcycles, drive vehicles, jump high, climb ropes and scale cliffs. This opens tremendous possibilities in the security environment. Robots can take action to rescue people from dangerous situations such as fires, explosions or accidents in locations that cannot be easily or safely reached by humans.

Software: Bringing the system together

Over the years, security software has evolved from single purpose systems to integrated systems. With standards, the evolution has continued to interoperable solutions. Physical security information management (PSIM) systems were an attempt to unify information into a common user interface.

Because of advancements in networking, data management, CPUs, GPUs and many other areas, today we see more than an evolution. Platforms have developed in interoperable environments fully utilizing the technological advancements that go well beyond the hopes of PSIM, moving the security industry into a new era.

It is an AI revolution in physical security.

John Carter (jcarter@reconasense.com) is co-founder, president and chief technology officer of ReconaSense.

This article originally appeared in the spring 2019 edition of SIA Technology Insights.