Page 21 - Security Today, May/June 2019

people inside the facility.
It can recommend or automate higher lev-
els of control to avoid life safety consequenc- es. Compared to a human security guard, an AI-based system can monitor more data, faster and more accurately without bias or distraction. It can leverage that intelligence to become risk-adaptive—adjusting access permissions based on risk levels.
A Shift from Role-Based to Risk-Adaptive Access Control
Risk comes in many forms. Dynamic by na- ture, it can increase exponentially in severity in a just matter of moments. Unfortunately, traditional access control systems are static by nature and cannot independently adapt to a changing environment. That inability can be detrimental to life safety and security.
Most traditional physical access control systems are role-based. Access is assigned
based on a person’s role within an organi- zation, assigning access to an access group representing a collection of doors that role can access at specific times. While this meets a very fundamental need, it controls access strictly by those static roles and does not adapt as the situation dynamically changes. If an incident occurs that could affect life safety, it requires a human, i.e., a security of- ficer, to react to an alarm or situation and make changes to the access control system in order to protect someone who could un- knowingly enter a bad situation or danger- ous environment.
Today, we see the emergence of risk- adaptive technology based on AI and new levels of interoperability. When applied to access, risk can be based on multiple crite- ria and access permissions can be adjusted as situations or individuals change. A risk- adaptive system can monitor key data points, activity and risk levels for an individual or facility. Let’s look at some examples.
Environmental Risks: Mitigating danger
An employee may have authorized access to a specific location, but there may be a reason at that particular moment that the employee should not enter. There might be a safety threat and the risk-adaptive access control system would recognize this and prevent him or her from entering.
Consider some of the high-risk situations and what could occur if an unsuspecting
person entered an area of risk. In critical in- frastructure facilities, for example, there is al- ways higher than average risk. Chemical spills, radioactivity, fire and other incidents are the dangerous examples to name a few. Those are relatively obvious risks and even legacy access control systems can provide some rudimen- tary measures to seal off areas of concern. However, without added intelligence and in- sight capabilities such as risk scoring and ar- tificial intelligence (AI) to identify these risks, the current access control systems cannot ad- just based on rising or sudden threats.
Beyond the facility, external threats such as weather events or potential riots in a particular area of the city, can impact the risk of enter- ing a building or create a need for access to a safer area within the building. In those cases, the system would need to recognize the exter- nal threat and adjust access permissions while directing the employees to find safe shelter.
Emergency Response:
Changing Access
Permissions on the Fly
Conversely, I may normally not be autho- rized to access a certain area, but because of a high-threat condition and who I am, I can now enter. The combination of these two factors provided an exception to the access permissions. An intelligent access control system can analyze that data and adjust per- missions on the fly.
A perfect example of this scenario would be in an emergency lockdown situation at a
WWW.SECURITYTODAY.COM 15
Development of AI and Neural Networks
John McCarthy, who is recognized as one of the fathers of artificial intelligence or “AI,” coined the term in 1955 defining it as “the science and engineering of making intelligent machines.” That definition is still accurate, and while AI has taken decades to mature, it is being applied to almost every industry today—solving harder and harder problems, and doing so at faster rates than humans.
A big part of the advancements we’re seeing in A.I. have to do with its ability to design com- plex machine learning systems known as artificial neural networks (ANNs). These ANNs make cars drive autonomously, diagnose disease, detect fraud, and are now aiding us in the physical security world.
In the physical security world, artificial neural networks are being used to connect all the avail- able dots about people, environments and situations to make better decisions for improved life safety at schools, workplaces and facilities.
The artificial neural network (ANN) is essentially the brain of an AI-based security solution. Similar to the human brain, the ANN receives sensory input from cameras, sensors, access control systems, etc. It can then identify potential danger or threats, and transmit actionable guidance and commands to the appropriate people and systems to address the risk. It can evaluate data inputs based on established policy, as well as acquired and saved knowledge.
When AI is applied to physical security, it enhances situational awareness by identifying anom- alous events, enabling automatic risk adjustments, adapting access permissions and alerting us to threat conditions before an event occurs.
While AI can seem like a distant vision, it’s here today and working to improve the physical security industry. Looking to the future, we can expect A.I. and ANNs to become even more intel- ligent, risk-adaptive and responsive to our security needs.
sdecoret/Shutterstock.com