Video and data analytics have been hot topics among manufacturers and on tradeshow floors for much of the past decade. We’ve been excited to see analytics catching on in security operations centers and enterprises, but there are still some challenges facing these technologies and their users. Security posed some questions about analytics, integration and technology trends to Joe Young, Director of Cloud Monitoring Services for G4S Secure Integration. Stay tuned for a second blog featuring insight from Joe on robotics, virtual reality and gaming culture for security next month.
On Data Analytics
SECURITY: How can end users use data to drive business decisions?
JOE YOUNG: Some customers are adopting big data and data analytics, but everyone is not there yet. Whether they have adopted or not, many see the value of capturing data and layering on advanced analytics to look at historical data, trend analysis, pattern and anomaly detection to make intelligent business decisions. One example that comes to mind is how end users can use data and analytics to efficiently staff a building and provide the right level of security based on real-time threat levels and deliver a better security experience to their employees and guest. Consuming internal and external intelligence feeds, data from your core security platform and data from non-core security platforms such as IT systems, HVAC, lighting, robotics, and building automation systems can allow you to create a predictive workforce management model to determine how many security officers, lobby ambassadors, and how many remote services would be needed on a daily, weekly or even a monthly basis. This is a data-driven approach to predictive workforce management.
SECURITY: How is security system data playing a role in this?
JY: Security system data traditionally was not looked at before. Usually when big data systems are being looked at, it's an IT department that spearheads this. They look at the IT side of the house, not physical side. It’s an evolution of IT and OT merging together to get a better snapshot of what is happening – real situational awareness beyond traditional security. Being able to consume different data points of a security system is key to figuring out how and when people flow through a building. There is a need to figure out how incidents and anomalies happen. Basically, the more pieces to the puzzle you can get means the more accurate the picture becomes which allows you to make intelligent decisions. You get clearer visibility when you are able to look at all these different aspects.
For example, we can look at basic burglar alarms. Traditionally, in the retail space, you can put in an alarm for the building that will protect that building when no one is there. Over the past few years, companies have hired data scientists and built data mining solutions. Now they can take traditional burglar alarm systems and leverage the front door action, allowing them to see and predict traffic patterns for busy times. They can turn around and start to use this data for things like proper staffing. For example, if a company is open from 9 to 5, but between 11 to 12 they see more door opens from the burglar alarm panel, they may need more staff during those times. That, combined with traditional POS systems or other types of shopper experiences, gives a true granular visibility to what is going on in their environment. These situations can range from this simple example to a much more complex situation involving consuming more data and targeted marketing to deliver a greater shopper experience.
SECURITY: How have other departments or functions within enterprises been getting involved in or benefiting from security analytics and data use?
JY: One department that can benefit from security analytics and data use is property managers looking at space utilization. Operations now have visibility to look at things such as access control data. For example, you can look at 10 different buildings and have access control in all those buildings with a hybrid of on-site and remote workers. Without the right data it is challenging to know who uses what buildings, what workspace is available in the building, whether the space is being used and what you should do with the space. Looking at simple access control usage, if no one uses the space, you might consider selling or subletting it based on basic access control, visitor management and space utilization information gathered through analytics.
This has been around a long time, but is still a great example of how operations can take advantage of this additional data to provide greater insight to their analytics.
SECURITY: How can analytics help end users better utilize their security budgets?
JY: Analytics can help an end user decide where to invest their money. It helps identify the risks to an organization, show the financial impact if that risk happens and allows you to proactively put counter measures into place to mitigate against that risk. Analytics are a proactive measure to mitigate risks and optimize budgets. I believe that leveraging data analytics to optimize a client's entire security program is the future of the security industry.
Look at how traditional security integrators design a system today. The system that is chosen and the security assets that are deployed are usually based on the sales reps and sales engineers current knowledge or comfort level or can be persuaded by a manufacturer that is working closely with that integrator. There is a shift towards a risk management model and part of this is powered by data analytics. There are two important parts of risk management:
- Risk assessment – which asks, “What are the threats to the clients people, property or assets?” It gives you an accurate view of the vulnerability to the organization.
- Risk control – with risk control, you use the risk assessment and data analytics to put the proper countermeasures, security systems and guarding resources in place to develop an optimized security program to mitigate against the risk that were outlined in the assessment
SECURITY: How can analytics help to track incidents and draw attention and resources to incident-prone areas?
JY: To keep it simple, I put analytics into two different categories: Identity Analytics and Incident Analytics.
Identity Analytics allows you to create a baseline on how people navigate through a building on an hourly, daily, or weekly basis. For example, if you know an employee comes in from in 9 to 5 every day, takes lunch at a certain time, does this task or that task at the same time every day, that creates a normal baseline pattern. But what if that employee comes in at night, what happens? Is he doing something malicious? Or did he lose his badge and it is the person that stole it that is in the building? This change in behavior create an anomaly which can be fed into an incident management system to notify an analyst that an abnormal behavior has been detected and further investigation is needed.
Incident Analytics looks at other data such as slips, trips, fall, external crime stats that include robberies, theft, vandalism, etc. Each incident is captured with a time/date stamp and a location. Continuous monitoring of this data along with the proper analytic algorithms can help detect patterns that the naked eye might not catch. From this data you have the ability to create heat maps that show where the incidents are occurring but more importantly you can predict where the next incident may occur in the future so you can properly deploy the right counter measure. This could be new security cameras, static officers, patrolling officers and even use of remote video services with voice down capabilities to deter against future incidents.
On Audio & Video Analytics
SECURITY: Please explain the difference between Focus of Action and Focus of Attention analytics, and what they mean for end users.
JY: I recently met with a video analytics company that broke down analytics into two groups: Focus of Action and Focus of Attention. This is a very clever way to look at how video analytics perform in a client's environment.
Focus of Action video analytics are a clear set of rules that have been written and when they are violated, have a set of standard operation procedures to respond to for each event. One example is an outdoor area that is protected with video analytics on a camera. When configuring the camera, you have the ability to pick from a list of available analytics such as, cross line detection, object left behind, object removed, color change, speed violation, loitering, direction violation, etc. For this example we will use cross line detection. When someone in the camera scene walks across the line, it triggers an event, and it sends an event to an operator in the command center to run through a checklist or standard operating procedure (SOP) of what needs to happen – this is a very structured and rigid process – “something happens and you do this.” Another example could be in airport, when an object like a backpack is left behind which could be a potential threat. You will have a new whole set of standard operating procedures to follow to deal with this type of event. Each event can have a different set of procedures attached to it depending on the client's needs.
Focus of Attention is a new style of video analytics that is coming to the market. Unlike the previous example, when installing the camera there are no analytics rules that need to be configured. Instead, the camera is equipped with intelligence and over time learns what the normal baseline scene should look like. If the camera detects some abnormal activity within the scene, it will create a notification which is sent to a security operator to focus their attention on this camera that there may be something abnormal going on in the scene. An example might be a camera in a mall that faces the escalators when you have normal traffic patterns of people going up and down the proper side of the escalator. However, if someone is going in the opposite direction, the camera would notify the security operator to look at the camera. Maybe it is a group of kids playing on the escalator or maybe someone has fallen down the escalator and is injured.
SECURITY: How are analytics changing how security operators and monitors work? How does this make security monitoring stations or security operations centers more effective?
JY: Monitoring live cameras is not the best use of an operator’s time. It is not cost effective to have your team of operators looking at live video all day long. Keeping operators engaged in that type of environment is virtually impossible. If we look at how monitoring centers are delivering remote video services today, most cameras are equipped with video analytics so as an event comes in operators have a specific set of instructions to follow allowing them to meet any kind of compliance or auditing situation. This approach is much more efficient. Operators are not sitting there looking at live camera feeds hoping to run across an anomaly. Instead, operators can be engaged doing other proactive tasks and can focus on an incident 100 percent as it happens and then return to what they were previously working on.
SECURITY: What trends are you seeing in audio analytics? Are there specific applications or sites where audio analytics are more valuable than others?
JY: These trends are similar to video analytics with lots of evolution especially as it relates to smarter technology. Gunshot detection is an area where we have seen quite a few advancements in delivery. Traditional systems required dedicated wiring to be run throughout a facility, making it challenging and expensive to deploy these solutions. New IP based systems allow a client to take advantage of their existing network infrastructure to deploy a new system; this is similar to when IP cameras first became available to the market. Audio analytics also have the ability to be loaded onto the edge of an IP camera giving clients the ability to deploy analytics like gunshot detection very easily and very cost effective. With the ease of deployment I think we will start to see new audio analytics arrive to market over the next few years. Each sound has a unique audio signature. Aggression detection is a good example. This analytic can help in any environment where a conversion gets heated between two people or someone yelling for help, this analytic can trigger staff or security resources to be dispatched to make sure the situation is under control.
