TechSec 2017 panel: mining for gold with big data

Panelists discuss the current state of big data within the physical security industry
 - 
Wednesday, March 8, 2017

DELRAY BEACH, Fla.—Companies are starting to look at the data that is already being collected by various access control and video surveillance as an opportunity for additional services or RMR. A panel at TechSec Solutions 2017, held here Feb. 27 and 28, explored where the value is and how or where companies should get started.

“If you are going to get into big data in the next 12 months in the physical security industry, you are going to be ahead of the curve,” panelist Steve Carney, senior director product marketing, video and integration platforms, TycoSP, said. “Being at that cutting edge and learning is a key element to longevity in what will probably be a large, lucrative part of our business going forward.”

The panel’s moderator, Brian Phillips, associate director, global security and resilience for Alexion Pharmaceuticals and one of Security Systems News’ “20 under 40” Class of 2016 End Users, opened the session with a basic question: What is big data and why should we use it?

Carney answered first: “It’s the aggregation and storage of data, but along with the term ‘big data’ is the analytics portion: what is happening to that data? That data is being mined for trend analysis, for patterns and for anomalies.”

This data is used to gain insight, Carney continued, “That insight can be [from] a forensic standpoint—what is happening now or in the past—but it can also be predictive; what could happen in the future and what should my business do to take advantage of those odds.”

Panelist Rick Urban, vice president of operations at Edge360—a company that designs, manufactures and installs situational awareness systems, said, “In the physical security industry, I think [big data’s] really just starting to scratch the surface.”

Though, Urban said that the industry fits “the three Vs” of big data: volume, velocity and variety of the types of data. “We certainly do that in this space, but we haven’t gotten big into the data fusion and the machine learning aspects. … I think that will start to play a bigger role in the ultimate goal, in my opinion, which is the predictive analysis—or the predictive analytics—within physical security.”

“True big data and predictive analytics are when you can take the human out and let the machine do everything, and we’re clearly not there yet,” Urban said.

Phillips said that his company is currently retaining a lot of its data, but that has not yet stretched to include data from physical security systems. “While it may not be the cheapest thing, it’s gotten a lot cheaper now to store some of that data,” said Phillips. “I think we’ll see it in physical security once we get a few pioneers to do it and talk about it, so that everyone else gets that comfort level that it’s going to work,” he continued.

Is big data just having a PSIM, Phillips asked the panelists, or is there more?

“I think there’s more,” Urban said, adding that events within PSIM systems, and the users’ responses, can be data points that end up helping big data or predictive analytic environments.

“PSIM is its own animal,” Carney said. “It’s not making the data, it’s not storing the data, it’s not necessarily analyzing the data, so it really does have a different function.”

Big data gathered from access control systems can help insider threat detection, Phillips said. “Take a person’s badge history, and create somewhat of a profile for that person, figure out their badging trends. If I come in on a typical day, Monday through Friday, 9 to 5, and I go exactly where my office is—why am I now at 6 p.m. or 7 p.m. trying to scan the CEO’s office, where I don’t have access? Or, why am I coming in at 2 a.m. on a Saturday?”

Big data is still emerging in the physical security space, the panelists agreed.

“First and foremost, it’s early days in this space,” Carney said. “If you are interested in getting in this, understand first where you want to be in the value chain—data has value in and of itself—and who are your customers.”

He continued, “If you’re dealing with large enterprises, one of the things that getting into big data does for you is it allows you access to that customer beyond just the security director; the IT department, facilities, they may have a supply chain—they are all interested in the type of data that can help them.”

Carney recommended one possible first step with big data: hiring an employee with a data-focused background. “It’s really about starting small, trying to find the value for your customer, … refining your understanding and the capabilities, proving it out, showing the value, and moving on. We really think that that’s how big data really proliferates through the physical security business.”

Big data can include more sources than just the systems that are internal to a location. Carney listed two offerings, Rapid Miner and DataRobot. “Those are off-the-shelf tools that will help aggregate data. Data comes in two flavors, structured and unstructured data. One’s easier to manage, one’s a little bit harder—but the tools out there will help.”

Phillips mentioned another tool: GDelt, an index of online and print news. “I think we get just as much out of open source as we do out of our paid services,” he said.

There are some pitfalls to big data, its storage and its usage. “I read a book on big data and one of the biggest pitfalls was drawing correlations that aren’t there. When you have all of these big sets of data, there is coincidence,” Phillips said.

“There are two primary concerns in the space of big data. One of which is something we all deal with every day and that is cybersecurity,” Carney said, and big data can extend a company’s risks with cybersecurity. “If you are using a public cloud or some other type of data repository, you have the third-party’s cybersecurity profile to be concerned about.” Privacy can be particularly concerning as countries have different laws around data privacy, he said.

Urban pointed out that the costs “to implement, maintain and operate these sorts of systems is still out of reach, I think for a lot of the smaller types of companies that might get benefit from these things.” Storage is another challenge, which is also related to cost, he added.

Closed access control systems can be another difficulty in big data within the physical security realm, Phillips pointed out. “Not all access control systems have an open and friendly API to integrate to. That can be very complicated.”

How does a company get started?

“There’s still a lot of confusion around what is the cloud, what is big data, what is predictive analytics, and I think a lot [of] understanding where to start is really about educating and trying to figure out what your needs are and what—exactly—each one of these things means,” Urban said.

“It’s a service revenue around data. First and foremost, your organization has to start thinking about how to manage service-oriented revenue, versus just product sale and maintenance,” Carney said.

“I would say start small, and find a good customer that you could partner with—you’ve partnered with before—and will work with you on shared insight,” Carney said.

Carney pointed out the relative value of data gathered from different sources. “Data is a wave of the future for us, and a video camera is by far the richest source of data there is in sensors out there between size, shape, speed, color, amount, duration.”

One attendee shared his experience with “midland data,” as opposed to big data; his company was able to use access control information to look at the number of people coming and going from a building, which gave the building owner a better idea of the size of real estate needed for their facility.

Carney responded, “I think the example is a good one, with regard to how value is derived and initially it will be operationally … optimizing activities where the data helps [to] give insight.”

Phillips added, “The midland data—if you will—is not to be forgotten.”