SFO: Steady Vision, New Tools

April 1, 2003
Nowhere has the image of counterterrorism activity been more vivid than at the nation's airports. The exposure for catastrophic human damage, the scope of commercial assets at jeopardy, and the potential for socioeconomic havoc occasioned by a wounded transportation system have combined to make airport security a priority. Additionally, September 11 and air hijackings generally have given aviation security an emotional legacy that is now part of our national consciousness.'

Yet these issues were real before the war on terrorism began some 18 months ago. San Francisco International Airport, like other airports around the country, already had in place equipment and procedures intended to detect, prevent, mitigate and recover from a wide range of threatening incidents.''

It has become clear at SFO and elsewhere over the past year that the challenge to move security to the next level is about more than pressing the pedal to the floor, more than just adding resources. It is about doing two things simultaneously: Holding the existing system to more rigorous standards and probing, planning and executing the development of more sophisticated and innovative systems for tomorrow.'

Mark Denari and Bob McKinley are assistant deputy airport directors at SFO, for aviation security and international terminal operations respectively, and have the kinds of roles that the public imagines to be at the center—or at least the front line—of an airport's efforts. Of course, they are quick to dismiss that kind of characterization, pointing to the responsibilities and accomplishments of other airport executives and TSA officials.'Their roles, nonetheless, combine several functions that are mission critical.'

Getting a Head Start
McKinley has focused on the explosive detection system (EDS) that is now installed in the international terminal and is currently being implemented in the domestic terminals. The system uses InVision Technologies' CTX™ equipment, which sells for about $1 million a pop. The decision to go with InVision was based upon two elements. The first was familiarity. Four InVision machines were already operating at SFO when the decision was made to install an automated system for checking 100 percent of the baggage that would come through the new international terminal. (This plan was in place before federal mandates were issued.) Additionally, InVision was local to the Bay Area and thus provided easy access to executives and other personnel whenever necessary.''

From the operational perspective, both SFO and InVision had experience working off the "European model" for screening for explosive devices, a multi-stage process of using two-dimensional X-rays, human inspectors and CTX machines. The new system envisioned by McKinley and Denari would build off that approach but would eliminate the X-ray tool and use the CTX for all luggage, not just the pieces an X-ray had identified as suspicious. Just as important, InVision had more experience than others with in-line inspections and knew the issues surrounding baggage conveyor systems. Not to be diminished either was the wider aperture of the InVision equipment, allowing larger bags to be treated routinely.''

Beyond the European Model
In an airport that inspects between 12,000 and 15,000 bags per day, the first issue after "Does the machine do what it says it will do?" is efficiency. To that end, McKinley and Denari used mathematical modeling techniques to determine the best plan of action. They varied their input and assumptions according to historical records, estimates of passenger load and baggage volume that correlated with facility size, and capacity of the conveyor systems. They experimented with different scenarios, such as what would happen if several machines went down during peak periods (say, from a bag jam or mechanical problem).''

The upshot of this planning is a program that is different from what exists in most of the country's other major airports. In a word, the system is about multiplexing—the ability to do remote screening of multiple bags while they are being processed at multiple CTX locations throughout the airport. Instead of having operators "tethered" to specific machines, the system at SFO has all operators—each a TSA employee—viewing CT images from a central screening room. The advantage is one of human resource allocation. Images ("bags") flow to operators as each is ready to view or inspect another piece of luggage. During slow periods (for example, afternoons as opposed to mornings or evenings), the number of operators can be reduced and some can be re-assigned. The operational point of view is that machine downtime doesn't demand underutilization of manpower.''

The operators, of course, only look at a small percentage of the bags that are moving through the system. These are ones that have been flagged by the CTX or, more precisely, by the artificial intelligence software that has recognized some attribute that needs a human evaluation—say, the shape or density of an object, or its puzzling position. The system, therefore, is automated with technology alerting humans for the exception. The result is 100 percent CTX redundancy for suspicious bags. If a bag's contents are still in question after an operator has evaluated its image, then it is directed to the BIR, baggage inspection room (sometimes described with the more graphic term “bomb room”), where operators follow up with closer physical inspection techniques.'

The challenge of monitoring the movement of thousands of bags every hour through what amounts to almost seven miles of conveyor systems at SFO begs for additional technology aids. RFID tags have been used before, but newer applications are being evaluated for better accuracy in determining the location of a bag in the system. Other software programs are being developed that give operators touch-screen tools for quickly pulling up threat images for any bag that needs the hands-on physical evaluation. "The game," said McKinley, "is to use your resources efficiently and to stay ahead of what bad guys might do."'

Physical Access Control
Denari, meanwhile, has been the manager responsible for developing and assessing the protocols that assure that SFO's operations comply with standards proposed by the TSA. His approach is built around a model that has three elements: programs, people and technology. Denari's team uses the model to understand the threat vectors that endanger the airport and how to respond to them.''

The process includes an assessment of the unique features of SFO: its geography, size, and waterfront exposure, as well as the issues associated with a large international, long-haul clientele. The programs and technologies suggested by the model are subject to stiff cost/benefit analyses and collaborative input from multiple groups (from executives to union workers). Drills to practice emergency procedures help assure an engaged worker population. In fact, each year SFO normally runs three or four drills related to air piracy, air crash and other extraordinary circumstances—more than what has been mandated by the TSA (or prior, the FAA).'

As part of the planning effort, Denari co-hosted in the Fall of 2002 a notable meeting with the San Diego airport to identify technological applications for an airport security operations command center. Participants included a select group of invited vendors, TSA officials, an assortment of military brass and representatives of other airports. They surveyed a host of issues, from escalation procedures to new technologies. He ended up selecting one participant, NEC, to help design a solution for controlled physical access.'

Much like McKinley's task with luggage, Denari has the problem of increasing security while maintaining an acceptable throughput of employees. With 25,000 to 30,000 workers (both employees and independent contractors) having access to the runways, for example, the challenge is daunting. SFO currently uses a system that has both an ID card and a biometric measure (hand geometry technology from Recognition Systems) working off the same platform. The redundant configuration was the first ever employed at an airport, but it has some shortcomings, not least of which is the inability to prevent or detect piggybacking by unauthorized personnel, an application for which it was not designed. Of course, the workable anti-piggybacking solution is in Denari's words, "the holy grail of access control systems."'

Smart SurveillanceNEC will soon pilot a smart surveillance system, a la the artificial intelligence model used by McKinley in the baggage domain, that will identify unauthorized intruders and will integrate within SFO's existing CCTV configuration. Cameras will have object recognition software on board, running on advanced video processors. Security operators would be notified when an exception to an accepted protocol took place, as in two people going through a gate when only one has permission or when someone is carrying a package when none is allowed. At the same time, the system's robustness has to include a low false negative ratio—that is, it can't jeopardize the access of the honest majority.'

An intelligent camera approach would diminish the need to have squadrons of employees looking at displays hour after hour waiting for something of interest to happen. To the degree that the intelligence can be moved to the camera (as in InVision's equipment), the efficiency of the system is increased. The backhauling of useless information decreases; the discriminating power of the technology network increases.''

The goal of an NEC-like system is to show a quantifiable increase in security and public confidence without incurring delays or inconvenience for employees or customers.''

About Attitude Both Denari and McKinley have to work within the economic constraints at SFO. Additionally, they each have to deal with a host of constituencies—collaborating with technology firms, integrators, union workers, TSA officials, airline employees, and, of course, community leaders. Each recognizes that in the final analysis it is securing the entire transportation network that is important, not just SFO. In short, an appreciation for interdependency and integration is at the root of their perspectives.'

Moreover, both communicate a desire to build systems that not only pass muster today but that also have the ability to absorb the lessons of experience, to learn from mistakes in pursuit of the better solution. Redundancy across solutions is good; redundancy and learning is even better. It is not only an intellectual honesty that is being honored with such a commitment but also a pledge to give society the security and peace of mind it deserves.


Nicholas Imparato, Research Fellow, Hoover Institution, Stanford University ([email protected]) and Professor of Management, University of San Francisco ([email protected]), is also a member of the policy and planning committee for the Bay Area Economic Forum's homeland security initiative.