4 ms·

A simple alarm(1) would have sent a patrol car out to investigate. Sorry, but no. They're covering miles of outdoor perimeter. The problem is that there are

by brk 14y ago

A simple alarm(1) would have sent a patrol car out to investigate. Sorry, but no. They're covering miles of outdoor perimeter. The problem is that there are very few technologies that can reliably detect things like people, while ignoring things like various animals (or heat from jet engines, etc.).

david_b 14y ago
I don't know when they erected this security system, but shouldn't this be a much smaller problem now with the progress in image processing? Put up a bunch of cameras and let humans weed out the false positives?
- brk 14y ago
  This system is/was actually using video analytics as part of the overall strategy. Video analytics can be very reliable, but you can't rely on humans to weed out the false positives very effectively. I've heard of (heard of, as in been involved in replacing them) video analytics systems that generated hundreds of false alarms per night on similar sized deployments. Any human operator is going to quickly become immune to the alerts when they happen at that rate.
noonespecial 14y ago
Ported Coaxial Cable Sensor (or leaky cable) has been around for a very long time(1) and was quite reliable even before the advent of computerized DSP to help with detection. (1)I think you could buy them commercially at least as early as the 70's
- brk 14y ago
  Buried coax and fiber detection systems tend to rank in the bottom 1/3rd of overall reliability (just above the systems that try to detect fence movement and tension changes). In fact, I rarely hear of new deployments considering these technologies (though I'm sure there are plenty of them, as the companies are still in business).
powertower 14y ago
> there are very few technologies that can reliably detect things like people, while ignoring things like various animals (or heat from jet engines, etc.). If a 100MM taxpayer funded system can't tell the difference between a jet engine heat output and a person's heat signature, then someone should be criminally liable for defrauding us. And as far as differentiating a deer from a human, it should have detected the moving object, sent the images to a person sitting in front of a computer, and he/she could have classified the object.
- brk 14y ago
  First off, the 100MM was a total expenditure across several airports, for multiple aspects of their security systems. It was not all spent on the fence perimeter system at JFK. JFK is a complex airport, I'm not aware of the specific length of the perimeter they are covering there, but it's likely between 8 and 15 miles of perimeter fence line. You need something with a high degree of reliability and false alarm immunity to make this work. This is especially true because the fact is that the probability of a real terrorist incursion is pretty low. So, the guards aren't going to stay actively engaged with reviewing alarms for very long if essentially 99.9% of what they see are all false alarm/no-action events.

by brk 14y ago

david_b 14y ago

I don't know when they erected this security system, but shouldn't this be a much smaller problem now with the progress in image processing? Put up a bunch of cameras and let humans weed out the false positives?

brk 14y ago
This system is/was actually using video analytics as part of the overall strategy. Video analytics can be very reliable, but you can't rely on humans to weed out the false positives very effectively. I've heard of (heard of, as in been involved in replacing them) video analytics systems that generated hundreds of false alarms per night on similar sized deployments. Any human operator is going to quickly become immune to the alerts when they happen at that rate.

noonespecial 14y ago

Ported Coaxial Cable Sensor (or leaky cable) has been around for a very long time(1) and was quite reliable even before the advent of computerized DSP to help with detection. (1)I think you could buy them commercially at least as early as the 70's

brk 14y ago
Buried coax and fiber detection systems tend to rank in the bottom 1/3rd of overall reliability (just above the systems that try to detect fence movement and tension changes). In fact, I rarely hear of new deployments considering these technologies (though I'm sure there are plenty of them, as the companies are still in business).

powertower 14y ago

> there are very few technologies that can reliably detect things like people, while ignoring things like various animals (or heat from jet engines, etc.). If a 100MM taxpayer funded system can't tell the difference between a jet engine heat output and a person's heat signature, then someone should be criminally liable for defrauding us. And as far as differentiating a deer from a human, it should have detected the moving object, sent the images to a person sitting in front of a computer, and he/she could have classified the object.

brk 14y ago
First off, the 100MM was a total expenditure across several airports, for multiple aspects of their security systems. It was not all spent on the fence perimeter system at JFK. JFK is a complex airport, I'm not aware of the specific length of the perimeter they are covering there, but it's likely between 8 and 15 miles of perimeter fence line. You need something with a high degree of reliability and false alarm immunity to make this work. This is especially true because the fact is that the probability of a real terrorist incursion is pretty low. So, the guards aren't going to stay actively engaged with reviewing alarms for very long if essentially 99.9% of what they see are all false alarm/no-action events.