Saturday, December 20, 2014

Sound of gunshots, Part III: Forensics

During the height of the French Revolution a Priest, a Lawyer and an Engineer were in line to be executed.

The priest was placed on the guillotine.  The blade came down and stopped an eyelash from the priest's Adam's apple.  The priest praised God's will and demanded to be set free.  The executioner complied.

Next, the lawyer was placed on the guillotine.  The blade came down and stopped an eyelash from the lawyers flaccid neck.  The lawyer quoted law stating that the State cannot execute a man twice.  He demanded that he be released.  The executioner complied.

Finally, the engineer was placed beneath the blade.  The blade came down and stopped within a thumb's breadth of his neck.  The engineer looked over at the executioner and said.....

"I think I see your problem."

As a conservative

As a conservative I have reservations about pointing out what appears to be technically feasible out of concern that I will be giving the heavy hand of Big-Brother "ideas".  But from what I have seen, Big-Brother is miles ahead of anything I will ever think of.  Therefore I am more inclined to share information than to withhold it.

Consider a few random observations:
  • Virtually every person in the demographic most likely to commit gun violence carries a smartphone.
  • If they did not have a smartphone, the government supplied them with one.
  • Their smartphones are always on.
  • Smartphones are regularly updated with software.
  • Smartphones have microphones.
  • Smartphones have GPS.
  • Smartphones can talk to each other.
  • Smartphones can be programmed to be the ultimate "flight recorder".
 Mathematically, the location of a gun-when-fired can be determined by the time:date stamps and GPS coordinates of three, nearby smartphones.  More smartphones can be used to more precisely locate the origin of the "BOOM", but three is almost always good enough.

The sound trace of a gun shot is sufficiently unique that it can be mathematically identified with a high degree of reliability.

Smartphones can be programmed to log their location whenever they "hear" a gunshot.  This information, conceivably, could be harvested with a warrant and without the knowledge of the smartphone carriers.

Any person carrying a smartphone that records a rapid succession of crisp, high-intensity echoes is either the shooter, the victim or an accessory.

Anatomy of a mugging

"Mark" wearing pink and walking into the light.  Accessory is wearing blue and is flanking him.  By Accessory's position, he has "marked" him as the target.  85% of population is right-handed, so strong-side is toward buildings.  Ambusher is wearing blue and is hiding in shadows.

Ambusher leaves shadow and gets light behind him.  Flanker boxes in Mark.  Ambusher intrudes on Mark's space bubble and irrationally berates Mark for some imagined insult.

A left-handed person has a possibility of pulling CW and crabbing aft and away.  Right-hander must expose weapon more (cannot shield with body) and can get pushed into buildings pinning weapon.  Mugger at bad-breath distance.  I may be wrong on this, regarding advantage of right-vs-left.  You can be sure professional muggers have it figured out.

Mark is likely to be shot if he resists.

So what kind of sound trace might be captured?

The shooter will have the original "Boom" one or two milliseconds after bullet leaves the end of the barrel.

Two-to-six milliseconds later (depending upon the distance between the shooter and victim) his phone will experience the echo off the victim and impact of bullet.  The frequencies associated with odd-quarter multiples ( 1/4, 3/4, 5/4, 7/4...) of the distance between the shooter and victim will quickly attenuate.  The even-quarter multiples (2/4. 6/4, 8/4...) will attenuate much more slowly.

Approximately eight milliseconds later his phone will experience a sharp. full-spectrum, prolonged echo off the sidewalk (four milliseconds down, four milliseconds back up.).  If there is an accessory, a smaller, muddy echo will show up nine or ten milliseconds after bullet leaves barrel.

In the scenario drawn, a full-spectrum echo will show up at 20 milliseconds from the trash gondola.

Most buildings are not as "clean" as shown in the sketch.  They have inset windows and doors.  The 90 degree angles of the sills and sides are optimum for reflecting noise (higher frequency content) right back to the point of origin.

Depending upon how deeply the windows and doors are inset and width of architectural features, the higher frequency content (maybe +4000Hz) will echo back in waves.

In most circumstances, a running acoustical record in the 50 milliseconds following a near-gunshot event (mike saturation) would provide enough circumstantial evidence to identify most or all of the people-of-interest.

These predictions will be tested and reported in a future post.  However, I will be using a grain-bag mounted on a stepladder, my garage and a woodpile instead of an authentic urban environment as a test bed.

Is it possible?

Sure it is possible.  We had hardware and software that could hear a Russian picking his nose 400 meters below sea level.  And that was back in 1969.

And, to the best of my limited technical capabilities, I see no technological barrier that would prevent this kind of system's implementation on the streets of American when it becomes politically advantageous to do so.

Forewarned is forearmed.


  1. Very possible and it's 'done' other places...

  2. Thank-you for the critique. You have BTDT under trying conditions. That one sentence is a heck of a validation.

    Again, thanks.



Readers who are willing to comment make this a better blog. Civil dialog is a valuable thing.