Re: Bare fibers

[Rabid Wombat <wombat@mcfeely.bsfs.org>]

On Tue, 23 Jul 1996, jim bell wrote:

> At 03:20 AM 7/23/96 -0400, Rabid Wombat wrote:
> >
> >> 
> >> Doesn't that make it vulnerable (detectable) to Tempest attacks?
> 
> >No.
> >Transmitting light via fiber doesn't emit EM.
> >Anyway, the original post, as I recall, was about keeping sensitive data 
> >on a second hard drive, connected via (very thin, therefore harder to 
> >notice) fiber. Tempest monitoring was not a factor.
> 
> 
> It occurs to me that a bare fiber could actually be (randomly) hung across 
> treetops, roofs, power lines, and various other structures, over a 
> many-block distance in suburban areas.    Such a fiber wouldn't be protected 
> very well, but it would probably last a few months.  It would also be 
> exceedingly hard to find its terminations, and tracing it would be a real 
> pain.  (It probably wouldn't be visible against a bright sky more than a 
> meter or two away.)

It also would have little structural integrity - if you attached it to
trees, which sway in the wind, you'd have a broken fiber in a short time.
(The fiber doesn't even need to break, per se; microscopic cracking,
usually at the cladding, will ruin your fiber) (bird strikes would also be
a big problem, mostly for the bird)

Fiber optic cable usually has a kevlar sheath, and exterior aerial grade
fiber generally has a fiberglass rod inserted between the inner sheaths
and the exterior jacket. Water is also a factor - fiber buried in areas
where moisture is likely to be present (almost all applications) is
usually installed with a silicon gel between the interior jackets and the
exterior; water otherwise adheres to the exterior of the cladding, and
expands due to freezing. This causes fine fractures in the cladding, which
makes it more refractive - increasing chromatic dispersion, and therefore
a higher db loss on the cable. 

:)

ob crypto/privacy: Anybody have a good idea for detecting a tap on 
exterior fiber? I'd expect an attacker to have to interupt connectivity, 
terminate both ends of a break, and insert an active device. Thoughts?


> 
> Jim Bell
> jimbell@pacifier.com
>