2) By making the cladding which surrounds the core very close to the core refractive index. The closer the two refractive indexes the better as any light that is less than a certain angle will be lost therefore only light that is at the right angle, preferably straight to the optical fibre, reaches the receiver.
I've been thinking about this problem today, and I found this page as part of a search for more information.
At first I thought that it was obvious why reducing the width of the fibre would reduce the multipath dispersion. But having tried the problem with maths, I'm not so sure.
I've tried to calculate the maximum possible path difference along a fibre of length l and width d that has a maximim transmission angle of [tex]\theta[/tex] to the normal of the core cladding boundary. I've come up with an expression for this path difference in terms of the length of the fibre... <Attached are my scribbles>
The problem with this formulae is that the path difference becomes a larger fraction of the total length of the fibre as the width of the fibre gets smaller.