guest@hyper.UUCP (guest) (03/02/88)
I think I can demonstrate the whole concept of spread spectrum stealthiness with the following simple BASIC program: Input "Value to be sent";V Input "Number of bands used";B Vx=V/B R=0 For J=1 to B R=R+Vx+(1-RND(0)*2) Next J Print "Value received =";R Print "Signal power in each band =";Vx End As you can see, the signal V has its energy distributed at B points with strength Vx. Vx is superimposed on the pre-existing random noise (here confined to a range of + and - 1) at each point J in B. As long as Vx is not too far below +and- 1, and there are sufficient bands, the random noise tends to cancel, leaving the original signal R, which is correlated here by simple addition. There is no need, however, to have the bands evenly spaced. In fact the spacing of the bands should be a changing function. Any other stealth xmitter would, in all probability, be using a different band spacing function at any given time. Thus there is no interference because there is no correlation. Correlation is the key to reception, interference and detection. So if the signal (in its discrete components) looks like noise, tastes like noise and feels like noise (to an uninformed receiver) then it must be noise. You can claim you have detected it, but I think not. John Logajan umn-cs!hyper!ns!logajan Network Systems Corp 7600 Boone Ave Brooklyn Park, MN 55428
paulf@umunhum.STANFORD.EDU (Paul Flaherty) (03/03/88)
I missed your reply to my first message, but it's clear that you don't have a fundamental grasp of Shannon's Law; Spread Spectrum raises the noise floor across the communication passband, and that increase in noise can be detected. The BASIC simuation that you've proposed is invalid for a number of reasons, not limited to: 1. The BASIC RND function is uniform; to model a radio channel, one needs Gaussian noise, plus impulse noise, plus Rayleigh fading on the signal. 2. No statistical test is proposed to separate the spreading code and the noise; however, even in the case that you have given, it can be shown statistically that the resultant is not uniformly distributed, and therefore a signal exists on top of the noise. Certainly, spread spectrum is less detectable than a narrowband signal, but to claim that it is next to impossible to detect is foolish. -=Paul Flaherty, N9FZX | Computer Systems Laboratory | "panic: getfs: bad magic" Stanford University | ->paulf@shasta.Stanford.EDU | --4.3bsd Unix