Worst-case analysis of direct-sequence spread-spectrum in multipath channels

We consider a direct-sequence spread-spectrum system operating in an indoors environment in the presence of multiaccess and multipath interference, and additive white Gaussian noise. We focus on the worst-case bit error probability of this system with a constraint on signal-to-interference ratio, and derive a Chernoff-type upper bound on this error probability. We evaluate the upper bound for a special case in order to gain understanding of the basic worst-case performance. We also compare the effects of the worst-case multipath interference with those of the worst-case multiuser interference of equivalent noise power, and observe that the worst-case performance under multipath interference is very similar to and only slightly worse than that under multiuser interference. We find out that the worst-case performance can be very good for a large number of chips per bit, whereas it is very poor for a smaller number of chips per bit, and for non-spread-spectrum systems.