Interference modeling in a direct-sequence spread-spectrum packet radio network

A technique for characterizing multiuser interference and background noise in a direct-sequence spread-spectrum network is introduced, and packet error probabilities are calculated. The multiuser interference over a packet in the network is modeled as a compound Gaussian multivariate random variable for moderate to large values of the processing gain. The conditional variance is dependent on the number of users and their interference powers. The method works for any interference with statistics of the block interference having a spherically symmetric distribution. The best performance results, in terms of the expected total interference power, are obtained for the case of a large number of interferers with comparable interference powers. As the number of interferers approaches infinity, the performance is the same as that for Gaussian noise. For a small number of interferers, the block error probability curve is broader than that for Gaussian noise. For small values of SNR, the probability of error is smaller than that corresponding to Gaussian noise, and for large values of SNR it is larger. Modifications to the basic direct-sequence scheme that improve performance in the case of one strong interferer are suggested