“…Interference models are an important analysis tool since such networks have become widespread and the need to predict their performance is obvious. Many of these models, however, estimate the probability of interference to be 1/N, where N is the total number of available hopping frequencies (Harris and Roberts, 2002;El-Hoiydi, 2001;Mazzenga et al, 2002;Mohamed and Pap, 1998;Nyberg et al, 2000;Popovski et al, 2003). This estimate of interference implicitly makes the following assumptions which may not be true for a particular network (and are certainly not true for Bluetooth): (1) frequencies are selected uniformly from the entire set of N available hopping frequencies, (2) frequency selection is independent (i.e., with replacement), (3) each network selects frequencies from the same set of frequencies (i.e., complete spectral overlap), and (4) each network makes its first frequency selection from this set at the same time (i.e., complete temporal overlap).…”