This thesis presents results of average and impulsive noise measurements inside five office buildings and retail stores. Measurements were made at 918 MHz, 2.44 GHz, and 4.0 GHz using a superheterodyne receiver with 70 dB dynamic range and a 3-dB RF bandwidth of 40 MHz. Omni-directional and directional antennas were used to investigate the characteristics and sources of radio frequency noise in indoor channels. Statistical analyses of the measured data are presented in the form of amplitude probability distributions, pulse duration distributions, pulse spacing distributions, and noise factor distributions. Simple mathematical models of these statistical characterizations are also presented. The measurements and analyses indicate devices with electromechanical switches (copy machines, microwave ovens, printers, and electric motors) are principal sources of impulsive noise in retail and office environments. The 918 MHz band was consistently the worst band throughout the measurement campaign. This is attributed to higher path losses at 2.44 GHz and 4.0 GHz, and to adjacent and cochannel interference from users near the 902-928 ISM band. Pulse duration statistics indicate that no significant differences exist between impulse durations in the measured bands. This suggests that impulsive noise inside buildings is very wideband, and that pulse durations are directly a function of the receIver bandwidth. Pulse spacing statistics also indicate that intervals between consecutive impulses are similar in each frequency band. This thesis developed a computer simulation algorithm to create sequences of impulsive noise events which have statistical distributions similar to measured data. The statistical results for simulated impulsive noise are compared to measured distributions to illustrate the accuracy of the simulation algorithm. / Master of Science