ii We examine the existing methods for evaluating the distribution of the sum of lognormal random variables, focusing on closed-form results. We find that there are no results in literature that are both simple and accurate. We then derive a new closed-form expression for the lower tail of the distribution, and use it to construct a new method using a power-lognormal distribution. We apply both basic moment-matching and our new method the problem of the total interference power in a cellular system. For both methods, we derive equations that find the interference distribution essentially in closed form, using minimal numerical integration. We apply both methods to the uplink and downlink in systems with and without power control, for various cellular layouts, channel models and user activity probability. We compare distributions obtained by Monte-Carlo simulation directly with those obtained by our method, and find very good matches in many useful cases. iii Acknowledgements

Abstract not found