Accuracy of temporal fade margin prediction for fixed wireless links exposed to urban traffic

The authors describe an empirical study to determine the accuracy of Ricean and Nakagami-m distributions for the prediction of fade margins, for 90 and 99% of temporal availability in fixed wireless links. They compare the results obtained assuming an envelope distribution (based on Ricean and Nakagami-m models) and those observed from measured data. Their results suggest that fade depths, for individual users predicted using a Ricean distribution, have a relatively small risk of being exceeded in practice. In contrast, a Nakagami distribution is somewhat `optimistic` regarding the extra power that will in practice be required by a specific user. They also present a more systems-oriented approach, where instead of evaluating the modelling accuracy for a single terminal, they consider a collection of users in a spatial region. This is relevant for sensor networks or relay-assisted infrastructures. They found that the larger than predicted temporal fades encountered in some areas will to a large extent be compensated by those that are in practice smaller than expected. Thus, in this case, the average margin that needs to be considered is similar to that predicted by either distribution.