Accuracy evaluation analysis for indoor measurement-based radio-wave-propagation predictions

An accuracy evaluation analysis of in-building path loss prediction models is presented, comparing the performance of measurement-based prediction against traditional indoor radio propagation models, such as Keenan-Motley, tuned with measurements. The analysis was conducted using extensive walk test data from various antennas at 1800 MHz collected in two types of buildings: an old Victorian manor-house built from brick and a modern purpose-built office building with open-plan construction. The measured data were split into two parts: a tuning set and an evaluation set, such that the data used to run the measurement-based prediction were excluded from the data used for the accuracy evaluation analysis. The results show a significant improvement in the prediction accuracy for the indoor measurement-based model over the tuned Keenan-Motley model.