Since the SURF project titled "Indoor localization based on Wi-Fi fingerprinting with deep learning and fuzzy sets" in 2017, we have been investigating large-scale multi-building and multi-floor indoor localization based on a single dataset for received signal strength indicators (RSSIs) and deep neural network (DNN) models for the integrated estimation of building, floor, and location with focus on the scalability of a DNN model and its outputs. In this project, we focus on inputs to a DNN model and study the scalable representation of RSSIs for DNN-based large-scale multi-building and multi-floor indoor localization.