Classification and Discretization of Shadowing Toward Low Storage Radio Map

This paper considers the shadowing classifier and shadowing discretization to create a low storage radio map. We have confirmed that the data size of the radio map can be reduced while high estimation accuracy keeps via the former method. However, the data size has not been evaluated using the latter method that simply discretizes a shadowing realization in each position. Thus, this paper discusses the effectiveness of the two methods in terms of the data size and estimation accuracy. The verification results show that the shadowing classifier can skillfully predict the radio propagation compared with the discretization-based method. Meanwhile, if the slight degradation of the estimation accuracy is tolerated, the discretization-based method may be useful.