Although the classification analysis based on the raw data set of PIR signals shows good recognition performance, other practical questions may need to be addressed. First, the memory
costs (i.e., the RAM size required) is important, which will be required to retain a set of classified examples for performing a supervised learning.