Object detection and classification based on RSSI

Hello,
I am using Panstamp NRG 2 for my project to detect and classify the objects that are interfere between communication path. Based on the variation of RSSI signal I must classify the objects that are detected. For that I should make RSSI signal sensitive such that it must be able to show the small variations caused by objects. I must classify objects like plastic, metals and wood etc. I tried using average filter and feedback filter to eliminate environmental noises but it also suppresses the variation caused by objected to be detected or it shows the same amount of signal attenuation for different objects (plastic and metal) which is not feasible for my experiment. Can somebody suggest me how to achieve this??

please ask me if I am not clear here.

Thanks in advance