Intra Spacecraft RFID Localization

In the related paper we explore two machine learning approaches to improve RFID tag localization in the highly reflective environment imposed by the International Space Station. We propose P-RFIDNet (Passive RFID RNN), a residual neural network (He, et al., 2015) [1] for localizing passive RFID tags in high multipath environments with fixed antennas. Furthermore, we show how transfer learning can be used to generalize P-RFIDNet to new RFID environments with limited training data. In addition to P-RFIDNet, we present REALMRFC, a random forest (Breiman,2001) [2] model with feature engineering performed by an RFID localization expert. We benchmark P-RFIDNet and REALMRFC using data from the RFID Enabled Autonomous Logistics Management (REALM) RFID system on the International Space Station (ISS). In this poster we hope to convey the challenges of localization within spacecraft and discuss the techniques we've used to model the data. We'll break down the data environment and discuss the modeling techniques used by P-RFIDNet and REALMRFC.