Raspberry Pi RFID Client

At the University of Michigan (UM) we, the SPQR lab, are interested in pushing the limits of practical application within the passively powered computational RFID space¹. So much so, that a joint effort between the University of Massachusetts Amherst and UM has successfully extended the intel WISP, an open source computational RFID platform. The result, a powerful and robust open source RFID tag, dubbed UMass Moo (or moo hereon), that increases the maximum bound of computation and space by leveraging the energy efficient TI MSP430F2618 microcontroller and an altered board design supporting capacious off-chip flash memory.

Post fabrication and distribution of the moos, there have been a few unique -- yet, practical -- applications requiring the moo, an RFID reader, and a client to configure the readers operational specification (i.e., what to collect and how).  For example, at UM we are conducting an experiment that requires the moo to be completely submerged in concrete, inside a cinder block, and below the ground as part of a foundation. Our experimental method is to collect accelerometer and temperature data from within the foundation wall; which is completely inaccessible (with respect to physical access) once placed.

Intro the Raspberry Pi. As briefly mentioned above, we require a client to configure the readers operational specification. This client should be accessible, reliable, and deployable. In a construction environment, a laptop might not make much sense. With respect to long term data collection, a headless and unobtrusive device could easily be mounted out of sight. Also, the client should be energy efficient if it is to operate over large time deltas. Sound familiar? All of these requirements resonate well with the feature set of the Raspberry Pi.

Our approach for the client was, therefore, to utilize dual Raspberry Pis (for redundancy in the face storage failure) that could: establish TCP connections to numerous RFID readers; execute an ARM binary that defined our best known reader configuration and data collection methods; recover from power and antenna outages, and; support advanced network configurations to remotely adjust readers. As far as we can tell, there aren't many individuals whom are using Raspberry Pis for RFID reader clients and so we thought we would share our efforts with the community. This effort includes the: cross-compilation of the LLRP Toolkit (in C) to ARM, which enables the Pi to configure the RFID reader; validation of Pi performance as a reader in a practical application, and; a simple application with command line arguments for configuring a reader. All of which has been posted on the web and can be accessed here: a prior blog post.

^{1. I had intended on sending this write-up to the Raspberry Pi website maintainers as an example of rPi use in the academic setting. It never made it, but is documented here none-the-less. Hopefully it inspires someone.}