TECHNICAL REPORT APRIL 2009 RFID Trees: A Distributed RFID Tag Storage Infrastructure to Backtrack Hikers in a Forest

In this paper, we propose embedding RFID tags in trees in a forest to track hikers. Hikers are equipped with RFID readers, which read from and write to tags. Specifically, as a hiker moves through the forest, his/her reader leaves his/her ID and increasing sequence numbers (SNs) in tags. This creates a digital trail that allows the hiker to backtrack his/her route. That is, when the hiker decides to leave the forest, he/she scans for tags, following a path of tags with his/her ID and decreasing SNs. During backtracking, if a tag with a valid ID-SN pair is not nearby the current hiker’s location, he/she has to wander around until he/she does find such a tag. Therefore, to minimize the backtracking time, we wish to avoid physical gaps in a hiker’s digital trail, which are created if an ID is repeatedly deleted in tags near each other. These deletions occur because ID-SN pairs are overwritten in tags. That is, since tag memories are constrained, if there is no more space in a tag, a hiker leaving his/her ID-SN pair first deletes an existing IDSN pair, according to one of three algorithms. In Random Selection (RN), the hiker randomly chooses an existing IDSN pair to delete. In High Frequency Selection (HI), each hiker keeps a record of the frequencies of IDs it has seen thus far from previous tag encounters. The hiker chooses the ID-SN pair with the highest ID frequency to delete. High Frequency + Waiting Selection (HI+WT), is an extension of HI. After a hiker deletes an ID, that ID is immune from being deleted for a fixed number of consecutive tag encounters. Simulations indicate that HI+WT minimizes the distance travelled by a hiker during backtracking.