Efficient Rostering of Mobile Nodes in Intermittently Connected Passive RFID Networks

This paper focuses on the problem of rostering in intermittently connected passive RFID networks. It aims to report a list of tagged mobile nodes that appear in given interested area(s) and time interval(s). Such rostering faces several unique challenges. First, the network consists of two dramatically different types of nodes: powerful static readers and extremely resource-constrained mobile tags. Communication can be established from a reader to a tag only, but not tags to tags or readers to readers. Therefore, the connectivity is very low and intermittent. Besides connectivity, the tag's computation power is also intermittent. It is available only for a short interval when the tag is powered up by a nearby reader, rendering any continuous functions impossible. Moreover, the capacity of tags is so limited that it becomes the critical network resource and communication bottleneck. To address the above challenges, we propose a rostering algorithm that employs a dynamic space-efficient coding scheme to construct hypothetic packet candidates, appraises their values according to information redundancy and tag mobility, and establishes a 0-1 Knapsack model to choose the best set of packets, which together maximize their total (redundancy-excluded) value, but do not exceed the capacity of a tag. We carry out experiments that involve 38 volunteers for nine days and perform large-scale simulations to evaluate the proposed rostering scheme.