Poster: HALL: High-accuracy and Low-cost RFID Localization in Large-scale environment

Abstract

Radio Frequency IDentification (RFID) localization systems are becoming increasingly coupled with various applications. Particularly, reference-tag assistant method with high-accuracy has become a mainstream. However, the main drawback of such systems requires to deploy a large number of reference tags, even including state-of-art work PinIt [1]. As a result, there exists serious tag collision, decreasing the identification throughput of reader, and massive reference tags in a large scale will aggravate this problem [2]. Further, most current methods have an underlying assumption that the signal is mainly propagating along the line-ofsight path, which is not true in most setups. In practice, the signals received by the antennas are the combination of various reflected signals. What’s more, despite the price of one tag is inexpensive (13 cents), to deploy large number of reference tags costs too expensive. For example, more than 50 universities in China have at least 3 million books in the libraries, and if we deploy the reference tags with distance of 15cm [1], the deployment cost will reach up to $0.78M. To deal with these problems, we introduces HALL which achieves both the high-accuracy and cost-efficiency in largescale environment. On one hand, by reducing the number of reference tags, the tag collision and deployment cost can be significantly reduced. On the other hand, we use the multipath profile to capture the multipath of signals [1], thus HALL is applicable in the non-line-of-sight environment. Thirdly, benefiting from the CS theory in sparse recovery, HALL acquires a high localization accuracy with measurements collected from only a few of reference tags. The key insight utilizing CS theory depends on the sparse property of the localization, i.e., the number of target tags to be located is much smaller than the number of locations.