Fast temporal continuous scanning in RFID systems

Abstract

This paper studies the problem of temporal continuous scanning for large-scale RFID systems, which is an essential operation to keep the inventory up-to-date. The existing solutions need to execute unknown and missing tag identification protocols separately, which are of low time-efficiency because the unknown tags disturb the identification of missing tags and vice versa. To this end, we design a Fast Continuous Scanning (FCS) protocol based on the proposed multiple categories filter which can detect unknown tags and skip the empty and collision slots for improving the efficiency of missing tag identification. FCS is faster than the prior methods because the proposed filter is helpful to decreasing the interference between unknown and missing tags. We also investigate the optimization of the involved parameters to minimize the execution time. Extensive simulation results demonstrate that the proposed protocol outperforms the state-of-the-art solutions by saving 39% ∼ 58.5% of the execution time.

Introduction

Compared with the traditional barcode systems, Radio Frequency Identification (RFID) technology possesses noticeable benefits of proving storage and computation ability [1], [2], not requiring a line-of-sight and low deploying cost. Owing to these features, RFID techniques have been widely applied in many object monitoring scenarios [3], [4], [5], [6]. A typical RFID system is composed of a host, an RFID reader and a large number of tags. An RFID reader has a dedicated source with a certain computing ability. Following the commands sent from the host, the reader quires the tags and transmit the received data back to the host [7], [8]. A passive tag is a microchip with an antenna; it is powered by the electromagnetic induction from magnetic fields produced by the reader and sends its response back to answer the reader’s query.

This paper concentrates on temporal continuous scanning, which contains a series of scanning operations performed at different time instants with the goal of obtaining the latest tag list in the interrogation zone of reader. It is a basic operation in dynamical system where the set of tags frequently changes with the time. For example, suppose a sports retailer stocks tens of thousands of tag-attached products in a warehouse, where items are frequently put on the shelf or moved out for selling. To keep an accurate inventory of the products, we let the reader regularly performs the scanning operation to identify tags in current system.

A straightforward solution to temporal continuous scanning problem is to invoke the tag identification protocols [9], [10] at each time point, which is of low-efficiency because the reader recollects IDs of remaining tags which have been identified in previous scans. To avoids repetitive work, Sheng et al. design a pioneer two phases solution called CU [11], which only needs to collect IDs of unknown tags. However, distinguishing unknown and identified tags incurs additional time. When there are few remaining tags between two scans, the time spent in classifying unidentified tags will overwhelm the time saved in not collecting IDs of remaining tags [12]. To address problem, Liu et al. proposed Adaptive Continuous Scanning scheme (ACOS), which first estimates the number of remaining tags and begin to identify unknown tags only if there are many remaining tags. However, when there are few remaining tags between two scans, ACOS still has a poor performance due to the additional estimation overhead and the time spent for tag identification. To overcome this deficiency, Liu et al. proposed a spot scanning protocol called LOCK [13], which is optimized for spot continuous scanning with few remaining tags. Existing solutions to continuous scanning are time inefficient in solving the temporal continuous scanning problem because they commonly focus on the identification of unknown tags. However, in the temporal continuous scanning both the missing and unknown tags are equally important because we need both of them to obtain the latest tag list. Although, we can use these continuous scanning protocols together with missing tag identifications [14], [15], [16] to achieve temporal continuous scanning, no matter which protocol is performed at first, it is of low time-efficiency because the presence of unknown tags interferes with the missing tag identification, and vice versa.

We develop a Fast Continuous Scanning (FCS) protocol to efficiently distinguish missing, remaining and unknown tags between two temporals scans. FCS first apply the proposed Multiple Categories Filter (MCF) to filter unknown tags as well as inform remaining tags of their responding orders. Then, it starts a time frame, during which it combines the tag’s reply slot index and fingerprint to distinguish unidentified, remaining and missing tags. The reader sends acknowledgment to label the unknown tags. Finally, the reader starts another time frame to collect IDs from these labeled unknown tags. To minimize the execution time of FCS, we provide theoretical analyses on parameters optimization, during which we face two major technical challenges. The first technical challenge is to optimize the length of MCF, including the number of slots and the number of categories. During the classification phase, since the number of remaining tags shrinks faster than the number missing tags, we formulate and solve an optimization problem with the goal of minimizing the average time cost of each remaining tag. The second challenge is to get the number of unidentified unknown tags, which is required in the process optimizing the execution time of the second phase. We propose a zero-cost estimator for unknown tags by observing the slot status in the last time frame.

Our major contributions can be summarized as follows:

• We propose a fast temporal continuous scanning protocol for simultaneously identifying the unknown tags and missing tags in dynamical systems.

• We evaluate the effectiveness and performance of the proposed protocol through detailed theoretical analysis and provide the insights on how to optimize the execution time under different scenarios.

• We conduct extensive simulations to evaluate the performance of the proposed protocol by comparing with the state-of-the-art protocols side-by-side. The results demonstrate that FCS can reduce the scanning time cost by up to 58 percent in performing temporal scanning.

The rest of this paper is organized as follows. Section 2 reviews the related work. Section 3 presents the system model and problem formulation. Section 4 describes the FCS protocol for continuous scanning problem. Section 5 elaborates on how to minimize the execution time and how to control the probability of false identification. Section 6 presents the simulations of FCS and compares it with the related work. Finally, we conclude this paper in Section 7.