Validate of RFID tool PROTOTYPE: Smart Shelf Case

Abstract

This paper is part of an RFID application for print cartridge inventory tracking and control. The objective is to present the technical feasibility of the Smart Shelf prototype through tests in order to achieve the best application performance in a production environment (i.e. all of the labels on the cartridges read). Six types of tags were tested on 120 tagged items equally distributed inside an intelligent platform prototype with four transmitting antennas located at the side walls and a commercial RFID reader. The test environment was built at RFID CoE Laboratory, unique EPCglobal Accredited Test Center in Brazil. The results indicated tags and antenna placements more appropriate for reading rate improvements in order to optimize application efficiency. It was concluded that the tests were necessary to map what is going on and how to improve the specific issues that interfere negatively in the reading rate numbers. There is a noticeable reading improvement derived from the preliminary results of the production environment testing.

1. Introduction

 The advent of RFID technology provides a greater degree of flexibility for system control/tracking and information dissemination. Wireless transmissions make data exchanges much more convenient and eliminate the existence of time (or lead time decrease) and spatial gaps between the plant and suppliers. Use of this technology has been developed and prototypes utilized in production and logistics management.

Recently, applications based in automatic replenishment RFID System have been adopted as a way to decrease out-of-stock among supply chain partners [7]; [8]; [9].

Besides, these solutions facilitate the integration in supply chains and reduction of bullwhip effect [5] and the enhancement of inventory replenishment management performance. In this case, the application of the RFID concept has been desired from the retailer to achieve just-in-time automatic replenishment upon the sale of a particular item.

The Smart Shelf solution turns out to be a smart solution for the management and inventory control, and optimizes the sale of cartridges. Human mistakes are reduced with RFID controlling the inventory and guiding the customer. The inclusion of RFID also streamlines the supply chain and marketing management by tracking sales, promotions and alerting suppliers when an inventory level has reached the replenishment level, avoiding item stock-out[8]. This paper adds to previous works, to improve shelf replenishment efficiency and prevent out-of-stock [1].

In this sense, tests were necessary to verify tags performance when applied to products stored in the Smart Shelf device. This was required due to the difficulties in the overall performance found on the first version of this solution (developed by another entity), where several challenges incurred such as low reading rates for the cabinet product capacity, equipment stability (especially regarding RFID Readers and PDA devices). These matters have been targeted throughout the project, in order to ensure the proper solution success.

The paper objective is to present these tests and give the best solution to implement the Smart Shelf in a production environment, through the types of tags indicated, their position on the products and the antennas positioned within the Smart Shelf used to achieve the 100 percent read rate.

2. Literature

Review Radio Frequency Identification (RFID) systems are one of the most anticipated technologies that will supposedly integrate high-speed processes across supply chain management [6]. In this sense, the inventory and information processing costs are considered as the major elements within the category of supply chain and logistics cost. It is a valuable reference for measuring the performance indicators in RFID-enabled supply chains [10].

The use of passive RFID devices for the tracking of pipe supports [4] and an application to test active RFID tag [3] was investigated. All of the problems with metal interference of the radio frequencies are common, as it pertains to read distance and kinds of material. Passive tags were used in the case described in this paper, the read distance, tag orientation created by the Smart Shelf, as well as cost, influenced the decision.

There are three important problems associated with tag detection in multiple tag RFID systems [2]: (a) accurately detecting RFID tags in the presence of reader interference (reader collision avoidance problem); (b) eliminating redundant tag reports by multiple readers (optimal tag reporting problem); and (c) minimizing redundant reports from multiple readers by identifying a minimal set of readers that cover all tags present in the system (optimal tag coverage problem).

Through experimentation, impacts and performance trade-offs of parameters including RF power, reading rate numbers and tag density were identified and utilized to create testing scenarios and hardware configurations.

3. Materials and Methods

The application is part of a partnership project between Hewlett-Packard and FIT, through the RFID CoE Laboratory, to implement a Smart Shelf for HP pen cartridges sales and inventory. The material used for the testing configuration was 6 RFID tag models from different manufacturers, 4 antennas and 1 RFID reader. The tag performance test results are shown through methods.

3.1 Material

The material used during the tests can be seen in Table 1.

Figure 1- Smart Shelf.

All the items were tagged with specific EPC numbers in order to facilitate the view of the reading map inside the shelf, locations were associated to an EPC number to allow for read coverage tracking.

3.2 Methods

The Smart Shelf works at RFID Brazil UHF band 1 (902 to 907,5 MHz) and all tests were conducted according to local regulations, using FHSS (Frequency Hopping Spread Spectrum) transmission technology and output power limited to 30 dBm.

The reading cycle lasted 2 seconds with 1 second delay between each reading cycle. Each reading cycle utilized all 4 antennas and cycled through them during the 2 second read cycle.

Duration testing was utilized to ensure the ability to maintain 100 percent reading in a production environment/condition. More than 5000 reading cycles were executed for each tag type.

In the beginning phases of testing tag placement was not defined and the tests were run with the tag located at two different positions on the unit under test. The positions will be described as position one and position two.

There were two tests rounds, each round presents some peculiarities: the first round was intended to compare results with other similar tests executed [1], after first round some possible improvements in the test methodology were detected and the changes were made.

The first test round was performed with five RFID tags, the tags are specified in this document as RFID Tag#1 to RFID Tag#5, Tag#6 was not available at the time. These tests were performed in position one and in position two using 27dBm power at the reader output.

The second round of tests was performed only with Tag#5 and Tag#6 due to the results of the first round tests and other specific reasons described in results. Only position one was tested, as position one became the defined tag placement for the unit under test. New software, developed by RFID CoE, was used to reduce the analysis time required in the former test procedure. This software highlights the areas in the Smart Shelf where reading was weakest allowing tester to quickly assess reading coverage and adjust accordingly. The tests were run at 27dBm and 30dBm reader output power.

4. Results

The results refer to tests realized in RFID CoE Laboratory. The results will be presented and compared separately due to the tests characteristics.

4.1 First Round

Table 2 shows the first round tests results. 

The first part (Table 2) refers the tag position in the cartridge internal flap. The second part shows the same tests, but with the tag applied at the cartridge top flap.

The main difference at this point is tag position and tag tested., Position two presented a better read rate in all cases.

Figure 2 resume the results of first round.

Figure 2 - Read Rate Round 1.

Although Tag#2 presented a good response it is not an option based on the tag's dimensions, it is too big to fit the application. Tag#3 and Tag#4 presented a low read rate.

Tag#5 had 100% of the tags considered as Detected tags. At the time of this test a tag that responded 50% of times was considered a detected tag, but 100% is the passing requirement. The results from this test derived the conclusion that Tag#1 and Tag#5 were the only tags that would be used in future testing.

Business strategies did not permit position two to be chosen as the tag placement for the unit under test although the tests results showed a better response from this position, so only position one will be used in test round two.

4.2 Second Round

Table 3 presents the second round tests results.

Tag#6 is a similar model of Tag#1, the main difference is in the manufacturing process, it is not possible to produce large amounts of Tag#1.

From this point on only the tags that had an acceptable performance in the first round of tests are being evaluated so it is acceptable to take the requirements to the level they deserve to be, 100% reading rate. Neither of the tags presented a perfect iteration.

Figure 3 facilitates the visualization that Tag#5 presented a better response in all the evaluated data and the number of tags read per iteration is almost two times the number of tags read per iteration when testing Tag#6.

Figure 3 - 27 dBm second round performance

The tags never detected also presented a significant difference, almost six times more tags were never read during the Tag#6 test when compared to the Tag#5 test. The perfectly detected tags bars show how far the tag is away from the 100% reading rate.

Figure 4 and Table 4 present the repeated testes with 30 dBm power at the reader output.

Figure 4 - 30 dBm second round performance

The results were better when using 30dBm power. The average number of tags read per iteration increased as well as the number of perfectly detected tags and the number of never detected tags decreased for both tags. It shows that the results although improved are not good enough. The best tag still has 6 tags that never replied, out of those 6 tags five were in a back location of the shelf.

During the tests, the two tags presented weak and non-reading spots located in different locations depending on the tag used. The areas requiring attention are those areas that did not respond or had few responses in both test cases. Most of the spots that were never or rarely read were in the back of the Shelf. It indicates that in order to improve the reading the shelf's antennas shall be slightly moved to the back.

5. Conclusion

Although the paper objective is not to present the application itself, or the Smart Shelf as a way of partners integration in the supply chain as inventory management, but to build scenarios to validate the application, the test results are indicative of improvement and care to be taken for its effective implementation in production.

The 120 RFID tags could not be read in 100% of the times when using the tags and the tag's positions described in this paper, to improve the readings the antennas placed inside the shelf should be adjusted to read the back of the Shelf or replaced with a different antenna designed to improve reading in the shelf. During the first test round the best tags were selected. Tag #1, Tag#2 and Tag#5 presented the best results. Tag#2 has an antenna design to big to fit the application so it was discarded and only the Tag#1 and Tag#5 were selected for future tests.

Tag#1 was switched to Tag#6, changing only the production process, because there were not enough Tags#1 to continue the tests. The second test round with Tag#5 and Tag#6 showed Tag#5 with almost two times the responding tags when comparing the average number of tags read per iteration and Tag#6 had almost 6 times more tags that were never detected.

When using more power the reading improved but no iteration was able to produce a 100% successful response. Tag#5 had the maximum number of tags read in a single iteration with 114 (six missing tags, five were located in a back location of the shelf).

When comparing the different tests results most of the spots without responses were located in the back of the shelf. Some adjustments must be made in order to improve those readings without compromising the current Smart Shelf front performance.

As future research, we intend to extend these results into the whole application. Preliminary results showed that the solution is viable, innovative to smart sales and important for integration in the chain. Tests and incremental developments are still being made to ensure reading of all units and to ensure that a store will have an intelligent solution counteracting the troublesome issues associated with human-based inventory tracking.

References

[1] BEZERRA, J. W. O.; MOREIRA, E. C.; SOUSA, M. F. A RFID Device to Improve Shelf Replenishment Efficiency. RFIDJournal. White paper. , acesso em 09 de Maio de 2009.

[2] CARBUNAR, B.; RAMANATHAN, M. K.; KOYUTÜRK, M.; JAGANNATHAN, S. GRAMA, A. Efficient tag detection in RFID systems, J. Parallel Distrib Comput. V. 69, p.180_196. 2009.

[3] GOODRUM, P. M.; MCLAREN, M. A.; DURFEE, A. The application of active radio frequency identification technology for tool tracking on construction job sites, Automation in Construction, v. 15, p.292 - 302. 2006.

[4] JASELSKIS, E. EL-MISALAMI, T. Implementing radio frequency identification in the construction process, ASCE Journal of Construction Engineering. Management, v. 129 (6), p. 680- 689. 2006.

[5] LEE, H.L., PADAMANABHAN, V., WHANG, S. Information distortion in a supply chain: The bullwhip effect. Management Science, 43 (4), 546-558. 1997.

[6] QIU, R. G.; TANG, Y.; XU, Q. Integration design of material flow management in an e-business manufacturing environment. Decision Support Systems, 42(2): p. 1104-1115. 2006.

[7] QIU, R. G. RFID-enabled automation in support of factory integration. Robotics and Computer-Integrated Manufacturing, v. 23,p. 677-683. 2007.

[8] REKIK, Y.; SAHIN, E.; DALLERY, Y. Analysis of the impact of the RFID technology on reducing product misplacement errors at retail stores. International Journal of Production Economics, 112(1), p.264 - 278. 2008.

[9] SAHIN E.; BUZACOTT J.; DALLERY Y. Analysis of a newsvendor which has errors in inventory data records. European Journal of Operational Research, 188(2), p. 370-389. 2008.

[10] WANG, S.-J.; LIU. S.-F.; WANG, W.-L. The simulated impact of RFID-enabled supply chain on pull-based inventory replenishment in TFT-LCD industry. International Journal Production Economics, v. 112, p. 570-586. 2008.

Autor: Paula Ruhnke Valério / José Geraldo Vidal Vieira (UFSCar)