Making Sense of RFID

Librarians everywhere are closely watching radio frequency identification (RFID) technology. An advance over using barcodes on library materials, RFID tags are being touted as a way radically to redesign how library materials are handled. From self-checkout to self-checkin to collection management, they reduce the need for staff intervention. With circulation up, budgets flat, and the pressure on to both maximize staff and improve customer service, RFID is seductive.

But it's expensive. The tags are vulnerable to wear and tear. The technology is not fully developed. It's unclear what sort of return on investment (ROI) it provides. And there are serious questions about privacy, with critics claiming that security flaws enable the surveillance of individuals and threaten the freedom to read.

Clearly, with such a complex technology, librarians need to do their homework and carefully weigh the pros and cons of RFID before deciding whether it makes sense for their library.

RFID can improve user self-checkin and checkout. This is because barcode readers require a line-of-sight (they use light or laser to scan the code). This limits checking to one item at a time and can make the process challenging. RFID scanners can read multiple items simultaneously. Anticollision algorithms ensure that each radio signal can be read accurately. Concurrent reads mean faster checking—no more lineups while someone aligns every piece of material—and improved customer satisfaction.

"There's a 'wow' factor, as well," says Margaret Hazel, principal technology librarian at the Eugene Public Library, OR. "The patrons think it's pretty darn cool that they can place items on a desk and see the titles appear on a screen, touch the screen, and leave the library."

Automated checkin is also a boon. Patrons get a receipt indicating the items returned. Hassling over fines can be reduced. RFID return bins can be configured to work with automated sorting bins. RFID also eliminates line-of-sight problems with this type of hardware.

Another benefit is faster, more accurate reshelving. Automated sorters can work with RFID to reduce the time and labor involved in rough-sorting returns. Fine-sorting can be improved with inventory wands, which can indicate books that are out of order. The wands beep to show a misplaced item and then show the area on the shelf to which to return the book. Most users configure the wand to show the proper place within four or five books. Finer levels of sorting are possible, but in practice the finest sorting requires longer read times.

Labor savings can lead to another benefit: more public service. Staff who would be at the circulation desk can be reassigned to support patrons in other activities. The University of Connecticut Libraries used RFID to help consolidate service points and reallocate staff, says Peter Murray, assistant to the director for technology initiatives. But evaluating whether more public service equals better public service can be tricky.

Inventory management is improved with RFID. Inventory, time-consuming and costly, often heads a library's "to do" list. Once automated with an inventory wand, it can be done more frequently. Robert Ferrari, stacks manager at California State University, Long Beach, says they had never performed an extensive inventory prior to having RFID. Now he inventories 5000 books per hour. The first time they did a partial inventory, Ferrari found 300 items they had recorded as lost or missing. The University of Nevada, Las Vegas (UNLV) has reported saving $40,000 in replacement costs for the 500 "lost" items it found after tagging its 600,00-plus collection. Frequent inventory and shelf reading—and more accurate holdings information—could increase patron satisfaction. "We use four [wand readers] and shelf read the collection about every six weeks," says Zep Sylvis, UNLV's stacks manager.

Improved inventory management can also benefit collection development. Currently in-house usage is measured with ad hoc procedures like marking ticks in the cover of an item when it's reshelved. With RFID, in-house usage is more easily monitored with the magic wands, and in-house usage is available for electronic report generation. Sylvis says UNLV has recently experimented with grouping in-house measures by subject area.

Another significant advantage is staff safety. According to a report commissioned by the San Francisco Public Library (SFPL), its circulation staff members perform 768 risky motions each hour while checking out items conventionally. The number of motions for checking in and sorting items is, respectively, 575 and 940 per hour. These motions include grasping, reaching, and lifting. SFPL spent $265,000 in direct costs for repetitive strain injuries over the past three years. Injured employees lost 260 days of work and had to have 500 hours of work modified to accommodate their injuries. SFPL anticipates that RFID will substantially reduce the risky motions and mitigate worker compensation expenses.

The most prominent negative issue swirling about RFID is privacy. There was little debate about privacy issues in early library installations, although consumer privacy groups had already sounded the alarm about RFID use in retail. The controversy exploded in the library world in 2003 when the SFPL Commission received objections at its hearings to consider RFID.

There are many ways in which library RFID tags are vulnerable to security risks. David Molnar and David Wagner, electrical engineers at the University of California, Berkeley, specialize in the security issues in library applications of RFID. Their recent article (see LinkList, p. 14) looks at specific threats to library RFID architectures.

Best practices have been developed by the Berkeley Public Library, CA, and the Ontario Privacy Commissioner. Among many recommendations, these practices include notifying patrons that RFID is being used, designating a responsible staff person to oversee security measures, and regularly auditing the privacy controls. Other best practices are in development by committees of the California Library Association and the Public Library Association.

The likelihood of a breach in library RFID security remains unknown, but the issues regarding tag security will continue to be discussed. While the details of the debate are beyond the scope of this article, any librarian considering RFID must be aware of privacy as a potential drawback. Privacy vulnerabilities exist at the tag level. Different tag standards have different susceptibilities. Understand the technology and its implications.

There are other disadvantages, too. Topping the list is cost. Book tags cost between 40¢ and 70¢ per tag. Media tags run about $1 each. The prices of tags are coming down, but library tags will not decrease as fast as those used in warehouse applications. Those tags work on different frequencies and are meant for use at the group level rather than the short-range item level.

A library with 100,000 books and 5000 media items would spend $45,000 on tags alone. This is for plain tags. Customized tags, with logos, for instance, will cost more. Hardware costs in the thousands, sometimes tens of thousands of dollars. For example, self-check machines can cost nearly $20,000, and inventory wands can cost between $4000 and $8000. Some vendors require the additional purchase of a computer server, or special software. None of the prices include ongoing maintenance contracts (between eight and 15 percent of the hardware price). There is also the cost of conversion to consider; either staff will have to reallocate their time from other functions or temporary staff must be hired.

Obvious tags on items are exposed to vandalism. "I'd love to know how our patrons have figured out how to get the tag out of the book without being seen," says Toni Oberman, head of circulation at Skokie Public Library, IL. "We find them in the rest rooms and lying on the shelves. It is not easy to remove them." Customized logos make the tags look like book plates and are less likely to be removed. However, vandalism can occur with other electromagnetic systems. No theft detection system is foolproof.

More minor is the problem of the RFID reader responsiveness. The accuracy declines as more items are read, and some can be missed altogether. This is because some RFID systems are based on a "reader talks first" type of communication, says Shai Robkin, president of Vernon Technology Solutions, an RFID vendor. "The security gates are attempting to read every tag in the field in order to determine if the designated security bit is either on or off, or to read the entire item identification number. Performance will suffer as there are more tags in the field. This degradation may not be apparent when there are only one or two tags in the field, but when you have five or more, the drop in detection is considerable, to the 70 percent range."

The main standards pertaining to library RFID are SIP2 and ISO 15693 and 18000-3. SIP stands for Standard Interchange Protocol. SIP2 manages the communication between the RFID system and the automated library system. All RFID vendors are SIP2 compliant, but some tweaking may be needed for things to work smoothly.

Vendors have added proprietary modifications to the protocol. Librarians have to investigate how well their vendor plays with their particular library system. The deficiencies of SIP2 led the National Information Standards Organization (NISO) to develop a protocol to encourage better interoperability. The result was Z39.83-2002, the National Circulation Interchange Protocol (NCIP). This standard is not yet implemented by all library RFID vendors.

The International Organization for Standardization (ISO) standards are used for communication between tags and tag readers. ISO 15693 was originally designed for access control cards. It defines the physical characteristics, air interface, and communication protocol but doesn't "account for what data is on the chip, how that data is formatted, or the read/write requirements of the RFID equipment." The ISO 18000 series of standards establishes RFID-specific communication protocols. The ISO 18000-3 standard will be employed with the 13.56 MHz RFID tags used in libraries.

While all of the RFID vendors are ISO 15693 compliant, there is no interoperability among them because proprietary protocols are added to the tags. ISO 18000-3 was recently approved and will be published in fall 2004. It should improve the ability to read a variety of vendor tags with one reader.

As those who experienced barcoding in its early days can attest, this is an issue for forward compatibility and interlibrary operations. If a vendor goes out of business or out of the library market, then it is difficult to recoup the investment if another vendor's tags or equipment won't interoperate. This will also be an issue for sharing materials among libraries. There may be a need to use barcode and RFID until the industry achieves mutual compatibility. Ask vendors if their equipment will read all tags and if their tags can be read by other vendors' machines.

Also ask vendors if they currently use ISO 15693 and plan to migrate to ISO 18000-3. These standards also impact the security of the RFID system. For example, the ISO 18000 standards specify how signal collision avoidance works, how data is written to the tags, and whether that data can be write protected. ISO 18000-3 comes in three different modes, which perform these tasks differently with different security vulnerabilities. Librarians cannot require vendors to address the threats unless they understand the privacy implications of the standards upon which the solution is based.

Deciding to implement RFID raises a new set of tasks: setting goals, evaluating other installations, selecting a vendor, planning conversion, managing the actual conversion, and assessing the results.

What do you want to achieve with RFID? "As part of our planning process, we established our criteria for the ideal security/self-checkout system," says Karen Saunders, assistant city librarian, Santa Clara City Library, CA. "The system had to simplify the checkin and checkout process for staff, increase efficiency, and minimize repetitive motion. The self- checkout component had to be easy for patrons to use. Magnetic media such as videos and DVDs needed to be secure, but we wanted to be able to handle the discharge function the same way as books."

Creating operational objectives from these ambitions is a next step. For example, RFID can accommodate multiple simultaneous checkouts. One objective could be to have the majority of your circulation done via self-check. This goal was set and achieved at the Cerritos Public Library, CA, which has 3000 people passing through its gates daily and an annual circulation of one million items. Sixty-two percent of its circulation is now done via self-check, says Ismael Arrayga, Cerritos's information technology supervisor. Goals are useful for system evaluation as they provide a way of comparing requirements with outcomes.

Speak with others experienced with RFID and "test-drive" their systems. See "Do Your Research! Questions for Colleagues," page 6, for ideas. Find out why they selected their vendor and their implementation process. Ask about how the system works in-practice. How was the customer service?

Invite vendors to make a presentation. Find out equipment costs, the range of readers, and the read response times. What type of tags do they use? If they don't provide all equipment, who do they partner with? Find out what differentiates the vendors from one another. Develop a checklist of your requirements and compare each solution using a chart. Detailed questions are available in "Do Your Research! Questions for Vendors," below.

The physical retrofitting of library materials is generally simple. Existing barcodes are scanned into a conversion station and programmed RFID tags result. Choosing a conversion workflow process is less simple. Some libraries tag all of their new books, then program the tags. Others start with subsections of their collection in order to get a feel for the process, become accustomed to the equipment, and field test operations.

The type of equipment you purchase will influence the conversion plan. For example, a combined self-check conversion machine would make it possible to bypass a formal conversion process altogether. The conversion can happen as patrons check out materials, and the most used components of the collection will receive the fastest conversion. After a set period of time, say a year, run a report to see what remains to be tagged and develop a staff workflow to accomplish it. It's possible to do conversion in the stacks using portable stations.

When it comes to the actual conversion, plan your staff training. Does your vendor provide a training session? Is there sufficient documentation? How will you manage quality control? One librarian found it necessary to hire temporary staff when student conversions had too many errors. Then there is allocating staff time to the project. Do you want to dedicate some staff doing conversion 100 percent time, or will you want to have a teams working in shifts? Shifts are probably better for avoiding repetitive strain injuries and maintaining quality control.

Equipment needs to be maintained after installation. As standards change, hardware and software might need upgrading. Regular updates should be included in your maintenance contract with the vendor. Other maintenance tasks include setting up a workflow for new items received by the library and performing evaluations of the RFID systems' effectiveness.

No published studies yet exist presenting quantifiable evidence that RFID provides greater gains than expenditures. This is partly owing to the limited number of libraries that have implemented the technology. Anecdotal evidence about savings is compelling. RFID vendor Bibliotheca says, "Mastics-Moriches Community Library, NY, carried out a time and cost analysis to compare the use of Electronic Article Surveillance (EAS) and RFID systems for circulation and found that the latter gave a saving in labor time of around 85 percent."

Space requirements bring different complications. If a library decides to implement an automated return or a sorting system, they will require space allocations and a power supply. This might demand renovations to public service areas, an additional expense.

Rollout is the time to track the effects of RFID. Gather the pertinent information to compare the library operations before and after implementation. This simple comparison should provide a rough idea of whether the library has benefited from the conversion.

Calculating a precise ROI may prove difficult owing to the intangible nature of some of the factors. ROI is typically defined as costs relative to profit generated. The high cost of purchase and the effort required to implement should provide the justification for attempting ROI as a matter of accountability. The difficulty is what to measure and how to quantify value.

ROI is measured as a percentage of investment over time. Three years is a good time frame for evaluating information technology projects. Concrete, measurable benefits may include those advantages touted during the consideration phase. Circulation statistics provide hard numbers as do reshelving costs. Inventory is another opportunity. For more ideas on calculating ROI, see "An ROI Laundry List," p. 8.

Other changes are difficult to quantify. How do you assign a figure to missed opportunity? There is no way to measure how many customers simply gave up when they couldn't find an item on the shelf prior to RFID.

Choosing and using RFID requires detailed effort, from initial investigation and vendor selection through planning and implementing the conversion to ongoing maintenance and evaluation. As with anything new, RFID is a risk. By doing your homework at least your risk is calculated.