Intelligent Inserter Control through File-Based Processing

By Bill Hart, Chief Systems Architect,
Pitney Bowes Production Mail and Software Systems

Once a mailing application grows to the point where every mailpiece is not constructed the same way, either in number of pages or selection of inserts, some form of "intelligent" inserter control is required so that each mailpiece can be uniquely processed. There are a large variety of control coding formats and control techniques used to implement intelligent inserter control, each with its own advantages and disadvantages. To sort through all of the details and marketing hype, it is essential to recognise that there are two key and separate ingredients to an intelligent inserter control system:

• How control information is printed and then scanned from the document, 1 D and 2D barcodes, OCR, OMR, and other scanning technologies.
• How the information is used to control the inserting process, either directly in a conventional system or indirectly by using electronic data in a file-base processing system.

How these two ingredients are selected and engineered to work together can make a significant difference in the performance and reliability of a mail processing system.

Coding Techniques
A variety of coding techniques have been used with inserting systems. Earlier techniques, such as OMR or "Optical Mark Reading", were developed uniquely for inserting applications, while others, such as Code 39 barcodes, have leveraged technology developed for other industries. Each coding technology provides a different set of trade-offs between capability, performance, cost, ease of application programming and impact on document design. The accompanying table summarises the differences between the various technologies.

In general, moving information automatically via paper is a difficult process. An electronic control system is much more at home reading from magnetic media than from a page. Variations in print quality, toner density, lighting, and paper motion can affect readability of all codes. As with the design of any communication system, it is important to pay attention to the coding design to ensure that it is robust and will continue to run well under the real world, non-ideal conditions of a production environment. Codes with built-in error detection capability, such as Code 39 barcode, can often offer superior performance.

One important aspect in designing a high performance scanning solution, is to keep the code as short as possible. Error rates increase geometrically with the length of the code string, so that doubling the number of characters will result in four times the error rate. It can be very difficult to keep scanning systems with large numbers of characters in the string running well. Choosing the right control method can significantly shorten the required code string.

Control Methods
In a traditional inserter control system the scan code information is used to directly control the inserting process. Thus the page count and feed selects are directly represented in the code on each page. The inserter merely has to read the data from the page and begin the processing. The inserter depends on being able to read the code on every page; an unscanned page usually has to be outsorted, often along with adjacent pages. For short codes this is generally not a problem, since error rates can be kept low. However, the more complex the inserting application becomes, the larger the code grows, creating an even greater error rate (remember the geometric error-rate problem). The code also occupies an even greater portion of the form's real estate, affecting appearance and wasting paper.

The alternative to direct control is file-based processing. The code string on the document is used to point to a larger information record in a data file that accompanies the mailing. This lets a relatively short code on the document (the Piece ID) point to a much larger data record.

Transmits Larger Quantities of Data
The most obvious advantage to file-based processing is that it provides a large amount of data to control the mailing process with only a small code on the page. The larger amount of data allows applications to be implemented that would be otherwise - impossible, such as in-line addressing. It is easy to provide include both delivery and return addresses in the data file so they can be printed on the envelope. Such an application would require several hundred bytes of data, more than could be communicated in any practical barcode.

An important feature of file-based processing is that the code on the page can be any of the Coding Technologies described previously. Because the code string can be short, any coding type (1 or 21) barcodes or OCR) will run more reliably. Coding selection is easier, becoming a tradeoff between code size, scanner cost, and application programming effort.

A basic file-based processing system stops here, just using the file to move a larger quantity of data than would be possible using scan codes alone. The Pitney Bowes Direct Connect- inserter control system uses the sequence information inherent in the file to significantly improve system reliability and integrity.

Scan Error Correction
Because the records in the Data File are in the same order as the documents, the control system always knows what page is coming next. It can "look ahead" in the data file to the next mailpiece and predict what will feed. This allows Direct Connect to correct for missing scan data, reducing dependence on high-read rate scanning. An exclusive Scan Sequence Verification algorithm ensures that the page sequence is reverified before any scan corrected piece is allowed to leave the inserter.

Eliminates Scanner Created Delays
Because it uses the scan data just to verify the correct processing sequence, a Direct Connect system does not have to wait for the scanner to finish to make a refeed decision. This allows the feeder to run at its maximum possible speed and lets the system designer place the scanner at a convenient location, where material presentation and operation access are optimised, rather than being forced to place it in a difficult location because of scanner timing restrictions.

Built-in Reconciliation
The Data File provides a built-in and highly accurate method of reconciling the entire mailrun. The Data File is essentially a list of every mailpiece that was generated by the application program that created the mailrun. The Direct Connect system checks off each piece from the list as it is processed, closing the loop between application and physical mail. This allows a regeneration list to be automatically created that includes both the damaged pieces and those pieces that never made it to the inserter, even if they were never printed! No longer does a mailer have to rely on operators finding all of the damaged mail to ensure it gets regenerated.

Because Direct Connect performs this reconciliation in real time, not as a post-processing batch exercise, it can support both realtime regeneration of damaged pieces, and separate run regeneration. In both cases, the regenerated mail can be run against the original Data File, ensuring a complete reconciliation of the total mailing process.

The real-time reconciliation capability of Direct Connect also provides tighter control on the mailroom process. If a mailrun is accidentally printed in duplicate, the duplicate run will be automatically detected as soon as it is processed on an inserter, even if the original run is currently being processed. Direct Connect also uses this capability to provide real-time late diverts, allowing the mailer to make a last minute decision to outsort a particular mailpiece. If that mailpiece hasn't been inserted, the system will automatically outsort it. Otherwise the system will indicate where to find it (by tray and piece numbers) in the out stream.

Automatic Resynchronisation
Mailing applications that require the matching of personalised documents from different print streams have traditionally been difficult to run productively. The manual resynchronisation process that must be performed whenever loading material or correcting a fault is both tedious and time consuming. Direct Connect can use the sequence information in the Data File to automatically correct mismatch situations, with no intervention from the operator. Direct Connect can also use the same technique to recognise and remove a printer backspace from any of the print streams, again without operator intervention.

Mailrun Planning
File-based processing also assists with facilities planning. The Data File can be prescreened to generate a "profile" of the mailrun, listing totals required for each insert, envelopes, and statement pages. This can be used to drive the material delivery and inventory management processes.

Observations
Just as the different document coding formats have trade-offs, the two control methods also offer trade-offs. A direct-control system does not require a network interface into the mail room and can be initially easier to setup. A file-based system can offer superior reliability, performance and accuracy, and can prove easier to operate because of the reduced dependence on scan accuracy. The table at the end of this article summarises these differences.

A variety of techniques have been proposed to replace a file-based system. For example a 21) barcode direct control system. Although 21) barcodes can move more information than a conventional 1 D barcode, they still rely on an accurate scan of each page to operate correctly. They do not enjoy the scan error correction and automatic reconciliation benefits of file-based processing. The extra programming effort and higher scanner cost makes these an unattractive solution when compared to a file based system driven by a more conventional code (barcode or OCR). A 21) barcode direct control system is not an alternative to file-based processing.

Some have tried improving the read accuracy of a direct-control system by printing the data for every page on two pages, so that each page contains its data as well as the data for the next page. The mathematics of scanning are against this one from the start. Since the number of characters to be scanned is doubled, the number of bad scans will go up four times! Although the adjacent scan error rate is theoretically lower, the systematic nature of most scanning errors eliminates this advantage. The net result is little if any improvement in performance with twice the visual impact.

OCR and file-based processing are a good match. OCR scanning has traditionally been expensive and unreliable. The scanners often exhibit substitution error (i.e. substituting an "S" for a "5") because of the inherent difficulty of having a machine recognise characters, as opposed to barcode symbols that are specially engineered to be easily by a machine read. A Direct Connect system can use its look-ahead error correction capability to easily compensate for OCR errors allowing a less expensive OCR scanner to be used. The OCR code format is normally compact and can easily be integrated into the total document design.

Many mailers have installed Direct Connect file-based processing systems, driven by the requirement for higher quality or data rich applications such as in-line addressing. However, as they have become familiar with the system they find that it is simply easier to use. The significantly reduced dependency on scanning quality makes both the inserters and the printers easier to operate. As more mailers become comfortable with network technology, they too will find that file-based processing is the control method of choice, because of its superior performance and quality.