The Issue of the Reliable Modem Escape Sequence Modems allow personal computers to communicate with other computers over telephone lines by transmitting data and files. There are two states of communication between an intelligent modem and the computer to which it is connected, the Command Mode and the Receive/Transmit Mode. The Command Mode allows that modem to receive instructions or commands from the computer, such as directing the modem to dial the phone, change modem settings, and hang up the telephone after a transmission is completed. The other state of communication is the Receive/Transmit Mode. It is during this state of operation that the modem sends or receives data or files which are exchanged between a personal computer and a remote computer. When the modem dials the telephone number and makes the connection, it goes from Command Mode to the Receive/Transmit Mode. The mechanism that causes the modem to go from the Receive/Transmit Mode back to the Command Mode is called an "Escape." The reliability of the escape mechanism is vital to the integrity of the system and the ability to predictably deliver the data or file being transmitted. When the escape occurs in the middle of a data transmission an aborted transmission results. This unintentional escape causes the interruption of the data flow, loss of time, increased telephone toll charges because the telephone line stays open even though no data is being transmitted, the inability of the software to use the modem until it has been manually reset, and most importantly, the file not being transferred. In computer jargon, the transmission "blows up"or "crashes" and the file transfer "aborts" or "bombs." For the personal user, this creates at the very least a significant inconvenience. Most users won't recognize the problem for what it is when it occurs. In businesses with large numbers of modems, this can create significant problems in operations for any company which relies on the transmission of data by modems to conduct its operations and make money. Preventing this inadvertent escape is a major design goal in reliable modem design. The Problem Prior to 1981, modems used a sequence of characters to escape from Receive/Transmit Mode to Command Mode. The state-of-the-art at the time used a series of control characters, sometimes "QUIT," to tell the modem to escape. Because these characters could appear in the data stream, this escape mechanism did not provide the user with a fail-safe method of escape. It is easy to see how a modem like this could inadvertently escape, creating just the problems described. The best idea at the time was to increase the length of the escape code and reduce the probability of it appearing inadvertently in the data. You could never be sure a file you were about to send would not contain the sequence of escape characters which would block the transmission. In order to alleviate this problem of inadvertent escape, it is necessary that the escape mechanism be transparent to data. That is, the possibility of the data alone triggering the escape should not exist. The Solution Was Found in 1981 Designing a modem completely transparent to data was the goal behind Dale Heatherington's efforts during the development of the first Hayes Smartmodem in the early 1980s. Dale was not satisfied with an escape mechanism which caused some data to be "unsendable." Dale's belief was that it was unacceptable to build a system where the modem was working as designed, i.e. "not broken," and the computer was not broken, and the software was not broken, but at some time, eventually, regardless of how improbable, the modem would fail to do its job. In effect, Dale redefined the problem as understood by the experts at that time and, by understanding the problem a different way, was on the path to a new solution. As part of his intensive research in the development of the original Hayes Smartmodem, Dale Heatherington solved this problem by surrounding the escape code, a sequence of characters, with guard times on both sides to alert the modem that the sequence is distinguished from a typical string of characters in a file transmission. This escape sequence eliminates the problems inherent in a data-dependent escape sequence because of its use of time and because it does not depend on the probability of character occurrence in a stream of data. It is virtually impossible for the Hayes escape sequence with guard time to appear in a file transfer and cause an unintentional escape using the common file transfer protocols. Computers just don't work that way. The problem of inadvertent escapes was solved. Dale Heatherington's invention led to the issuance of United States Patent Number 4,549,302, the Improved Escape Sequence With Guard Time, often called the Hayes '302 Patent, and corresponding patents in a number of countries. The Hayes '302 Patent ensures that modems escape or change to the Command Mode of operation reliably and without the possibility that data alone could ever trigger the escape. In over eleven years of use of the Hayes '302, Hayes has never received a complaint about an unintentional escape. In addition, this method was copied by almost everyone in the industry making it one of the most widely adopted defacto standards. With the Improved Escape Sequence With Guard Time you have reliable modem control. With reliable modem control, you don't have to know how to: o Understand and know how to modify file and document data formats. o Understand the details of how your communications software controls your modem. o Learn AT commands, syntax and escape sequence parameters. o Bypass your communications software and directly communicate with your modem. o Deal with serial cable pin outs and control signals. The New/Old Problem In September of 1991, another modem manufacturer requested that Hayes examine and test a particular modem to determine if the escape mechanism incorporated by the modem was covered by the Hayes '302 Patent. During the testing process, Hayes discovered that the modem incorporates a "new" escape sequence that is based purely upon a series of characters which appear in the data stream. This "new" escape mechanism is called Time Independent Escape Sequence or TIES. The name appears to derive from the way in which the escape sequence works because it does not make use of time as the Hayes '302 does. TIES depends entirely upon the appearance of the escape sequence in the stream of data being received by the modem. The TIES escape mechanism is very similar to the escape mechanism in use at the time of the invention of the Hayes '302 because an escape can be triggered by the data being sent as part of a file transfer. Historically, that old escape sequence was unreliable, and it is very likely that TIES will fall to a similar fate. TIES cannot be considered by anyone who understands it to be progress in the ongoing development of modem design. TIES - What Is It? The simplest escape sequence for TIES is "+++AT" where "+++" stands for any escape character and "" represents carriage return or the character set in the modem registers by the AT command set which designates the end of the command. Whenever that series of characters appears in the data stream, the modem "escapes" or changes from Receive/Transmit Mode to Command Mode of operation. In effect, what happens at that point in the transmission is that the flow of data stops! The flow of data has halted simply because the characters which make up the escape sequence have appeared in the data being transmitted. The appearance of these characters does not necessarily mean that you can pick up a file and read until you find the escape characters. The escape characters may appear as a mixture of file data and file transfer protocols which would not appear in printed text. In either event, the receipt of those characters by the modem will cause it to escape. Since software compatibility was an important criteria, the TIES escape sequence was selected to use the same sequence of characters that a computer sends to a Hayes modem to cause it to first escape and then execute a command. However, the TIES escape mechanism does not use guard times. The TIES modem appears to work with most existing software, but by disregarding the guard times the data transparency of the TIES modem is fatally flawed. In essence, the TIES escape sequence was designed to fool computers into thinking that they were communicating with a Hayes modem or a modem incorporating the Hayes '302 Patent, but TIES modems cannot fool the data being sent to prevent the TIES block from stopping the data transmission. TIES - What does this mean to you? Depending on where the escape characters appear in the data, the receipt of those characters by the modem could create unintentional escapes, prematurely halting the flow of data. For instance, if the file being transferred has been completely transferred and the TIES escape characters appear at the end of the data, there is no problem. The receipt by the modem of the escape sequence at that time would not interrupt the flow of data as all the data in that file has already been transmitted. However, if the TIES escape characters appear at any place during the transfer of a file other than at the end of the data, when those characters are received by the modem, the modem will escape, blocking the transmission of the file at the TIES block. At that time, the modem may need to be reset, and the transmission of that file must begin again. Every time the file reaches the point where the TIES escape characters occur, the modem will experience a TIES block and escape. As long as those characters appear in the data, at any point other than at the end of the data, the file can never be successfully transmitted through a TIES modem. No matter how many times you try to transmit the file containing those escape characters in the data stream, there is a 100% probability of a false escape. When the TIES block occurs, the typical modem user operating that modem will not be able to determine why the transmission of the data has stopped. Each and every time that the user attempts to send the file containing the escape sequence characters, the transmission of data will halt, and the user will be unable to determine the problem. In some instances, depending on the data following the inadvertent escape, the modem may actually have to be turned off and back on again before the modem will be able to operate. Further, a modem incorporating TIES cannot be repaired because it is not broken. There is an inherent limitation engineered into the product, i.e. the absolute inability to transmit the sequence of data that TIES defines as its escape sequence. The only practical solution to this problem of unreliability is to replace the TIES modem with one that does not incorporate the TIES escape sequence, unless of course you are willing to stand by the modem and reprogram the escape character and retransmit the data every time the TIES block occurs. Since many modems are used in unattended operation, this is completely unacceptable. And why would anyone who knows about this want to take the risk? TIES - Why You Haven't Heard About it? In December of 1991, Hayes Customer Service received a telephone call from a user of a modem manufactured by another modem company. This individual was using Hayes Smartcom software and was having a transmission-related problem. While attempting to assist this individual in solving this problem, Hayes learned that the modem was using the TIES escape sequence. This was the first time that Hayes had actual knowledge that the TIES escape sequence was being introduced into the marketplace. Manufacturers who are incorporating this technology have not been publicizing their use of TIES. There have been few articles written about this new escape sequence, and the majority of those articles are the result of Hayes expressions of concern over the reliability of TIES. Hayes was surprised to learn of the commercial implementation of this escape sequence because of those reliability questions. In fact, TIES appears to be the reincarnation of the problem that was solved by the invention of the Hayes '302 patented technology in 1981. Why are modem manufacturers not publicizing their use of this new technology? If there are no inherent reliability problems, why does it seem that modem manufacturers using this technology are hiding it and not touting its benefits from the rooftops? Why is it that modem manufacturers who use TIES want to keep consumers in the dark about their adoption of this new escape sequence? The answers to all these questions are rooted in financial concerns. Hayes, on the other hand, knows that consumers want to be completely and accurately informed about the reliability of the communications products they purchase. Modem manufacturers who use TIES should know that informed consumers will not knowingly want to buy products with TIES. The Philosophy of Reliable Modem Design Data alone should never cause a modem to escape. Users rely on the manufacturers of computer equipment and modems to build the equipment in such a way that users can count on its operation to conduct business and increase personal productivity. When the data being sent by a computer user can cause an interruption of use, the user does not get the performance expected. For a manufacturer to intentionally include a mechanism in a modem where data transmission alone can cause the escape is a great disservice to the computer industry, the public using the products, and eventually the manufacturer. The manufacturer should be concerned that dissatisfied customers who learn about this defect will at least have some serious questions about buying such a product. Once customers understand the issue they will no more want a modem that escapes when data is being sent than they would want a computer which miscalculates when processing a particular program sequence. The defect is simply unacceptable if there is a readily available, well-understood way to avoid the problem. Manufacturers who care about their responsibility to their customers will avoid products which contain deliberate defects since their products may create legal liability for that manufacturer. Development of the TIES Test Anticipating that there would be other telephone calls to Hayes Technical Support about the first known TIES modem and realizing that there may be other modems which incorporate this new escape sequence, Dr. John Copeland of Hayes was asked to formulate a test for use by Hayes Technical Support to determine if a modem being used by a caller was a modem incorporating the TIES escape mechanism. The test was easy to develop because TIES will cause a file to abort whenever a certain sequence of characters is present, and this sequence of characters is contained in the data stream. In fact, the characters themselves are nothing more than data. The test data file developed by Dr. Copeland includes all possible sequences of the form "xxxATy" (the TIES escape sequence). The character "x" can be any one of 128 different data characters. The character "y" similarly can be any one of 128 different data characters. This results in 16,384 possible sequences which are each repeated twice in the test file. The TIES test data file was developed only to determine whether a modem would escape on a TIES data sequence. Contrary to the statements of some members of the press, the file is not a software package. The test file is simply an ASCII data file to assist users in determining if their modem implements TIES. Some people have charged that the test data file is rigged. This is not a rigged test. It is necessary that the test contain all possible combinations of characters in order to be effective because some software packages you use may set the escape code to be of different values. Not "Hayes Compatible" Since the introduction and overwhelming success of the first Hayes Smartmodem, other modem manufacturers have touted their products as "Hayes compatible" and have assured purchasers that the modems they buy are "Hayes compatible." Consumers have come to rely on "Hayes compatibility" as a standard consideration in their purchases of communications equipment. Consumers who buy a TIES modem might assume that the modem is "Hayes compatible" because it uses AT commands, only to learn later that the modem might have been designed with a serious defect. TIES modems are not "Hayes compatible." Why Is This Issue So Important? Hayes believes that the introduction of TIES into the installed base of modems will seriously compromise the integrity of the industry. The degree of damage to the industry and to the confidence in the installed base of modems depends on the frequency that this problem will occur if the use of modems with TIES becomes widespread. There is no way of accurately predicting the probability of inadvertent escapes occurring in a TIES modem. One manufacturer of a chip set incorporating the TIES technology has admitted that when using a 9600 bps modem in continuous operation, there is a probability of an inadvertent escape every seven years while transmitting random data. The problem with these calculations is that data in the real world is never random. We know from research in data compression that data is not random. The fact the data is compressible demonstrates that it is not random. As a result the statistical probability is even greater that the ASCII characters used in the TIES escape sequence would occur in data. Further, once a byte sequence has special meaning, it appears much more frequently than would a truly random sequence of the same length. For example, Dr. Copeland scanned the hard disk in his office and found three files that would escape a TIES modem: a communications program EXE file, an Internet message discussing modems, and a log file from a telephone-line security program. Based on these results, the mean time to blockage at 9600 bps for a TIES modem would only be 14 hours for Dr. Copeland's data. The probability of an inadvertent escape may either increase or decrease depending on the data being transmitted. Probability figures are easily manipulated to reflect the outcome desired by the entity doing the calculations. For this reason, Hayes has not calculated the probability of inadvertent escape by a TIES modem, and we believe the calculations made by the modem chip manufacturers selling TIES in their chip sets represent the absolute minimum under the best conditions. For those who require an answer to the question of "how often," they can turn to the figures provided by the chip manufacturer itself, in the attempt to minimize the importance of this problem. Placing that calculation in terms of the Rreal world,S in a company using 700 TIES modems communicating at 9600 bps, there would be an inadvertent escape twice a week. Twice a week a data file important to the business would not be sent because it could not get through the TIES modem. Hayes maintains that any inadvertent escape is unacceptable. Two inadvertent escapes each week is outrageous! And if the data is being sent at 38.4 kbps, as many new modems are now capable of doing, this TIES block will occur more than once per day, only you cannot predict which of the systems will be hit. The following chart represents the minimum frequency of TIES blocks that you should expect with a TIES modem depending on the number of modems and the speed at which they are operating. TIES Impact on Business - From Personal Frustration to Corporate Chaos Fortunately, TIES blocks have yet to show up in great volume. However, as more and more TIES modems are sold, it is predictable that TIES blocks will start to occurr with ever-increasing frequency in businesses where numbers of modems exist. Personal Frustration. Take for example, your remote sales staff. What is the traveling salesperson going to do when he or she tries to log an order into the system at 9:00 at night and suffers a TIES block? Is that person going to attempt to adjust the RS232 serial cable pin outs or reconfigure the modem control from a hotel room? No, that is unreasonable to expect. Or say, for instance, sales support is attempting to send the monthly sales forecast spreadsheet to the regional offices. Only this time the transmission keeps crashing because one of the fields happens to include a number that when translated for transmission happens to be the TIES escape sequence. No matter how many times the transmission is attempted, it will never go through. Paralysis. TIES could also have a dramatic impact on entire computer systems that had incorporated software which is incompatible with the TIES escape sequence. Imagine the scenario where over the weekend your company upgrades its software using a program that includes some data that is incompatible with TIES. On Monday morning, when employees try to send material to remote offices of the corporation, the transmissions all fail. Your entire company will suffer a TIES block. Think of the time and money required to correct such a problem, not to mention the lost productivity while the problem is being found and fixed. Sabotage. The introduction of TIES into the installed base of modems has far-reaching consequences beyond TIES blocks. TIES is a threat to the security of your business. Its implementation makes it easier for an individual to sabotage the work of a company. If a company is using TIES modems to transmit data in its day-to-day operations, it would be very easy for a disgruntled employee to insert the TIES escape mechanism into the company's data. The employee could even change the software used by the company to insert this sequence of characters in every file being created at the company. Even more problematic is the ability of an employee to sabotage the employer's business by the insertion in spreadsheets of a few numbers which will cause the TIES modem to fail. The TIES escape sequence can appear in a transmission of either binary numbers or ASCII characters. There are many number combinations in data that will cause an inadvertent escape in a TIES modem. Chaos. TIES creates a window of opportunity for a new type of computer virus. In the last year, we have seen the business world shaken by the existence of a number of computer viruses that attack data directly in the computer. TIES offers illegal hackers the ability to attack the integrity of data by inserting a character string in unused parts of the data file to prevent its future transmission. Because TIES relies entirely on the existence of its escape characters in the data stream, it would be possible for a hacker to introduce a virus that would carry those escape characters into files without changing the program data itself. The file would be incapable of transmission because the TIES modem would inadvertently escape when it reached those characters every time it was transmitted. With this perspective in mind it is clear that TIES is a step backward that also could potentially create chaos in computer communications if it is deployed to any great extent. On many levels, TIES could actually affect productivity throughout the business community. There is an installed base of modems estimated at over 25 million. Assuming that these modems were TIES modems and using the probability calculations provided by the TIES chip manufacturer, there would be almost 3 million inadvertent escapes a year. Even assuming typical modem usage and a mixture of modem speeds, there would be more TIES blocks in one year than the number of people killed on U.S. highways. As modems with this inferior technology are introduced into the marketplace, their effect on productivity has the potential to be far-reaching. Keep in mind these calculations of probability are based upon what we believe to be the theoretical absolute minimum. Hayes fully believes that calculations using random data grossly underestimate the problem presented by the introduction of TIES modems into the installed base of computing equipment used around the world. Arguments for Acceptable Levels of Failure Because manufacturers have thus far refused to tell the public that they are using this new technology, it is very difficult to identify who is using TIES. However, the manufacturers of the chip sets necessary to implement TIES and the modem manufacturers who admit that they use this new escape sequence concede that it is not as reliable as the technology offered by the Hayes '302. There are repeated instances of these companies making statements to this effect. The question remains; what level of reliability is acceptable? Hayes believes that the industry should not accept any reduced level of performance if there is readily-available technology that offers proven reliability. Why is Hayes Interested In Educating the Public About Escape Sequence Reliabity? Hayes developed an education campaign targeted to corporate and business America because we know this is where the greatest damage is likely to occur if TIES control modems become widely installed. Without action the problem eliminated by Dale Heatherington in 1981 will be in the future of the modem industry. Data alone should not cause a transmission to abort! Modem manufacturers who have made the decision to begin implementing TIES have not informed the consuming public about TIES, their decision to use this escape mechanism, nor the inherent problems associated with the use of data alone as an escape signal. We believe that many of the manufacturers who are using TIES do not even understand the true nature of the problem and some just don't care and want the public to believe it doesn't matter. Since business, government, and industries increasingly rely on information technology infrastructure to conduct their business with applications like electronic mail, electronic data interchange for orders and invoicing, telecommuting to reduce traffic and fossil fuel consumption, and the rapid growth of information services for education and entertainment, it is becoming a regular part of our everyday lives. When major events occur that affect the infrastructure like the telephone system crashing in the Northeast U.S., the major scare over computer viruses, or similar threats to the use, availability or reliability of this strategic and tactically critical resource, the public confidence severely diminishes. Since its founding in 1978, Hayes has sought to educate consumers about many topics related to modems and their usage. TIES is just one more of these issues. Hayes has learned through the years that consumers want to know about these issues and appreciate any information provided that will enable them to make informed choices in their purchase of communications products. If someone wants to buy a modem with TIES, they should be in a position to be an informed consumer and should be prepared to deal with the consequences when the TIES block happens to them.