Introduction

The ability of the electronic industry to continue to produce more powerful and cheaper products every two years or less depends on a strong and competitive Electronic Design Automation EDA sector. In the last few years the EDA industry has suffered through a number of legal battles. EDA vendors have sued each other for either copyright violations stemming from the theft of software codes, or for patents infringements. Although companies must protect their Intellectual Property (IP), the dynamics of the EDA industry and the nature of its products make it difficult to do so under existing patent law and business practices.

Current IP protection methods in EDA

Companies doing business in the United States can protect their IP in three ways, depending on the nature of the property. Patents protect inventions and improvements to existing inventions. Trademarks protect words, names, symbols, or devices or any combination thereof, used or intended to be used in commerce to identify and distinguish goods. Copyrights protect literary, artistic, and musical works. In addition trade secrets are an efficient way to retain a competitive advantage in the market. Utility patents protect how an object functions. EDA companies obtain utility patents when they want to protect a software product from the competition. According to the U.S. Patent and Trademark Office (USPTO) a utility patent may be granted to anyone who invents or discovers any new and useful process, machine, article of manufacture, composition of matter, or any useful improvement thereof. The patent grant excludes others from making, using, or selling the invention in the United States. A utility patent is valid for a maximum of twenty years.

EDA companies almost never use trademarks to protect their products, although many of their marketing programs result in trademarks to protect brand names. Although the definition of what is protected by copyright predates the introduction of programmable computers, the Copyright office has remedied the situation by stipulating that computer programs and most “compilations” may be registered as literary works. A work created on or after January 1, 1978 is protected for the author’s life plus an additional 70 years after the author’s death. In 1980 software programs received copyright protection. Although a copyright protects the author from plagiarism, it does not protect the concept, idea, or even the story line. In addition seventy years is too long a time in the EDA industry, therefore the use of copyrights to protect software programs is rare.

Shortcoming of patents as an IP protection method

Patents in the electronic field constitute a gray area. The US Government codified most of the definition of what can be patented and what constitutes a patent before the invention of the computer and software programs. Thus the courts have had to adapt the existing definition to cover software. In order for a software program to be a candidate for patent protection it had to be defined as a process. Gregory J. Kirsh in an article first published on www.GigaLaw.com (Reference 1) states that: “Software operating on a computer causes the computer to perform a process, and the process can usually be represented by any one of a multitude of different (even if functionally equivalent) sequences of software code.” Mr. Kirsch goes on to justify the use of patents to protect the software on this basis. Two things are important in this statement. Firstly, since the software causes the execution of a process, it falls within the definition of a patentable invention. Secondly other developers can reproduce the same process by developing equivalent but different source code. If a second programmer can reproduce the same process by independently developing a different but equivalent computer program, then any computer scientist will agree that the neither program can be patented since they do not uniquely represent a process. Or, to take the other side of the argument, both can receive a patent. This latter argument would mean that every computer program, by its own nature, would receive a patent upon its release without having to apply to the USPTO.

Webster Dictionary Unabridged Second Edition defines a process as: ”a particular method of doing something, generally involving a number of steps or operations”. Dictionary.com adds that in Computer Science a process is: “a running software program or other computing operation. A part of a running software program or other computing operation that does a single task.”

A software program is an implementation of a formula or an algorithm. It performs a series of logical operations that implement the algorithm required to solve a specific problem given a unique set of requirements. Thus it can be called a process. But the Computer Science definition carefully limits the use of the word process to the dynamic state of the program: that is it is a process when it is executing. No one can develop a useful software program without a formal and precise definition of scope and intent. In other words one must know what problem must be solved, what are the processes that must be executed, and the sequence in which the execution must take place. Richard Sietmann points this out very eloquently in his Letter to the Editor in IEEE Spectrum, a publication of the Institute of Electrical and Electronic Engineers (IEEE), the largest worldwide association of professionals in the electronics and computer fields (Reference 2). Clearly to computer scientists the description of the algorithm and its associated requirements do not constitute a process.

The U.S. Supreme Court in 1981 in its decision in the Diamond v. Diehr held that computer software was patentable provided that any claims to the software product were not merely a procedure for solving a mathematical formula. As any computer scientist can attest, the purpose of a computer program is exactly a procedure to implement the solution of a mathematical or logical problem. Congress intended that a patent would protect a unique implementation of an idea or a formula that uniquely described a natural transformation.

As an increasing number of electronic products contain software, the entire electronic industry is facing ambiguities when dealing with patents. Arguably the ongoing case with the greatest possible financial implication in this industry is the legal battle between RIM, the producer of the Blackberry personal digital assistant, and a very small company that specializes in holding patents, NTP, Inc. The latter holds patents that arguably were infringed by RIM in its wireless e-mail application. In spite of the fact that the USPTO has reviewed those patents and preliminarily rejected them, the fact that the patents had been granted previously, probably by mistake, is forcing the two companies into a very expensive legal battle.

The character of the EDA industry is changing. From an industry where almost all of the vendors were based in the United States to one that has seen a significant number of new companies started in the last few years in Europe, including the ex East European satellites of the Soviet Union, India, China, and Japan. The process for granting and challenging a patent is unique to each region. Thus, an American company challenging a patent granted in Europe, for example, would find the process quite different than what US courts follow. Steven J. Frank, a partner in the law firm of Goodwin Procter LLP, in Boston describes the differences in some details in a recent article in the IEEE Spectrum (Reference 3).

Impact of patents on the EDA industry

The EDA market is highly segmented by narrowly defined application areas. Although the entire industry only generated slightly more than $4 billion in 2004, research firm Dataquest identified over fifty unique market segments. Such fragmentation allows one company, especially one of the largest four who together account for over 75% of total revenues, to control one or more segment by patenting the fundamental algorithm addressing that particular design or analysis problem. Customers are hurt in two ways by this state of affair: they are most likely to pay higher prices and the product does not improve as when it must compete with similar tools. In this case the patent restrict instead of promoting competition and innovation. When a competitor enters this segment of the market, the patent owner often sues the newcomer for patent infringement. The USPTO suffers both from a larger than predicted volume of patents applications as well as a lack of sufficient personnel trained in the specific technical details within the EDA industry. By choosing to protect software with patents, the EDA and the electronic industries have forced the USPTO and the courts to artificially modify a definition that was never intended to apply to the protection of abstract entities like algorithms. Legal battles have a negative effect on the industry. Dr. Wally Rhines, chairman and CEO of Mentor Graphics and chairman of the Electronic Design Automation Consortium (EDAC), observes: “Litigation is disruptive to all parts of an organization. It detracts from product development, customer support and management time generally.” And Albert Stritter, Vice President of Design Automation at Infineon Technologies remarks: “R&D development is reduced due to high litigation costs. EDA tools users must prepare contingency plans. In case of an adverse decision migration efforts jeopardize product development cycle time and increase engineering costs.”

When Cadence purchased Quickturn it inherited a patent infringement case that Mentor Graphics had brought against Quickturn. In February of 2003 a federal court in San Francisco ruled against Mentor and declared that the six Mentor patents involved were invalid. The patents all covered hardware methods used in designing hardware emulators used in developing Integrated Circuits (ICs) and the related software. A judgment in favor of Mentor would have significantly restricted any other company from entering the emulation market. In the fall of 2004 Synopsys sued Magma Design Automation for patent infringement. This lawsuit is interesting because its resolution, directly and indirectly, will impact how both companies and individuals use and protect their intellectual property (IP).

Most engineers that develop patentable solutions while working for an EDA company are very rarely properly compensated. On the other hand, if they leave to form a startup that eventually is acquired by one of the big four, a common exit strategy for successful small EDA companies, he or she stands to gain significant financial benefits from a patent they received while at the new company, as long as all of the work was performed at the new company. Since the EDA industry encourages the creation of small companies as a way to test both the robustness of an engineering solution as well as the existence of a market, the EDA company is paying a significant price for innovation, while the individual engineer risks to be sued for stealing company secrets if he ever discussed his ideas while still employed by his previous company. The patent infringement suit brought by Synopsys against Magma Design Automation is an example of what can happen when an employee moves from one company to another. Dr. Lukas van Ginneken, worked at Synopsys from June 1995 to May 1997 when he left to join Magma, who was founded a month before. While at Magma he did work on a method called “constant delay synthesis” and Magma eventually was awarded three patents covering this method. Synopsys claims that Dr. Ginneken started working on the patents while he was still a Synopsys employee and therefore the patents belong to Synopsys. The problem is that an individual engineer uses his knowledge in the performance of his work and cannot “turn off” part of his brain when moving from one company to another. It is reasonable, of course, to ask that the engineer not divulge proprietary or competitive information gathered while at his previous company, but it is not rational to demand an impossible behavior of a professional: to forget part of what he has learned in preparing for and carrying on his technical duties.

The case of Dr. Ginneken is even more striking because constant delay synthesis was his primary area of work first at IBM, then Synopsys and then Magma. He considers the method as a philosophy that helps to optimize and re-order existing algorithms. And there is serious doubt that a philosophy is patentable. The USPTO admits that it does not have enough qualified employees to properly analyze all of the patent claims submitted, and EDA patents are often arcane for even the most informed generalist. Dr. Rhines points out: “It’s extremely difficult for any agency to acquire sufficient expertise to appropriately evaluate technology that is as complex and specialized as EDA.” Just because a professional has a degree in electronic engineering or physics does not mean that he can decide whether constant delay synthesis is a method, an ordered sequence of algorithms, or a philosophy.

Conclusion

In 1999 Cadence sued Avanti for stealing computer codes. The case went to trial with both a criminal and a civil case concluding with a verdict. Cadence won its case and in November 2002 Synopsys, who in the mean time had purchased Avanti, was ordered to pay $265 million as part of the settlement. Because the code was protected by a trade secret, copying the code was a criminal offense. Cadence had not patented the code yet it received as much protection, and arguably faster legal resolution, than if it had used a patent to protect its code.

In November 2002 Silvaco sued Circuit Semantics, Inc. (CSI) also for theft of software code. The two parties reached an out-of-court settlement according to which CSI paid an undisclosed amount to Silvaco. As in the previous case, the code was only considered a part of the company trade secrets and had not been patented. Both cases show that companies do not need to resort to patents in order to maintain their products proprietary and to be able to successfully compete in the EDA market. Patents in the EDA industry are overly restrictive, disrupt the flow of innovation the industry has developed for its own benefit, and will complicate legal matters as foreign EDA vendors become significant players and resort to their own national patent offices to protect their products.

REFERENCES:

1. Kirsch, Gregory J., Software Protection: Patents versus Copyrights, www.gigalaw.com/articles/2000/kirsch-2000-03.html
2. Sietmann, Richard, Fuzzy Patents, IEEE Spectrum NA, September 2005, page 6.
3. Frank, Steven J., Patent Reform Cacophony, IEEE Spectrum NA, December 2005, Page 58-59.

To comment on this article send email to:gmoretti@gabeoneda.com