A lawyer's view: IP implications of SoC

By Robert F. Gazdzinski
Gazdzinski & Associates, San Diego
(Reprinted from EETimes.com On-line Magazine - June 2000)

With the rapidly rising popularity of low-cost, high-volume consumer goods such as cellular telephones and pocket electronics, the cost and size of their IC components have become more significant considerations. Improved fabrication processes and increased demand for I/O and peripheral capability mean more and more historically discrete devices have been incorporated or integrated directly on a single die. While providing many tangible benefits in terms of reduced cost and size and in many cases enhanced processing performance and lower power consumption, these system-on-chip (SoC) devices also can create intellectual property concerns.

The recent trend in IC design has been away from the more general-purpose, multi-application discrete components toward more narrowly focused and well integrated devices that are specifically adapted to the end user's required functionality. In many cases nonessential functions are excised from the design, thereby reducing the complexity and gate count of the device to the minimum required. The benefit of such reduction and simplification is generally a smaller, less costly and more power-efficient device whose logic design can be synthesized or modified more easily as well. However, this integration and simplification can be a double-edged sword as the device becomes more narrowly focused on accomplishing a given function or process, it may also have greater exposure to the patent rights of third parties.

Under the U.S. patent laws, a valid and enforceable patent provides the owner or assignee the right to exclude other parties from making, using, selling or having made the patented invention. So the patent gives its owner no affirmative rights per se to manufacture or practice a given invention, but rather is wholly exclusionary. According to 35 U.S.C. §271(c), one who knowingly offers for sale or sells a component or apparatus that has no "substantial noninfringing use"-that is, no substantial use other than one that infringes on a patent-can be held liable for infringement. So where a manufacturer produces a specialized SoC design that is particularly adapted to perform an infringing function, that manufacturer can be liable as a contributory infringer.

Clearly, the more elements of the patent claim or claims that a given single product satisfies (the product having no other substantial non-infringing uses), the more the producer of that product contributes to the infringement. However, contributory infringement has, under certain circumstances, been held to exist even in cases where the component that contributes to the infringement is only a modest portion of the whole device or method specified in the claim.

Contributory infringement

This relief for so-called contributory infringement is based on the policy of protecting the patent owner's rights against attempts by others to circumvent them by dividing an invention into components, none of which meets all of the elements of the patent holder's claim.

For example, consider the case of a general-purpose digital signal processor used within a cellular telephone to, among other things, code speech. Assume that the use of the DSP with the other components of the telephone infringes on a patent for a method of communication utilizing speech coding. Having a broad functionality, the general-purpose DSP could be extracted from the cellular telephone and readily reprogrammed for use in another (noninfringing) application; that way it would likely have little or no exposure to the patent under a contributory infringement theory.

However, if a DSP core is integrated on a single die with other components such as memory, I/O interface, front-end or back-end processing and RF transceiver or CDMA channel spreader and specially adapted for speech coding and other functions within the telephone, the potential for liability based on a theory of contributory infringement could be significantly increased.

Similarly, the arguments for inducement of infringement under 35 U.S.C. §271(b) can become more compelling in the context of highly integrated devices such as a SoC. In simple terms, inducement under §271(b) is established when a party or parties actively induces or otherwise facilitates infringement by combining their device or service with that of others. So the manufacturer that provides the telephone less the DSP and also provides explicit instructions on selecting and configuring a DSP to create a specific configuration or functionality that would infringe a method claims of a patent, could be liable for infringement under a theory of inducement. Method claims, as the name implies, list steps or actions that are taken to accomplish a specific result. The "result" may vary widely, ranging from producing or manufacturing a product to providing a specific functionality, such as coding the speech signal via the DSP of the cellular phone in the example.

Additionally, when combining functions on a single device and selling that device as a discrete product, the likelihood of infringement of a patent holder's method or process claims could become more likely as well. The law requires that for a method claim to be infringed, there must either be a single infringing party performing the steps of the method, or some relationship between multiple parties each performing one or more of the steps.

Infringement breakdown

Consider a hypothetical method claim covering five steps necessary for speech coding. If the physical components necessary to perform those five steps are manufactured and sold by five separate entities, with the end user or consumer assembling the device and performing the patented method only when the device is assembled, only the end user could be liable for infringement of the method claim unless there is infringement under either inducement or contributory infringement theories. However, if the five components are integrated into an SoC that is assembled into the telephone and tested by a single entity, that single entity could be liable under a direct infringement theory because the method of coding speech is being used when the assembled device is tested.

Even if the single entity did not actually test the device, but merely assembled it and shipped it to the end user, that entity could have liability under an inducement theory since it is ultimately causing the end user to infringe the method claim when the latter operates the phone. Hence, there is often somewhat of a "reverse synergy," in that the combination of the individual components into a single device can provide greater potential for exposure than the sum of the pieces taken individually. Similarly, some apparatus or device claims may also be more readily infringed by SoC products as compared to discrete components.

The so-called first-sale doctrine is well established in the law. It provides that an authorized sale of a patented invention-that is, a device that would infringe one or more claims of a patent-places that product beyond the reach of that patent. The purchaser of the product may then use or resell the product with relative impunity. This principle applies equally to the sale of a patented product manufactured by a licensee or even sublicensee acting within the scope of its license. However, the waters can become muddied somewhat when defining just what constitutes a "sale." The U.S. Circuit Court's decision in Intel Corp. vs. ULSI System Technology Inc. highlighted the need for explicit proscription of foundry services within license agreements and contracts. But with SoC, the picture becomes more complicated.

Consider that an SoC device comprises the intellectual property of a number of unrelated parties. Portions of the device may feasibly be licensed under certain terms, while others may not or may be licensed under different terms. But what if a third party seeks to provide a "licensed" foundry with its own design of the SoC device, much as in the Intel case, and the design arguably infringes the claims of patents assigned to those unrelated parties?

Then the existence of foundry clauses within the licenses granted by each party to the foundry may be critical to determining whether the foundry can rightfully produce the design for the third party. Clearly, the more IP that is implicated (such as in SoC devices), the greater the potential for problems. Further, the Federal Circuit's 1991 decision in Intel Corp. vs. United States International Trade Commission ("Atmel"), which stands for the proposition that a third party can only resell products purchased from a foundry that is licensed to produce products for third-party resellers, adds complexity.

Despite all that, vendors of SoC devices or designs in some cases offer indemnification to the customers to shield them from potential exposure to third-party patents. Indemnification provides that the producer or seller of a device will "stand in the customer's shoes" when the latter is alleged to infringe a patent. However, in the case of SoC devices, several licensable technologies may be employed within a single chip, thereby potentially increasing the exposure of the vendor.