As v. Iancu

PGS GEOPHYSICAL AS, Appellant v. ANDREI IANCU, UNDER SECRETARY OF COMMERCE FOR INTELLECTUAL PROPERTYAND DIRECTOR OF THE UNITED STATES PATENT AND TRADEMARK OFFICE, Intervenor

2017-1582

United States Court of Appeals for the Federal Circuit

June 18, 2018

NOTE: This disposition is nonprecedential.

Appeal from the United States Patent and Trademark Office, Patent Trial and Appeal Board in No. IPR2015-00313.

JESSAMYN SHELI BERNIKER, Williams & Connolly LLP, Washington, DC, argued for appellant. Also represented by DAVID I. BERL, DAVID M. KRINSKY, JAMES MATTHEW RICE, CHRISTOPHER ALAN SUAREZ.

THOMAS W. KRAUSE, Office of the Solicitor, United States Patent and Trademark Office, Alexandria, VA, argued for intervenor. Also represented by NATHAN K. KELLEY, MONICA BARNES LATEEF, MEREDITH HOPE SCHOENFELD.

Before LOURIE, CLEVENGER, and REYNA, Circuit Judges.

CLEVENGER, Circuit Judge.

PGS Geophysical AS ("PGS") appeals the final written decisions of the Patent Trial and Appeal Board ("the Board") in an inter partes review ("IPR") proceeding instituted by WesternGeco LLC ("WesternGeco").¹ In its first decision, the Board invalidated claims 1, 4, 10 and 11 of U.S. Patent No. 6,026,059 ("'059 Patent") as being anticipated or obvious in light of the prior art. Further, after granting WesternGeco's request for rehearing, a majority of the Board invalidated dependent claim 2 by associating its limitation to one step in independent claim 1 (which was taught by the prior art), over PGS's arguments that the limitation applied to a different step in claim 1 (which was not taught by the prior art). We agree with the Board as to the invalidity of claims 1, 4, 10 and 11, but disagree with the majority of the Board as to claim 2.

BACKGROUND

The '059 Patent concerns three-dimensional seismic surveying and processing of the resultant data. Seismic survey data is generated and acquired using source-receiver pairs; a series of "sources" are physically placed in an array relative to a series of "receivers." The sources emit a "shot" via vibrations or explosions, which travels through the target geology and bounces off geological features before returning to the receivers. Figure 1 below shows a representative seismic survey diagram and "generalized waveform response"—known as a "trace"—picked up by the receiver. See J.A. 144. When a shot bounces off of a geological feature, it produces a spike in the trace signal's amplitude relative to the background noise. Data processors then collect these trace signals and utilize a variety of techniques to increase the resolution and accuracy of the survey, in essence turning discrete signal spikes into subsurface maps.

Image materials not available for display.

Fig. 1. Simplified diagram of seismic principle used in exploration

WORLD OIL | APRIL 1994 85

One metric by which data processors measure the resolution of a survey is through the signal-to-noise ratio—the ratio of signal strength (i.e. signal carrying relevant information) to background noise. For three- dimensional seismic surveys, data processors often increase this ratio using a process called binning, which groups traces together by some shared feature. For instance, traces may be grouped into common midpoint bins (CMBs)—containing traces that have the same lateral midpoint between their source-receiver pairs—or common reflection point bins (CRPs)—containing traces that have the same subsurface reflection point between their source-receiver pairs. CMBs are generally used for simple sub-surface geometries, whereas CRPs may be used for more complex geometries.

Each bin has a particular "fold," which is the number of traces within the bin. Each trace within the bin also has a particular "offset"—the distance between the source and receiver that produced the trace—and "azimuth"—the angle between the offset line and some reference axis. The figure below depicts an overhead view of a bin, where each line passing through the reference point at "2" represents a single trace. See J.A. 642. By "stacking" (i.e. summing) numerous traces having a common reference point (i.e. midpoint or reflection point), the amplitude of the signal becomes more pronounced relative to the amplitude of the noise, thereby increasing the signal-to-noise ratio and overall resolution of the survey.

Image materials not available for display.

However, bins generally contain non-uniform offset and azimuth distributions, as shown in the figure. In other words, traces may be more concentrated at certain offsets or azimuths, rather than evenly distributed about the reference point. In the figure, bin 2 contains numerous traces from mid-distance source-receiver pairs, but far fewer traces from both the nearest and furthest source-receiver pairs. According to the '059 Patent, these non-uniform distributions negatively impact the analysis of the stacked trace data. In particular, variations that arise when normalizing² the amplitude of each trace subsequently impact the amplitudes of the stacked traces.

The '059 Patent purports to solve the problem of non-uniform offset and azimuth distributions by having traces evenly distributed around a bin's reference point. According to the '059 Patent, each bin is assigned a coordinate system, where each trace in the bin is given coordinate values based on the source-receiver pair's position in the array. These coordinate values may then be plotted in a "spider" graph, as shown in Figure 6 of the '059 Patent (annotations added).

Image materials not available for display.

FIG. 6

Once the coordinate system has been assigned, the '059 Patent teaches a process of generating a series of smaller "sub-bins" within the coordinate system. According to the '059 Patent, the goal is for each sub-bin to contain the same number of traces; each sub-bin is "regularized."³ For example, each sub-bin in Figure 6 containstwo traces. The process of generating sub-bins with the same number of traces ensures each bin has sufficient offset and azimuth diversity to increase the signal-to-noise ratio, and that those traces are uniformly distributed within each bin so as to avoid problems caused by amplitude normalization.

WesternGeco filed a petition requesting an IPR of claims 1-12 of the '059 Patent. The Board instituted the IPR only as to claims 1-5, 10 and 11. Although the Board erred in failing to institute the IPR on every claim WesternGeco challenged, SAS Inst., Inc. v. Iancu, 138 S. Ct. 1348, 1354 (2018), neither the Appellant nor the Intervenor complain about this failure, PGS Geophysical AS v. Iancu, Nos. 16-2470, 16-2472, 16-2474, slip op. at 11-13, 2018 WL 2727663 at *5-6 (Fed. Cir. June 7, 2018) (noting that the Board's partial institution decision is a waivable error). Claim 5 was upheld, and is not implicated in this appeal.

Independent claim 1 of the '059 Patent discloses a method of generating bins with regularized sub-bins. The relevant claims at issue read as follows:

1. A process for generating a bin of common midpoint traces from a three dimensional seismic survey data set, each of the traces having a shot location and a receiver location associated therewith, the process comprising:

gathering from the data a plurality of traces having a common reference point . . . ;

assigning a coordinate set to a plurality of traces in the common reference point bin, wherein the coordinates are associated with the shot position and the receiver position associated with the traces . . . whereby a coordinate designated set of traces is defined; and

organizing the coordinate-designated set of traces into a set of bins having a regularized number of traces.

2. A process as in claim 1, wherein a plurality of the coordinate-designated set of traces have the same coordinates.

3. A process as in claim 2, further comprising adding a plurality of traces having the same coordinates.

4. A process as in claim 1, wherein each trace has a unique set of coordinates.

10. A process as in claim 1 wherein the common reference point comprises a common midpoint.

11. A process as in claim 1 wherein the common reference point comprises a common reflection point.

'059 Patent, col. 5 l. 48-col. 6 l. 4; col. 6 ll. 20-23.

In its first written decision, the Board found that claims 1, 4 and 10 were anticipated under 35 U.S.C. § 102 by U.S. Patent No. 4,933,912 ("Gallagher"). Gallagher discloses a method of improving the signal-to-noise ratio in CMB data processing by ensuring the selected traces have diverse offsets and azimuths. Gallagher, col. 1 ll. 51-54. Gallagher selects particular traces by: (1) choosing a desired number (n₁) of folds in a particular CMB, (2) assigning a coordinate system to the bin's trace data, (3) generating a number (n₂) of angular sections (i.e. lines A and B, below) in the coordinate system, and (4) generating a number (n₃) of concentric shells (dashed concentric circles, below) in the coordinate system. Id. at col. 5 l. 15- col. 6 l. 67. Preferably, n₂ and n₃ are selected such that n₂ x n₃ = n₁. Id. at col. 8 ll. 3-4. In other words, if a CMB with a fold of sixteen is desired, Gallagher teaches generating four sections (i.e. quadrants) and four concentric shells, thereby producing sixteen section-shell regions (i.e. sub-bins). After creating the section-shell regions, Gallagher teaches (5) selecting a single trace from each section-shell region, and (6) stacking the selected traces. Id. at col. 6 l. 68-col. 7 l. 41; col. 8 ll. 48-55. If one region does not contain a trace (because no eligible source or receiver was physically within the post hoc region), a second trace may be selected from any other region. Id. at col. 7 ll. 49-59. As shown in Figure 1 of Gallagher (annotations added), the use of sections ensures the selected traces have varying azimuths, while the use of concentric shells ensures they have varying offsets.

Image materials not available for display.

FIG. 1

Throughout the IPR, PGS attempted to distinguish Gallagher by arguing that the "organizing" step of claim...