UNITED STATES INTERNATIONAL TRADE COMMISSION
`WASHINGTON, D.C.
`
`Before The Honorable Doris Johnson Hines
`Administrative Law Judge
`
`In the Matter of
`
`CERTAIN VIDEO CAPABLE
`ELECTRONIC DEVICES, INCLUDING
`COMPUTERS, STREAMING DEVICES,
`TELEVISIONS, CAMERAS, AND
`COMPONENTS AND MODULES
`THEREOF
`
`
`
`Investigation No. 337-TA-1379
`
`RESPONDENTS’ OPENING CLAIM CONSTRUCTION BRIEF
`
`AMAZON-1016
`7,532,808
`
`

`

`TABLE OF CONTENTS
`
`(a) 
`
`4. 
`
`2. 
`
`3. 
`
`2. 
`3. 
`4. 
`5. 
`6. 
`
`3. 
`
`B. 
`
`Page
`INTRODUCTION ................................................................................................. 1 
`BACKGROUND ................................................................................................... 3 
`A. 
`Technology Background ............................................................................ 3 
`Building Blocks: Frames, Pixels, Blocks,
`1. 
`Encoders/Decoders ........................................................................ 3 
`Shrinking the Data: Data Compression .......................................... 4 
`Types of Motion: Global Motion Versus Object Motion .............. 4 
`Copying Pixels: Motion Compensation and Motion Vectors ........ 5 
`Touching Up Visible Errors: Residual Information ....................... 8 
`Macroblock Coding Modes: Motion and Residual
`Information .................................................................................... 8 
`The ’808 Patent: Allegedly Improved Global Motion Coding .................. 9 
`1. 
`Overview of Relevant Claims ........................................................ 9 
`2. 
`The ’808 Patent Changed Prior Art That Disclosed
`Methods to Code Global Motion With And Without
`Residual Information. .................................................................. 10 
`The ’808 Patent Allegedly Improves Coding the Motion
`Part of Global Motion. ................................................................. 11 
`The ’808 Patent Encoder Can Derive Global Motion and
`Can Use Residual Information. .................................................... 12 
`The ’808 Patent Decoder Can Derive Motion and Can Use
`Residual Information. .................................................................. 14 
`LEVEL OF ORDINARY SKILL IN THE ART ................................................. 15 
`LEGAL STANDARDS FOR CLAIM CONSTRUCTION ................................. 16 
`ARGUMENT ....................................................................................................... 16 
`“skip coding mode” (Claims 1, 6-7, 9-10, 15-16, 20, 23-25, 32, 34,
`A. 
`43-44, 51-54, 62-63) ................................................................................ 17 
`Respondents’ Construction Is Consistent With the
`1. 
`Dependent Claims, While Complainants’ and Staff’s
`Construction Invalidates Them. ................................................... 17 
`Respondents’ Construction Is Consistent with All
`Disclosed Embodiments............................................................... 19 
`Complainants’ and Staff’s Construction Impermissibly
`Elevates One Embodiment Over a Claimed Embodiment. .......... 20 
`CONCLUSION .................................................................................................... 22 
`
`I. 
`II. 
`
`III. 
`IV. 
`V. 
`
`VI. 
`
`
`
`-i-
`
`
`
`

`

`TABLE OF AUTHORITIES
`
`Page
`
`CASES
`
`Curtiss-Wright Flow Control Corp. v. Velan, Inc.,
`438 F.3d 1374 (Fed. Cir. 2006)................................................................................................19
`
`In re Katz Interactive Call Processing Patent Litig.,
`639 F.3d 1303 (Fed. Cir. 2011)................................................................................................19
`
`Littelfuse, Inc. v. Mersen USA EP Corp.,
`29 F.4th 1376 (Fed. Cir. 2022) ..........................................................................................19, 21
`
`Multilayer Stretch Cling Film Holdings, Inc. v. Berry Plastics Corp.,
`831 F.3d 1350 (Fed. Cir. 2016)................................................................................................19
`
`Ortho-McNeil Pharm. v. Mylan Lab’ys, Inc.,
`520 F.3d 1358 (Fed. Cir. 2008)................................................................................................19
`
`Phillips v. AWH Corp.,
`415 F.3d 1303 (Fed. Cir. 2005)..........................................................................................16, 21
`
`Tate Access Floors, Inc. v. Interface Architectural Resources, Inc.,
`279 F.3d 1357, 61 U.S.P.Q.2d 1647 (Fed. Cir. 2002) .............................................................17
`
`Teleflex, Inc. v. Ficosa N. Am. Corp.,
`299 F.3d 1313 (Fed. Cir. 2002)................................................................................................21
`
`Teva Pharmaceuticals USA, Inc. v. Sandoz, Inc.,
`574 U.S. 318 (2015) .................................................................................................................16
`
`Wright Medical Technology, Inc. v. Osteonics Corp.,
`122 F.3d 1440 (Fed. Cir. 1997)................................................................................................18
`
`-ii-
`
`

`

`I.
`
`INTRODUCTION
`
`Complainants’ and Staff’s construction of “skip coding mode” should be rejected because
`
`it cannot possibly be correct—it would require invalidating four dependent claims and exclude
`
`embodiments from U.S. Patent No. 7,532,808.
`
`The ’808 patent is directed to improved coding of “global motion,” i.e., when a scene in a
`
`video frame moves because the camera pans in a certain direction. The ’808 patent specifically
`
`alleges to improve upon “Global Motion Vector Coding” (“GMVC”).1 Ex. 1 (’808 patent) at
`
`13:31-33.2 GMVC uses two coding modes. In both modes, the encoder creates global motion
`
`vectors to instruct the decoder how to move the entire frame. Ex. 4 at 2-3. One GMVC mode also
`
`sends residual information—used to touch up visible errors—from the encoder to the decoder,
`
`while the other mode does not.
`
`In contrast to GMVC’s global motion vectors, each of the ’808 patent’s independent claims
`
`(1, 7, 10 and 16) describes a “skip coding mode” wherein motion vectors are derived at the decoder
`
`rather than sent by the encoder. These independent claims are silent on whether the “skip coding
`
`mode” utilizes residual information. But each independent claim is subsequently modified by a
`
`dependent claim whose only addition is to state expressly that “no residual information is
`
`provided.” ’808 patent at cls. 30, 41, 48, and 60 (emphasis added). Thus, for these dependent
`
`claims to have any meaning at all—indeed, for them even to exist—the “skip coding mode” in the
`
`independent claims must necessarily allow residual information to be used in the decoding process.
`
`By reading residual information out of the independent claims completely, Complainants’ and the
`
`Staff’s construction not only ignores these four dependent claims, it renders them a nullity and
`
`invalidates them contrary to black letter precedent.
`
`1 The ’808 patent incorporates GMVC by reference. ’808 patent at 12:52-57.
`2 Exhibits to this motion are attached to the concurrently filed declaration of Thomas N. Millikan.
`
`

`

`
`
`
`
`Complainants’ and Staff’s construction must be wrong also because it is inconsistent with
`
`the specification. The specification discloses embodiments where residual information can be
`
`provided. For example, the ’808 patent discloses embodiments in Figures 6 and 7 in which an
`
`encoder and decoder can provide residual information. ’808 patent at 16:51-55 (The encoder
`
`“codes the motion vector information and control information and forms a single bit stream of
`
`coded image information, 135 comprising the entropy coded motion vector, prediction error [(i.e.,
`
`residual information)] and control information.”) (emphasis added), 21:18-22 (“The prediction,
`
`that is the blocks of pixel values derived from the reference frame, are passed from motion
`
`compensated prediction block 740 to combiner 230 where they are combined with the decoded
`
`prediction error [(i.e., residual information)] information (if any).” (emphasis added)).
`
`Respondents’ proposed construction preserves these embodiments, while Complainants’ and
`
`Staff’s proposed construction excludes them.
`
`Finally, Complainants’ and Staff’s construction improperly imports a limitation from an
`
`embodiment into the claims. One statement in the specification describes not sending residual
`
`information with motion information. But that lone passage merely describes one embodiment,
`
`not the entire invention. Indeed, importing such a limitation would contradict a main tenet of claim
`
`construction, and the fact that it does confirms that Complainants’ and Staff’s construction cannot
`
`be adopted.
`
`For all of these reasons, the ALJ should adopt Respondents’ construction that “skip coding
`
`mode” means “a coding mode in which a zero (non-active) motion vector or a nonzero (active)
`
`motion vector is associated with each skip mode macroblock, depending on the characteristics of
`
`the motion in image segments surrounding the macroblock in question, and residual information
`
`can be provided for each skip mode macroblock.”
`
`2
`
`

`

`
`II.
`
`BACKGROUND
`A.
`
`Technology Background
`
`
`
`In the following sections, Respondents present a high-level summary of the technical
`
`concepts involved in video coding and that underlie the present claim construction dispute, such
`
`as an explanation of how “residual information” (i.e., “prediction error”) is used to touch up visible
`
`errors.
`
`1.
`
`Building Blocks: Frames, Pixels, Blocks, Encoders/Decoders
`
`Digital video is a series of still images shown in quick succession to give the illusion of
`
`movement. ’808 patent at 1:15-19. Each image is a “frame” comprised of millions of individual
`
`“pixels” that emit color and brightness. Id. at 1:32-40.
`
`Sometimes, multiple pixels in a region are grouped together into “macroblocks” (16x16 pixels) or
`
`“blocks” (8x8 pixels). Id. at 1:58-67. The image below illustrates a frame broken down into
`
`macroblocks, blocks, and pixels.
`
`
`
`3
`
`

`

`
`
`
`
`
`
`To generate digital video, an “encoder” is used to encode frame/pixel data into a digital
`
`format that can be transmitted. On the receiving end, a “decoder” decodes the data so it can be
`
`displayed. Id. at 3:47-52.
`
`2.
`
`Shrinking the Data: Data Compression
`
`Encoders have long employed “data compression” techniques to reduce the data that must
`
`be encoded and transmitted. Id. at 2:36-55. Typically, those techniques leverage “redundant” data
`
`that has already been transmitted—for example, by copying previously-sent pixels or sending only
`
`“residual” information to fix visible errors after copying pixels. Id. at 3:11-43. Each of these are
`
`discussed in more detail below.
`
`3.
`
`Types of Motion: Global Motion Versus Object Motion
`
`When a camera moves by panning or zooming, it causes an entire frame to move. That
`
`movement is called “global motion.” Id. 1:26-29. For example, in our car video, if the camera pans
`
`to the left, the entire image shifts to the right, revealing additional information (e.g., another tree),
`
`and hiding other information (e.g., (a part of the sun):
`
`4
`
`

`

`
`
`
`
`
`
`Most macroblocks in Frame 1 and Frame 2 cover the same visual image, just shifted. These
`
`similarities are a form of temporal redundancy. Id. at 2:42-45.
`
`Other times, the camera is still and an object in one frame—e.g., a car driving across a
`
`road—is moved in a subsequent frame. Id. at 2:43-44. This is object motion. Id. For example, the
`
`car below is driving past the trees between Frame 1 and Frame 2. In this case, much of the
`
`background has not shifted, but the macroblocks showing the car have shifted.
`
`4.
`
`Copying Pixels: Motion Compensation and Motion Vectors
`
`To exploit temporal redundancy, a first frame is sent with instructions to copy macroblocks
`
`from the first frame to a second frame and move the blocks according to the object’s movement.
`
`Id. at 3:15-21. This technique is called “motion compensation” or “motion compensated
`
`
`
`5
`
`

`

`
`prediction.” Id. at 3:11-25. The copy instructions indicating where to copy the blocks from in the
`
`
`
`previous frame are called “motion vectors.” Id. 3:18-23.
`
`For global motion, an encoder can estimate how much movement (e.g., panning) has
`
`occurred between frames and create a global motion vector. Id. at 13:30-44. Then, the encoder can
`
`code macroblocks to use the global motion vector. Id. In the example below, rather than encoding
`
`and transmitting all the identical macroblocks in both frames, the encoder can transmit just Frame
`
`1 (called a reference frame3). Then, for a subsequent frame (Frame 2, called a predicted frame4),
`
`the encoder can send a global motion vector and a macroblock coding mode for each of the yellow
`
`macroblocks telling the decoder to copy and shift the macroblocks from Frame 1 according to the
`
`global motion vector. A separate global motion vector is not sent with each macroblock of Frame
`
`2. Id. at 12:53-59.
`
`Similarly, for object motion, the encoder can transmit the reference frame, then, for the
`
`subsequent frame, send a motion vector for each macroblock where the object moved. Id. at 3:18-
`
`21. For example, the encoder can send a motion vector for the macroblock representing the front
`
`
`
`
`3 A reference frame that served as a reference for a predicted frame can also be referred to as an
`INTRA coded frame or I-frame. Id. at 3:3-6.
`4A predicted frame can also be referred to as an INTER coded frame or P-frame. Id. at 3:25-27.
`
`6
`
`

`

`
`of the car. To recreate the front of the car in the predicted frame, the decoder takes the previously-
`
`
`
`received macroblocks representing the car and copies the macroblock representing the front of the
`
`car to the location indicated by the motion vector.
`
`
`
`Sometimes there is no motion between frames. In that instance, macroblocks are typically
`
`copied from the same location in the previous frame and the motion vector is zero. Id. Returning
`
`to the car example, the background did not move between frames. For those elements, the encoder
`
`sends a reference frame and instructs the decoder to copy the macroblocks from the same location
`
`for subsequent frames. Id. at 10:64-67. The prior art called this “copy mode” or “skip mode.” Id.
`
`at 10:63-69; 12:54-58.
`
`
`
`7
`
`

`

`
`
`
`
`While the above techniques save bandwidth, sometimes the image in the reference and
`
`predicted frame changes a little and the copy of a macroblock is imperfect. In those instances,
`
`updates of the predicted macroblocks (i.e., touch ups) may be needed to fix visible errors.
`
`5.
`
`Touching Up Visible Errors: Residual Information
`
`To touch up visible errors, a decoder uses “residual information” (i.e., “prediction error”)5
`
`to improve image quality. Id. at 3:28-38. The residual information represents the difference
`
`between the predicted version of frame 2 and the original version of frame 2. Id. at 3:31-35. For
`
`example, light from the sun may reflect differently on the car between the two frames, causing a
`
`shiny spot on the front. The encoder can send residual information to the decoder that represents
`
`the difference between the macroblock in the reference picture without the shiny spot and the
`
`macroblock in the predicted picture with the shiny spot. Id. at 3:31-35.
`
`
`
`Residual information can be used with global motion and object motion. Id. at 12:54-58, 3:31-35.
`
`6.
`
`Macroblock Coding Modes: Motion and Residual Information
`
`Motion-compensated macroblocks can be assigned coding modes that identify whether the
`
`macroblock has motion information and residual information. Id. at 10:40-65. For example, the
`
`
`5 The terms “prediction error” and “residual information” are interchangeable in the art. See
`Declaration of Dr. Joseph Havlicek at ¶¶ 24-34.
`
`8
`
`

`

`
`GMVC prior art system describes
`
`
`
`three macroblock coding modes
`
`(“GMVC_16,”
`
`“GMVC_Copy,” and “Copy”) that vary in whether they use global motion vectors and/or residual
`
`information. GMVC_16 uses global motion vectors and residual information, GMVC_COPY uses
`
`global motion vectors without residual, COPY uses no motion vectors and no residual. Id. at 12:54-
`
`58. In Joint Model Number 1—another prior art system incorporated by reference in the ’808
`
`patent (id. at 10:34-39)—a “16x16” coding mode can use motion vectors and residual information,
`
`and a “skip mode” (hereinafter “JM1 Skip Mode”) uses no motion and no residual information. Id.
`
`at 10:40-50 and 12:26-37. These macroblock modes are summarized as follows:
`
`Mode
`GMVC_16
`GMVC_COPY
`COPY
`JM1 16x16
`JM1 Skip Mode
`
`Global Motion
`X
`X
`
`
`
`
`Motion
`
`
`
`X
`
`
`Residual
`X
`
`
`X
`
`
`
`
`B.
`
`The ’808 Patent: Allegedly Improved Global Motion Coding
`
`The ’808 patent purportedly changed how the prior art global-motion coding modes
`
`derived motion information—instead of coding global motion vectors, the macroblock derives the
`
`motion from neighboring macroblocks. See, e.g., ’808 patent at Abstract (“A method of motion-
`
`compensated video encoding that enables a video sequence with a global motion component to be
`
`encoded in an efficient manner.”). But the ’808 patent recognizes that residual information is still
`
`needed. ’808 patent at 3:28-32 (“Motion-compensated prediction alone rarely provides a
`
`sufficiently precise representation of the image content of a video frame and therefore it is typically
`
`necessary to provide a so-called ‘prediction error’ (PE) frame with each INTER-coded frame.”).
`
`1.
`
`Overview of Relevant Claims
`
`9
`
`

`

`
`
`
`
`Because claim construction begins with the claims, Respondents want to provide the ALJ
`
`with a table showing the relationship between the claims relevant to this dispute.
`
`1. A method of
`encoding a video
`sequence, the method
`comprising:
` assigning a skip
`coding mode to a first
`segment of a first
`frame of the
`sequence;
`…
`
`30. A method
`according to claim 1,
`wherein no residual
`information is
`provided for the first
`segment in the
`encoded bitstream.
`
`
`7. A method of
`decoding an encoded
`video sequence, the
`method comprising:
` receiving an
`indication of a skip
`coding mode for a
`first segment;
`…
`
`10. A video encoder
`for encoding a video
`sequence, the encoder
`comprises:
` a coding controller
`for assigning a skip
`coding mode to a first
`segment;
`…
`
`41. A method
`according to claim 7,
`wherein no residual
`information is
`provided for the first
`segment in the
`encoded bitstream.
`
`49. An encoder
`according to claim
`10, wherein no
`residual information
`is provided for the
`first segment in the
`encoded bitstream.
`
`16. A video decoder
`for decoding an
`encoded video
`sequence, the decoder
`comprising:
` a demultiplexer
`for receiving an
`indication of a skip
`coding mode
`assigned to a first
`segment;
`…
`60. A decoder
`according to claim
`16, wherein no
`residual information
`is provided for the
`first segment in an
`encoded bitstream.
`
`2.
`
`The ’808 Patent Changed Prior Art That Disclosed Methods to Code
`Global Motion With And Without Residual Information.
`
`The ’808 describes problems with the motion information in GMVC, the prior art method
`
`for coding global motion mentioned in the introduction.6 ’808 patent at 13:31-33. In GMVC, a
`
`reference frame is sent with information to describe global motion and additional macroblock
`
`modes are used to indicate when global motion vectors are used for a given macroblock. Id. at
`
`13:33-36. Specifically, GMVC discloses two global motion macroblock modes (highlighted
`
`below).
`
`
`6 The ’808 patent incorporates VCEG-020 by reference. ’808 patent at 12:52-57.
`
`10
`
`

`

`
`
`
`
`
`
`Ex. 4 at 2 (emphasis added). In GMVC_Copy, the encoder uses an equation to derive motion and
`
`“there is no residual coding.” Id. at 3 (emphasis added). But in GMVC_16, “only the transform
`
`coefficients of the motion compensated residual are coded into the bitstream, not its motion
`
`vector.” Id. (emphasis added). In other words, GMVC disclosed two modes for global motion, one
`
`of which used residual information and the other did not.
`
`GMVC was less complex than other global motion compensation methods, but still added
`
`encoder complexity to estimate motion. ’808 patent at 13:36-42. The method also generated extra
`
`global motion information that could be large for small resolution video. Id. at 13:42-44.
`
`3.
`
`The ’808 Patent Allegedly Improves Coding the Motion Part of Global
`Motion.
`
`Given the state of the art, the ’808 patent’s primary goal was to “combine the simplicity of
`
`local motion compensation with the coding efficiency of global motion compensation….” Id. at
`
`14:6-8. Rather than build a standalone system, the inventors were focused on incorporating their
`
`motion information technique into the draft standard, called “Joint Model Number 1,” or JM1 in
`
`the ’808 patent. Id. at 19:28-51. As mentioned above, JM1 had the JM1 Skip Mode that sent no
`
`motion information and no residual information. The ’808 patent takes the “skip mode” term from
`
`JM1 and then adds its motion information technique. The ’808 patent describe the motion
`
`information for its “skip coding mode” as:
`
`11
`
`

`

`
`
`
`
`
`
`
`
`’808 patent at 17:45-52, 19:28-51.
`
`(a)
`
`The ’808 Patent Encoder Can Derive Global Motion and Can
`Use Residual Information.
`
`The ’808 patent discloses an embodiment in which an encoder can derive motion and use
`
`residual information, as shown in annotated Figure 6 below.
`
`12
`
`

`

`
`
`
`
`
`
`The process for deriving a motion vector starts with the green boxes and arrows at the bottom of
`
`the figure. Motion estimation block 630 (in green) derives a “skip mode motion vector” for a
`
`macroblock and passes it through motion field coding 140 (also in green) to motion compensated
`
`predictor 150 (in blue). Id. at 17:23-54 (annotated). The “skip mode motion vector” is derived by
`
`analyzing “the motion of image segments (e.g., macroblocks and/or sub-blocks) surrounding the
`
`macroblock to be coded.” Id. at 17:57-60. More specifically:
`
`[If] the encoder determines that the motion in the region surrounding the current
`macroblock has a global characteristic, skip mode coding is adapted to take account
`of this (by generating an associated non-zero valued skip mode motion vector
`representative of the motion). Alternatively, if no such motion is present, a zero
`valued motion vector is generated causing the skip mode as modified by the
`invention to operate in a conventional manner i.e. a zero valued skip mode motion
`vector causes a macroblock to be copied directly from the reference frame.
`
`13
`
`

`

`
`Id. at 18:2-11. Figure 8 (including blocks 802, 803, and 804) shows an embodiment for analyzing
`
`
`
`the motion in surrounding macroblocks. Id. at 18:24-19:10
`
`The encoder can also use residual information, as shown in the blue boxes and lines.
`
`Picking up with motion compensated prediction block 150 in blue, it applies the “skip mode
`
`motion vector” to the macroblock received from the frame store 120 to generate a prediction for
`
`the current macroblock. Id. 16:22-31. The motion compensated prediction block “passe[s]” the
`
`prediction “to combiner 116,” which subtracts the prediction for the macroblock against the actual
`
`macroblock, and the result is used to generate “a set of prediction error blocks for the macroblock”
`
`(i.e. residual macroblocks). Id. Those then make their way through transform 104 and quantizer
`
`106 into bit-stream 635. Id. at 8:14-18.
`
`4.
`
`The ’808 Patent Decoder Can Derive Motion and Can Use Residual
`Information.
`
`On the receiving end, the ’808 patent describes an embodiment in which the decoder can
`
`derive motion and use residual information, as shown in annotated Figure 7 below.
`
`14
`
`

`

`
`
`
`
`
`
`For motion, the decoder does not have a “motion estimator” to analyze surrounding macroblocks
`
`per se, but the patent teaches that motion compensated prediction 240 (shown in green) has a
`
`motion analysis block (blocks 802, 803, and 804 shown in Figure 8).
`
`Motion compensated prediction block 240 can derive a motion vector from surrounding
`
`macroblocks, obtain the pixels from a previous frame stored in frame store 250, and generate a
`
`prediction for the current macroblock (shown by the blue arrow). Id. at Fig. 7, 9:56-60, 15:40-42,
`
`20:26-38. The decoder shows prediction error information (i.e., residual information) received by
`
`bitstream 635, passing through inverse quantizer 210 and inverse transform 220, then to combiner
`
`230 where it is added to the predicted macroblock. Id. at 10:8-12.
`
`III. LEVEL OF ORDINARY SKILL IN THE ART
`
`A person of ordinary skill in the art (“POSITA”) at the time of the alleged invention of the
`
`asserted patents would have had a (1) Bachelor’s degree in electrical engineering, computer
`
`15
`
`

`

`
`engineering, computer science, or a comparable field of study, and (2) approximately two to three
`
`
`
`years of practical experience with video and/or image processing or coding. Additional experience
`
`can substitute for the level of education, and vice versa.
`
`IV.
`
`LEGAL STANDARDS FOR CLAIM CONSTRUCTION
`
`It is a bedrock principle of patent law that the claims of a patent define the invention to
`
`which the patentee is entitled the right to exclude. Phillips v. AWH Corp., 415 F.3d 1303, 1312
`
`(Fed. Cir. 2005). Claim terms are generally given their ordinary and customary meaning, which is
`
`the meaning that the term would have to one of skill in the art at the time of the invention, in light
`
`of the entire patent, including the specification and prosecution history. Id. at 1312-13, 1321. The
`
`patent specification “is always highly relevant to the claim construction analysis.” Id. at 1316.
`
`“[T]he prosecution history can often inform the meaning of the claim language by demonstrating
`
`how the inventor understood the invention and whether the inventor limited the invention in the
`
`course of prosecution, making the claim scope narrower than it would otherwise be.” Id.
`
`In some situations, a “court will need to look beyond the patent’s intrinsic evidence and to
`
`consult extrinsic evidence in order to understand, for example, the background science or the
`
`meaning of a term in the relevant art during the relevant time period.” Teva Pharmaceuticals USA,
`
`Inc. v. Sandoz, Inc., 574 U.S. 318, 331 (2015). Extrinsic evidence is “all evidence external to the
`
`patent and prosecution history, including expert and inventor testimony, dictionaries, and learned
`
`treatises.” Philips, 415 F.3d at 1317.
`
`V.
`
`ARGUMENT
`
`The parties and Staff only dispute the construction of “skip coding mode,” a term found in
`
`the independent claims of the ’808 patent.
`
`16
`
`

`

`
`
`
`
`A.
`
`“skip coding mode” (Claims 1, 6-7, 9-10, 15-16, 20, 23-25, 32, 34, 43-44, 51-
`54, 62-63)
`
`Staff’s Construction
`A coding mode in which a
`zero
`(non-active) motion
`vector or a non-zero (active)
`motion vector is associated
`with
`each
`skip mode
`macroblock, depending on the
`characteristics of the motion
`in
`image
`segments
`surrounding the macroblock
`in question.
`
`Complainants’
`Construction
`A coding mode in which a
`zero
`(non-active) motion
`vector or a non-zero (active)
`motion vector is associated
`with
`each
`skip mode
`macroblock, depending on the
`characteristics of the motion
`in
`image
`segments
`surrounding the macroblock
`in question.
`
`Respondents’ Construction
`A coding mode in which a
`zero
`(non-active) motion
`vector or a nonzero (active)
`motion vector is associated
`with
`each
`skip mode
`macroblock, depending on the
`characteristics of the motion
`in
`image
`segments
`surrounding the macroblock
`in question, and
`residual
`information can be provided
`for each skip mode
`Macroblock
`
`
`As the ALJ can see, the parties’ and Staff’s proposed constructions are largely identical,
`
`with Respondents adding one phrase (shown in underline). The identical parts show the parties
`
`and Staff agree motion information can be provided in “skip coding mode.” But the parties and
`
`Staff disagree whether residual information can also be provided in that mode, with Respondents
`
`contending it can. As explained below, the ALJ should adopt Respondents’ construction because
`
`it is supported by embodiments in dependent claims and is consistent with embodiments in the
`
`specification. But adopting Complainants’ and Staff’s strained construction would require the ALJ
`
`to invalidate four dependent claims (claims 30, 41, 49, and 60), disregard embodiments, and import
`
`a limitation from the specification into the claims.
`
`1.
`
`Respondents’ Construction Is Consistent With the Dependent Claims,
`While Complainants’ and Staff’s Construction Invalidates Them.
`
`“Claim interpretation begins, as always, with the language of the claims.” Tate Access
`
`Floors, Inc. v. Interface Architectural Resources, Inc., 279 F.3d 1357, 1370, 61 U.S.P.Q.2d 1647
`
`(Fed. Cir. 2002). Relatedly, “[a court] must not interpret an independent claim in a way that is
`
`17
`
`

`

`
`inconsistent with a claim which depends from it.” Wright Medical Technology, Inc. v. Osteonics
`
`
`
`Corp., 122 F.3d 1440, 1445 (Fed. Cir. 1997). Here, Respondents’ claim construction analysis
`
`begins, as it should, with the claim language. Starting there, the ALJ can see Respondents’
`
`construction accounts for embodiments recited in dependent claims, while Complainants’ and
`
`Staff’s construction would require the ALJ to invalidate those four dependent claims.
`
`Independent claim 1 (highlighted below) recites a method for encoding global motion—
`
`indeed, “motion” appears eight times in the claim.
`
`The method broadly recites “assigning a skip coding mode to a first segment.” It does not exclude
`
`residual information and it cannot because dependent claim 30 specifies an embodiment in which
`
`“no residual information is provided for the first segment”:
`
`
`
`’808 patent at cl. 30 (highlighting added). Because dependent claim 30 specifies that no residual
`
`information is provided, independent claim 1 must allow providing residual information.
`
`
`
`18
`
`

`

`
`Otherwise, claim 30—and identical claims 41, 49, and 60—would be meaningless and not comply
`
`
`
`with 35 U.S.C. § 112 ¶ 4 (“[A] claim in dependent form shall … specify a further limitation of the
`
`subject matter claimed.”); Curtiss-Wright Flow Control Corp. v. Velan, Inc., 438 F.3d 1374, 1380
`
`(Fed. Cir. 2006) (“[R]eading an additional limitation from a dependent claim into an independent
`
`claim would not only make that additional limitation superfluous, it might render the dependent
`
`claim invalid.”). The law strongly disfavors such an outcome. Ortho-McNeil Pharm. v. Mylan
`
`Lab’ys, Inc., 520 F.3d 1358, 1362 (Fed. Cir. 2008) (“[T]his court strives to reach a claim
`
`construction that does not render claim language in dependent claims meaningless.”); Littelfuse,
`
`Inc. v. Mersen USA EP Corp., 29 F.4th 1376, 1380 (Fed. Cir. 2022) (“A claim construction that
`
`leads to [a dependent claim being meaningless] is generally disfavored.”).7
`
`Accordingly, the ALJ should adopt Respondents’ construction.
`
`2.
`
`Respondents’ Construction Is Consistent with All Disclosed
`Embodiments.
`
`Respondents’ construction should also be adopted because it is consistent with all
`
`embodiments disclosed in the ’808 patent. In re Katz Interactive Call Processing Patent Litig.,
`
`639 F.3d 1303, 1324 (Fed. Cir. 2011) (Stating principle that “there is a strong presumption against
`
`a claim construction that excludes a disclosed embodiment…”).
`
`
`7 While conferring, Complainants identified Multilayer Stretch Cling Film Holdings, Inc. v. Berry
`Plastics Corp., 831 F.3d 1350 (Fed. Cir. 2016) to support their position that claim differentiation
`must cede to the ’808 patent’s definition of “skip coding mode.” But Complainants misunderstand
`the issue. In Multilayer, there was a direct inconsistency between a Markush grouping that
`identified four different layer materials and a dependent claim that identified a fifth material not
`in the Markush grouping. Id. at 1358-62. Here, there is no inconsistency because the independent
`claims are silent on residual information and, as the specification shows, it is entirely consistent
`for “skip coding mode” to have both motion and residual information.
`
`19
`
`

`

`
`
`
`
`Specifically, the independent claims (e.g., claim 1) cover embodiments that can have
`
`residual information and dependent claims (e.g., claim 30) cover Complainants’ preferred
`
`embodiment that cannot have residual information.
`
`On top of that, the ’808 patent discloses multiple embodiments in the prior art a