Case 1:16-cv-00455-RGA Document 394 Filed 02/07/18 Page 1 of 4 PageID #: 29726
`
`M O R R I S , N I C H O L S , A R S H T & T U N N E L L L L P
`1201 NORTH MARKET STREET
`P.O. BOX 1347
`WILMINGTON, DELAWARE 19899-1347
`
`(302) 658-9200
`(302) 658-3989 FAX
`
`STEPHEN J. KRAFTSCHIK
`(302) 351-9378
`(302) 498-6233 FAX
`SKRAFTSCHIK@MNAT.COM
`
`
`
`The Honorable Richard G. Andrews
`United States District Court
` for the District of Delaware
`844 North King Street
`Wilmington, DE 19801
`
`February 7, 2018
`
`VIA ELECTRONIC FILING
`
`Re:
`
`Acceleration Bay LLC; C.A. Nos. 16-453 (RGA); 16-454 (RGA); and 16-455 (RGA)
`
`Dear Judge Andrews:
`As the Court requested (Ex. A, Jan. 29, 2018 Hearing Tr. (“Tr.”) at 78-79), Defendants
`write to address the expert testimony provided at the hearing concerning Term 4 (“means for
`connecting”). The testimony confirmed that the alleged “First Embodiment” is irrelevant to the
`claims at issue and, even if relevant, is not sufficient to accomplish the stated function of Term 4.
`
`Figs. 3A and B are Irrelevant
`
`Plaintiff has argued that “[t]he first embodiment is disclosed in Figures 3A and 3B and
`described in the corresponding description [at 5:33-55].” D.I. 345 at 1 (all D.I. citations refer to
`C.A. No. 16-453). Dr. Mitzenmacher testified that Figs. 3A, 3B depict a node z “connecting to
`the broadcast channel [that] is shown in Fig. 3A.” Tr. at 29:1-3. He also testified that the first
`paragraph of claim 13 “defines the plurality of broadcast channels” and that Term 4 relates to
`connecting to one of the defined plurality of broadcast channels. Id. at 29:5-17. But as Dr.
`Mitzenmacher also admitted, the “plurality of broadcast channels” are defined in the claims to
`require that “the number of participants is at least two greater than m thus resulting in a non-
`complete graph.” Tr. at 29:24-30:21; A-1 (‘344 patent) at 30:20-21; A-2 (‘966 patent) at 30:52-
`54. Fig. 3A, however, is a “complete” graph where the “number of participants” is 5, one more
`than m (4), as Dr. Mitzenmacher also admitted. Tr. at 30:5-9, 18-31. Thus, based on both the
`plain language of the claims and Dr. Mitzenmacher’s admissions, the Fig. 3A graph cannot be
`one of the “plurality of broadcast channels” of claim 19 in either the ‘344 patent or the ‘966
`patent.
`
`Although Dr. Mitzenmacher pointed out that the Fig. 3B graph meets the language of the
`claims for a “plurality of broadcast channels” (id. at 30:14-17), the fact that the channel is “non-
`complete” only after node z joins is irrelevant because the “means for connecting” relates to the
`graph being joined, not the graph that results from the joining, as Dr. Mitzenmacher admitted.
`Tr. at 29:5-17. The testimony of Dr. Kelly on this same point was unequivocal: Figs. 3A, 3B are
`not relevant to the claims at issue. Tr. at 76:21-77:17.
`
`

`

`Case 1:16-cv-00455-RGA Document 394 Filed 02/07/18 Page 2 of 4 PageID #: 29727
`The Honorable Richard G. Andrews
`February 7, 2018
`Page 2
`
`
`Accordingly, Figs. 3A, 3B depict a computer connecting to a small regime “complete”
`graph, whereas Term 4 is directed to connecting to a channel “of interest” from among a
`plurality of “non-complete” channels. A-1 at 30:23-25. Thus, the alleged first embodiment is
`not relevant to the claims at issue and cannot be used as corresponding structure for Term 4.
`
`The Alleged Algorithm of Figs. 3A, 3B is Insufficient
`
`Even if Figs. 3A, 3B were relevant to the asserted claims, the corresponding disclosure is
`clearly insufficient. Each of the “three steps” is no more than a “black box” disclosure which
`would improperly expand the scope of Term 4 to cover virtually any algorithm for connecting.
`ePlus, Inc. v. Lawson Software, Inc., 700 F.3d 509, 518 (Fed. Cir. 2012).
`
`Alleged Step One. Plaintiff contends that the first step in the algorithm is “locating the
`
`broadcast channel.” D.I. 345 at 3. Plaintiff relies on the following portion of the specification:
`
`Each computer is aware of one or more “portal computers” through which
`that computer may locate the broadcast channel. A seeking computer
`locates the broadcast channel by contacting the portal computers until it
`finds one that is currently fully connected to the broadcast channel.
`
`A-1 at 5:37-42; see D.I. 345 at 3. This thin disclosure says nothing about how the seeking
`computer is able to make contact with the portal computer, which is no simple task. As Dr.
`Kelly explained: “there are tens of thousands of ports that the seeking computer may have to dial
`to actually make a connection with the portal computer.” Tr. at 66:3-17; see also id. at 34:9-11
`(Dr. Mitzenmacher agreeing that “a portal computer, a typical computer would have tens of
`thousands of ports.”). Dr. Kelly explained that “there has to be some mechanism to deal with
`that – such a large number of potential ports. And the patent goes on to explain that there are
`ways to deal with that. It’s the port ordering algorithm, for example.” Id. at 66:18-22. Dr. Kelly
`also testified that “it is just not practical for a seeking computer to dial[]those [ports] one after
`the other. It would take far too long.” Id. at 67:1-3. Indeed, the patent specification supports
`Dr. Kelly’s opinion, stating that the dialing of each port would be a “problem.” A-1 at 12:41-52.
`Dr. Kelly further testified that “you’ve got to find some mechanism for identifying, or increasing
`the likelihood that you will identify the port. And that’s what the port ordering algorithm is used
`for.” Tr. at 67:7-9. This is exactly what the patent teaches. See, e.g., A-1 at 12:49-52.
`
`Yet another problem not addressed by the disclosure in col. 5 is that a seeking computer
`
`needs to know when to stop searching for a portal computer and conclude that a portal computer
`cannot be found. See, e.g. A-1 at 12:58-65. In other words, because of the sheer number of
`possible ports that could be searched by a seeking computer, there must be a mechanism for
`determining when the search should end. Id. Dr. Mitzenmacher conceded that “you might not
`want the computer to do that [i.e., keep endlessly searching].” Tr. at 39:5-40:11. Although col.
`5 is silent on this issue, other parts of the specification detail how the seeking computer stops
`searching when a certain “search depth” is reached. See, e.g. A-1 at 12:33-65; 18:57-19:5.
`
`Accordingly, at least for the reasons that both the “port ordering algorithm” and the
`
`“search depth” disclosures in the specification are “integral to performing the stated function” of
`
`

`

`Case 1:16-cv-00455-RGA Document 394 Filed 02/07/18 Page 3 of 4 PageID #: 29728
`The Honorable Richard G. Andrews
`February 7, 2018
`Page 3
`
`locating a portal computer, the brief disclosure in col. 5 should not, by itself, be considered
`sufficient corresponding structure for the “connecting” function of Term 4. Gemstar-TV Guide
`Intern., Inc. v. ITC, 383 F.3d 1352, 1362 (Fed. Cir. 2004) (Structure that is “integral to
`performing the stated function” constitutes corresponding structure.).
`
`Alleged Step Two. Plaintiff contends that the second step in the algorithm is
`
`“identifying the neighbors for the connecting computer.” D.I 345 at 3. The entire portion of the
`specification on which Plaintiff relies for this step is the following sentence: “The found portal
`computer then directs the identifying of four computers (i.e., to be the seeking computer's
`neighbors) to which the seeking computer is to connect.” Id., A-1 at 5:42-45. This single
`sentence says nothing at all about how the portal computer is able to find the neighbors for the
`seeking computer. Dr. Medvidovic addressed this lack of disclosure by stating that it is “a
`streamlined and straightforward process for the portal computer to identify four computers to
`which the seeking computer can connect.” D.I. 346 at 5. He did not say what that “process” is
`or explain why it is “streamlined and straightforward.” Dr. Mitzenmacher similarly provided no
`explanation at the hearing for why this sentence in col. 5 is a sufficient disclosure, other than
`arguing that it is “understandable” to a POSITA. Tr. at 19:25-20:16.
`
`In fact, the patent specification explains that it is anything but “straightforward” for the
`
`portal computer to identify the neighbors for the seeking computer because the portal computer,
`like all participants, has only local knowledge. See, e.g. A-1 at 13:23-36. Local knowledge
`means that any given participant only has knowledge of its own neighbors, not “global
`knowledge” of the entire network. Id.; see also Tr. at 44:2-45:16; 46:23-47:9 (Dr. Mitzenmacher
`agreeing that the portal computer has only local knowledge.) The patent explains that “[t]his
`local knowledge makes it difficult for a portal computer to randomly select four neighbors for a
`seeking computer.” A-1 at 13:34-36. Dr. Kelly explained that “a portal computer that has been
`connected to by the seeking computer only knows about a subset of the rest of the network… and
`you simply can’t use that local knowledge to find the neighbors, because that ends up with
`elongating the network.” Tr. at 68:4-23; see also A-1 at 6:64-7:29 (explaining that, although a
`“seeking computer could connect to the broadcast channel by connecting to computers either
`directly connected to the found portal computer or directly connected to one of its neighbors…
`[a] possible problem with such a scheme” is elongation of the network.).
`
`The patents address this “local knowledge” problem by describing a mechanism whereby
`
`the found “portal computer sends an edge connection request message through one of its internal
`connections that is randomly selected.” A-1 at 13:37-39. Then, the “receiving computer again
`sends the edge connection request message through one of its internal connections that is
`randomly selected.” Id. at 13:39-42. “This sending of the message corresponds to a random
`walk through the graph that represents the broadcast channel.” Id. at 13:42-44; see also id. at
`18:53-56; 19:66-20:44; 23:34-24:38 (describing in detail the algorithm for finding a neighbor).
`This sending of messages to identify neighbors is the only technique described in the patents for
`overcoming the “local knowledge” problem inherent in the claimed m-regular systems. As such,
`it is clearly integral to the connecting function.
`
`Accordingly, because the found portal computer has only local knowledge, the step of
`
`locating neighbors is not at all “straightforward” as Plaintiff contends, and the sending of the
`
`

`

`Case 1:16-cv-00455-RGA Document 394 Filed 02/07/18 Page 4 of 4 PageID #: 29729
`The Honorable Richard G. Andrews
`February 7, 2018
`Page 4
`
`“edge connection request” described in various other portions of the specification is “integral to
`performing the stated function” and thus should be included as corresponding structure for the
`“connecting” function of Term 4. Gemstar-TV Guide, 383 F.3d at 1362.
`
`Alleged Step Three. Plaintiff contends that the third step in the algorithm “requires that
`
`each of the four computers to cooperate with the seeking computer to connect to the broadcast
`channel.” D.I 345 at 3. The entire support for this step is at col. 5, lines 45-48: “Each of these
`four computers then cooperates with the seeking computer to effect the connecting of the seeking
`computer to the broadcast channel.” Id. This single sentence says nothing about how the
`identified neighbors will cooperate “to effect the connecting.” Dr. Medvidovic merely asserted
`that the “seeking computer and the four [neighbor] computers can connect with each other using
`basic networking procedures.” D.I 346 at 5. Dr. Mitzenmacher provided no more of an
`explanation at the hearing. Tr. at 19:12-24.
`
`Once again, the process disclosed in the patent specification is more complicated than
`Plaintiff’s experts contend. Fig. 17 provides the actual algorithm for adding a neighbor and
`includes critical details such as adjusting the “connection state” of the seeking participant as it
`adds neighbors to avoid a problem with forming “disjoint broadcast channels.” A-1 at 12:66-
`13:6; 23:3-8. Furthermore, the add neighbor routine must adjust the respective lists of neighbors
`to reflect the new connections, or else the broadcast channel cannot function to send information,
`which relies on these neighbor lists. Id. at 23:8-11. Dr. Kelly explained these critical details of
`the add neighbor process. Tr. at 69:4-70:8. Dr. Mitzenmacher admitted that it is “important” to
`update the list of neighbors as new neighbors are added. Id. at 56:8-57:1.
`
`Conclusion. Plaintiff relies upon Figs. 3A, 3B as a distinct “first embodiment” that is
`sufficient corresponding structure for the “connecting” function of Term 4, but these figures,
`being directed to adding a node to a complete graph, are not even relevant to the non-complete
`graphs of the claims. Furthermore, the sparse description in col. 5 is not sufficient to accomplish
`the connecting function of Term 4. There are many critical details that are addressed elsewhere
`in the specification and which are clearly “integral” to the claimed connecting function and thus
`should be included in any corresponding structure for Term 4.
`
`Defendants respectfully ask the Court to amend its previous claim construction to clarify
`that Figs. 3A, 3B, along with column 5, lines 33-55, do not alone provide sufficient
`corresponding structure for the “means for connecting” of Term 4.
`
`
`
`
`
`
`
`
`
`
`
`
`
`Respectfully,
`/s/ Stephen J. Kraftschik
`Stephen J. Kraftschik (#5623)
`
`SJK:ncf
`Enclosure
`cc:
`Clerk of Court (Via Hand Delivery)
`
`All Counsel of Record (Via Electronic Mail)
`
`