SecureNet Sols. Grp. v. Arrow Elecs.

1

SECURENET SOLUTIONS GROUP, LLC, Plaintiff, v. ARROW ELECTRONICS, INC., Defendant.

Civil Action No. 22-cv-01254-KLM

United States District Court, D. Colorado

June 5, 2023

ORDER

Kristen L. Mix, United States Magistrate Judge

This matter is before the Court on Defendant's Motion for Judgment on the Pleadings Regarding Patent Invalidity Pursuant to 35 U.S.C. § 101 [#39] (the “Motion”). Plaintiff filed a Response [#48] in opposition to the Motion [#39],^[1] and Defendant filed a Reply [#53]. The Court has reviewed the briefs, the entire case file, and the applicable law, and is sufficiently advised in the premises. For the reasons set forth below, the Motion [#39] is DENIED.^[2]

I. Summary of the Case^[3]

“[T]he determination of patent eligibility requires a full understanding of the basic character of the claimed subject matter . . . .” Content Extraction &amp Transmission LLC v. Wells Fargo Bank, Nat'l Ass'n, 776 F.3d 1343, 1349 (Fed. Cir. 2014). Accordingly, the Court goes into great detail regarding the patents underlying this lawsuit before addressing the issue of patent eligibility raised in the Motion [#39].^[4]

This lawsuit was filed on May 19, 2022. Compl. [#1]. In short, Plaintiff sues Defendant for patent infringement based on “the sale and offer for sale of various end-to-end [information technology (IT)] and [operational technology (OT)] solutions in the [internet-of-things] space.” Id. at 1. Plaintiff asserts that Defendant offers and sells technologies for profit which are protected by and infringe on patents owned by Plaintiff, including U.S. Patent No. 9,344,616 (the “'616 patent”) “correlation engine for security, safety, and business productivity,” issued May 17, 2016; U.S. Patent No. 10,862,744 (the “'744 patent”), “correlation system for correlating sensory events and legacy system events,” issued December 8, 2020; and U.S. Patent No. 11,323,314 (the “'314 patent”), “hierarchical data storage and correlation system for correlating and storing sensory events in a security and safety system,” issued May 3, 2022 (collectively, the “Asserted Patents”). Id.

In 2007 (the priority date for the Asserted Patents), smart surveillance systems gained commercial adoption and started to replace traditional security systems, causing challenges for large-scale data analysis. Id. ¶ 7. Plaintiff states that the claims of the Asserted Patents address a need arising specifically within the field of computerized security systems. Id. ¶ 6. Inventors Daniar Hussain and Dr. John Donovan conceived of the inventions after a three-day workshop with the IT Department for the Oakland County, Michigan confederation of police departments—the largest confederation of police departments in the country. Id. ¶ 7. During the workshop, police described several major problems with known police IT systems. Id.

The claims of the Asserted Patents disclose technical solutions to some of these challenges, such as reducing errors and false positives via a particular computerized process. Id. ¶ 8. In particular, the inventors conceived of systems and methods for using integrated cameras, sensor networks, and other data sources with a correlation engine that correlates two or more events weighted by the attribute data of the data sources.^[5] Id. Such correlations could effectively connect crime-related events to specific sensor data, other legacy system data,^[6] 911 calls, anonymous tips, and video records. Id.

Plaintiff provides a high-level depiction of one embodiment of the invention as illustrated in Figure 1 of its ‘744 Patent:

(Image Omitted)

Id. ¶ 9. In this embodiment, various types of security-related data are collected from various sources, such as video cameras and other sensory devices, a video tip system, a card access system, a personnel system, and a vehicle information module. Id. These data describe “primitive events,” which are atomic,^[7] indivisible events from any subsystem, such as people entering a designated area, a vehicle driving the wrong way in a designated lane, a package left behind in an area, a person screaming, glass breaking, or a gunshot. Id.

The primitive events are then normalized by the normalization engine.^[8] Id. ¶ 10. The normalization engine normalizes the primitive events into a “normalized event 115,” which is in a standardized format the system can recognize. Id. The specification^[9] points out that each type of sensory device may have its own normalization engine, or, instead, one normalization engine, as shown in Figure 1 above, may have multiple modules^[10] for each type of sensory device. Id.

In the described embodiment, normalized events 115 are placed in event queue 116 for processing by correlation engine 117. Id. ¶ 11. The basic function of the correlation engine is to correlate two or more primitive events, combinations of primitive events and compound events, and combinations of compound events. Id. Carrying out this function is complex. Id. Figure 2 here shows how the correlation engine works in one embodiment of the invention:

(Image Omitted)

Id.

As described by the specification, the correlation engine receives normalized events from the normalization engine. Id. ¶ 12. These normalized events are then filtered by a privacy filter 204, which applies a set of privacy rules defined by a system administrator. Id. For example, a privacy rule may instruct the system to ignore all primitive events between certain time periods, or to disregard other categories of data. Id.

After applying the privacy filter, the correlation engine applies the business filter, which applies a set of business rules set by a system administrator. Id. ¶ 13. The objective of the business filter is to eliminate unnecessary false alarms by disregarding events when they are not significant based on normal business processes. Id. For example, in a security system designed to guard a data center, a business filter could be configured to ignore primitive events taking place during hours when the data center is scheduled to be serviced. Id.

After the correlation engine has filtered primitive events based on the privacy and business rules, it evaluates the remaining primitive events for the presence of “compound events”—events that are composed of one or more primitive events. Id. ¶ 14. An example of a compound event is tailgating, i.e., where two or more persons enter a designated area as detected on the video data using a person-counting algorithm, when only one corresponding swipe/access card is detected by the legacy access control system. Id. Compound events may include primitive events from one sensor, from two or more sensors, or even from two disparate types of sensors. Id.

Next, the correlation engine uses a “correlation module 210” to correlate both the primitive and compound events across geographical space. Id. ¶ 15. For example, the correlation engine may identify multiple tailgating events in different parts of a facility, or the loitering of two different vehicles in different parts of a campus. Id. The correlation module also correlates events across time, by comparing events detected presently with events detected in the past. Id. Examples include detection of the presence of the same individual allowing another to tailgate at different times, or the same person loitering or being stopped multiple times by security. Id.

The correlation engine's forensic analysis of events is depicted in greater detail in Figure 10 of the specification:

(Image Omitted)

Id. ¶ 16. This figure depicts various sets of video data (i.e., V₁ through V_i; the large circles in the first column), with each subset corresponding to data obtained from a particular video camera. Id. The dashed circles inside V₁ through V_i each represent a subset of video data. Id.

This figure also depicts various sets of meta-data (i.e., M₁ through M_i; the large circles in the second column). Id. ¶ 17. Each set of meta-data is indexed and points to at least one set of video data. Id. As depicted in this figure, each set of meta-data corresponds to one and only one set of video data, but the relationship between metadata and video data may be one-to-many, many-to-one, as well as many-to-many. Id.

Finally, Figure 10 depicts various sets of attribute weight data (i.e., W₁ through W_i; the large circles in the third column). Id. ¶ 18. The sets of attribute weight data are sets of vectors which represent weights associated with subsets of the meta-data M₁. Id.

These weights may be multi-dimensional. Id. For example, a two-dimensional weight may represent the attribute weights associated with (i) the reliability of a particular video camera for motion detection, and (ii) the reliability of that camera for gunshot detection. Id. This would enable the system to account for the fact that a camera might have high motion detection reliability and low gunshot detection reliability, or vice-versa. Id. The specification provides an equation that can be used for determining weights for attribute data:

(Image Omitted)

Id. In this equation, the weights “ωi may be a weighted average of attribute data (αi). Id. The symbol “ωi” refers to relative weights of the attributes (αk), which are themselves weights associated with the data sources. Id.

The specification further discusses how attribute weights may be recalculated to yield the weighted attribute data of (i) two or more events occurring substantially simultaneously, or (ii) any of two or more events occurring substantially simultaneously. Id. ¶ 19. Respectively, the formulae are as follows:

(i) W(M₁∩M₂) = W(M₁) ∙ W(M₂)

(ii) W(M₁∪M₂) = W(M₁) + W(M₂) - W(M₁) ∙ W(M₂)

Id.

As the compound events and correlated events are detected, the correlation module stores them in the events database...