PTAB

IPR2025-00049

Digital Global Systems Inc v. DeepSig Inc

Log In
Key Events
Petition
petition Intelligence

1. Case Identification

2. Patent Overview

  • Title: Correcting Radio Signal Distortion Using Machine Learning
  • Brief Description: The ’469 patent discloses techniques for mitigating radio signal distortion by using a nonlinear pre-distortion machine learning model, such as a neural network, to generate a pre-distorted signal that counteracts impairments introduced by transmitter components like amplifiers.

3. Grounds for Unpatentability

Ground 1: Claims 1-8, 10, 12-15, and 17-24 are obvious over Jüschke in view of Holt.

  • Prior Art Relied Upon: Jüschke (EP 2,538,553 A1) and Holt (Patent 10,552,738).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Jüschke taught the core elements of the challenged claims, including using a "pre-distortion neural network processor" to mitigate transmitter impairments (e.g., from amplifiers) and training that network by comparing input and output signals in a feedback loop. However, Jüschke's feedback loop is internal to the transmitter. Holt was argued to supply the missing element that led to allowance: training a machine learning model based on data from an actual receiver. Holt taught a communications system with an encoder (transmitter) and a decoder (receiver), where a loss function is calculated based on the difference between the original input signal at the encoder and the final output signal from the decoder. Petitioner asserted that adding Holt's receiver-based training data to Jüschke’s pre-distortion system rendered the claims obvious. Specifically, this combination taught updating model parameters based on a distance metric computed between the original transmit signal and a signal actually obtained from a radio receiver.
    • Motivation to Combine: A Person of Ordinary Skill in the Art (POSITA) would combine Jüschke and Holt to improve the accuracy of Jüschke’s training model. Both references address signal processing in wireless communications using neural networks. Petitioner contended that using actual received signal data, as taught by Holt, is a known technique for creating more accurate training data that accounts for real-world transmission and reception effects. This would have been seen as a simple and predictable substitution to improve the performance of Jüschke’s similar system.
    • Expectation of Success: A POSITA would have had a reasonable expectation of success because Holt's use of received and recovered signal samples would provide a more accurate data source for training than Jüschke’s internal transmitter feedback loop. Both systems were described for use in similar devices (e.g., base stations), making the integration straightforward.

Ground 2: Claims 2, 22, and 24 are obvious over Jüschke in view of Holt and Dzierwa.

  • Prior Art Relied Upon: Jüschke (EP 2,538,553 A1), Holt (Patent 10,552,738), and Dzierwa (Patent 10,122,479).
  • Core Argument for this Ground:
    • Prior Art Mapping: This ground was asserted as an alternative for claims 2, 22, and 24, which relate to updating model parameters for a "particular deployment scenario" or based on "environment condition[s]." Petitioner argued that the combination of Jüschke and Holt already rendered these claims obvious. However, to the extent that combination was found insufficient, Dzierwa was introduced. Dzierwa taught a machine learning system for observing an RF environment and using an "optimization module" to account for environmental parameters such as noise, antenna hardware, and communication protocols (e.g., LTE, CDMA). Petitioner argued that Dzierwa’s teachings on using environmental data to optimize signal processing would supply the limitations of claims 2, 22, and 24.
    • Motivation to Combine: A POSITA would be motivated to incorporate Dzierwa's environmental optimization into the Jüschke/Holt framework to further enhance the pre-distortion model's performance. Since all three references operate in the same field of machine learning for wireless communications, combining Dzierwa's technique for gathering more complete environmental data would be a recognized and predictable way to improve the functionality of Jüschke's "flexible radio transmitter." This would allow the model to adapt not just to the transmitted signal but also to the specific conditions of its deployment.
    • Expectation of Success: A POSITA would expect success in this combination because implementing Dzierwa's optimization module and environmental data into the Jüschke/Holt system would predictably improve performance by providing more complete and context-aware data for training the pre-distortion model.

4. Arguments Regarding Discretionary Denial

  • Petitioner argued that the Board should not exercise its discretion to deny institution.
  • Fintiv Factors: The petition asserted that Fintiv factors strongly favored institution, as the ’469 patent was not subject to any parallel district court litigation, and the petition presented strong grounds for invalidity.
  • Advanced Bionics Factors: The petition also argued that denial would be inappropriate under the Advanced Bionics factors because the prior art references relied upon (Jüschke, Holt, and Dzierwa) were not previously applied or considered during the original prosecution of the ’469 patent.

5. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-8, 10, 12-15, and 17-24 of the ’469 patent as unpatentable.