Obviousness

Obviousness Analysis of US Patent 8,441,721 under 35 U.S.C. § 103

This analysis addresses the obviousness of US Patent 8,441,721 (hereinafter '721 patent) under 35 U.S.C. § 103, considering the provided prior art. The '721 patent generally relates to systems and methods for controlling Raman amplifier pump power by projecting measured deviations into a space of correctable changes.

Claims at Issue

The independent claims of the '721 patent define the scope of the invention:

• Claim 1 describes a method including receiving measured channel powers, determining deviations, projecting the deviations into a space that defines Raman gain profiles achievable, and determining power setting values based on the projected deviations.
• Claim 9 describes a method including removing a feature which cannot be corrected from measured deviations and optimizing power setting values based on the space of correctable changes.
• Claim 13 describes a method including receiving power measurements, creating deviation measurements, determining correctable deviations (as a correctable portion of a partition of deviations), and determining a pump setting based on these correctable deviations.
• Claims 19, 20, 23, and 24 are apparatus and computer-readable medium claims corresponding to the method claims, incorporating the core features of projecting or removing uncorrectable deviations.

The core innovative aspect across these claims lies in identifying and isolating the "correctable" portion of the deviation measurements by projecting them into a subspace defined by the Raman gain profiles achievable by the pumps, or equivalently, by removing features that cannot be corrected by the Raman pumps.

Primary Prior Art Reference

US Patent Application Publication 2004/0036954 (Freund) is highly relevant prior art. The '721 patent itself explicitly states that "a linear programming method of determining optimal pump configurations may be utilized for broadband Raman amplifier control. In one embodiment employing this type of Raman amplifier control, disclosed by the teachings of U.S. Patent Application Publication 2004/0036954... a linear program based on a linearized model of Raman gain is created. The linear program can be solved using a standard linear program solver and the solution provides Raman pump settings to obtain an optimal gain profile."

Freund thus teaches:

• Controlling pump powers of broadband Raman amplifiers.
• Receiving optical channel power measurements.
• Comparing these measurements to target powers to determine deviations.
• Using an optimization algorithm (specifically, linear programming) based on a linearized model of Raman gain to determine optimal Raman pump settings.

These elements align with the initial steps of the '721 patent's claims, such as receiving measured channel powers, determining deviations, and determining power setting values for pump lasers. Other prior art, such as US20070258132A1 (AT&T) and US20030151799A1 (Innovance), also teach similar feedback dynamic gain control for WDM systems employing Raman amplifiers, often utilizing linear programming or other optimization techniques based on signal power profiles.

Motivation to Combine and Obviousness

The '721 patent identifies a critical problem with conventional Raman amplifier control schemes, including those like Freund's: "a single channel power measurement error may cause conventional linear programming methodologies of Raman amplifier control to fail to converge to a suitable solution." The '721 patent further notes that "adjustments to the Raman pump lasers cannot produce an arbitrary gain profile for the channels" and that "a channel power distribution which cannot be corrected by adjusting the Raman pump powers can cause significant convergence difficulties, or worse, cause the system to converge to an incorrect pump setting."

A Person Having Ordinary Skill in the Art (PHOSITA) working on Raman amplifier control systems, such as those described by Freund, would be aware of these inherent limitations: that Raman pumps have a limited ability to shape the gain profile and that not all desired or measured gain deviations can be perfectly corrected. Facing problems like poor convergence, instability, or suboptimal solutions caused by uncorrectable error components (e.g., measurement noise or errors), a PHOSITA would be motivated to improve the robustness and efficiency of the control algorithm.

The linearized Raman gain model, which Freund explicitly uses, inherently defines the "space of achievable Raman gain profiles" for a given set of channels and pumps. In linear algebra, projecting a vector onto a subspace is a well-known mathematical technique for isolating the components of that vector that lie within the subspace, effectively removing any components that are orthogonal (uncorrectable by changes within that subspace). Techniques like Gram-Schmidt orthogonalization for finding orthonormal bases and constructing projection matrices are standard mathematical tools.

Therefore, a PHOSITA, seeking to improve the stability and convergence of Freund's Raman amplifier control system in the face of uncorrectable deviations, would find it obvious to apply a known mathematical technique of projecting the measured deviations onto the subspace of achievable Raman gain profiles. This would involve:

1. Utilizing Freund's system for receiving measured channel powers, determining deviations, and using an optimization problem to determine pump settings.
2. Interpreting Freund's "linearized model of Raman gain" as defining the "space that defines Raman gain profiles achievable with a set of channels and pump lasers."
3. Applying a projection technique (common general knowledge in mathematics/signal processing) to the deviation measurements before feeding them into Freund's optimization algorithm. This projection would isolate the "correctable deviations" or "remove a feature which cannot be corrected by the adjustment of a plurality of Raman pumps," as described in the '721 patent. The remaining "projected deviations" or "correctable changes" would then be used by the optimization algorithm to determine pump settings.

This combination directly addresses the problems acknowledged by the '721 patent (convergence difficulties, susceptibility to measurement errors, attempts to correct uncorrectable features) by logically applying a known mathematical solution to a known problem within the specific context of Raman amplifier control.

Conclusion

The independent claims (1, 9, 13, 19, 20, 23, and 24) of US Patent 8,441,721 would be rendered obvious by a combination of US20040036954A1 (Freund) and the common general knowledge of a PHOSITA regarding linear algebra and control system design. The motivation for such a combination arises from the recognized limitations of Raman gain shaping and the desire to improve the stability, robustness, and convergence of iterative Raman pump control algorithms, especially in the presence of uncorrectable errors or noise. A PHOSITA would understand that projecting error signals onto the subspace of physically achievable corrections is a standard approach to address such issues in control systems.