Obviousness

Obviousness Analysis of US Patent 7676007 Under 35 U.S.C. § 103

This analysis identifies combinations of prior art references that would render the claims of US Patent 7676007 obvious to a person having ordinary skill in the art (POSITA) at the time of the invention (priority date July 21, 2004). The core innovation claimed by US7676007 lies in addressing the phase ambiguity inherent in optimal beamforming vectors (or the non-uniqueness of precoding matrices) when performing interpolation for limited feedback MIMO-OFDM systems.

General Motivation for Combination

A POSITA in wireless communications would be continually motivated to improve spectral efficiency and reliability while reducing feedback overhead in MIMO-OFDM systems. Given the known correlation of channel conditions across adjacent subcarriers in OFDM, using subsampling and interpolation to reduce feedback for channel state information (CSI) (such as beamforming vectors or precoding matrices) would be a natural and expected engineering approach. The challenge addressed by the patent is the effective interpolation of these vectors/matrices despite their inherent phase invariance or unitary non-uniqueness. A POSITA would recognize this challenge and seek methods to enhance interpolation accuracy by addressing this ambiguity.

Combination of Prior Art for Obviousness

The obviousness of US7676007 stems from combining the following known concepts and the explicit problem identification within the patent itself:

Combination 1: For Beamforming Claims (e.g., Claims 1, 19, 40, 41)

1. MIMO-OFDM Systems with Transmit Beamforming and Full Feedback: It was well-established in the prior art to utilize Multiple-Input Multiple-Output (MIMO) systems with Orthogonal Frequency Division Multiplexing (OFDM) to exploit diversity and array gain over frequency-selective channels. In such systems, transmit beamforming with receive combining was a known technique to maximize Signal-to-Noise Ratio (SNR) or capacity. For non-reciprocal channels (e.g., FDD systems), the receiver would estimate the channel and feed back optimal beamforming vectors for each OFDM subcarrier to the transmitter. [cite: Source: Patent Detailed Description, "Background" and "MIMO-OFDM Systems with Limited Feedback" sections] References such as Dighe (P. A. Dighe et al., 2003) and Tse (C.-H. Tse et al., 2000), cited in the patent, describe aspects of MRT/MRC which forms the basis for optimal beamforming. [cite: Source: Patent Detailed Description, "MIMO-OFDM Systems with Limited Feedback"]
2. Reducing Feedback via Subsampling and Interpolation: The significant feedback overhead associated with sending beamforming vectors for all subcarriers in MIMO-OFDM was a recognized problem. A POSITA would be motivated to reduce this overhead. Given the inherent correlation between adjacent OFDM subchannels (explicitly acknowledged in the patent as the basis for feedback reduction: "the neighboring subchannels of OFDM is substantially correlated"), it would be an obvious solution to subsample the subcarriers, send feedback only for a subset of them, and then use interpolation at the transmitter to estimate the beamforming vectors for the un-fed-back subcarriers. [cite: Source: Patent Detailed Description, "Further Description"]
3. Spherical Interpolation for Unit Vectors: Beamforming vectors are typically unit-norm complex vectors. Methods for interpolating unit vectors on a sphere, known as spherical interpolation, were established in the art. The patent itself cites references like Buss (S. R. Buss and J. P. Fillmore, 2001), Shoemake (K. Shoemake, 1985), and Watson (G. S. Watson, 1983) for spherical interpolators. [cite: Source: Patent Detailed Description, "One Embodiment of the Spherical Interpolator 305"] A POSITA would naturally consider applying these techniques to interpolate beamforming vectors.
4. Addressing Phase Invariance in Beamforming Vectors: The patent explicitly identifies a critical challenge when applying conventional spherical interpolators to beamforming vectors: "when w(k) is the optimal beamforming vector maximizing the effective channel gain, e jΦ w(k) also maximizes the effective channel gain. In other words, the optimal beamforming vector is not a unique point but a line on the unit sphere." [cite: Source: Patent Detailed Description, "One Embodiment of the Spherical Interpolator 305"] A POSITA, recognizing this fundamental property of beamforming vectors and the limitations it imposes on direct interpolation (e.g., leading to arbitrary interpolation paths and distortion if phases are not aligned), would be motivated to introduce a phase alignment mechanism. It would be obvious to incorporate a phase rotation parameter (θl) to "remove the distortion caused by the arbitrary phase rotation of the optimal beamforming vectors" [cite: Source: Patent Detailed Description, "One Embodiment of the Spherical Interpolator 305"] prior to or as part of the interpolation process. Optimizing this phase parameter at the receiver to maximize standard performance metrics, such as minimum effective channel gain or capacity (as described by equations (5), (6), (12), and (13) in the patent), would be a routine optimization task for a skilled engineer. Feeding back these phase values as "interpolation information" along with the subsampled beamforming vectors would be a logical step to enable the transmitter to perform the improved interpolation.

Therefore, the claimed features of using phase values as interpolation information for beamforming vectors to reduce distortion during interpolation, particularly in a spherical interpolation context for MIMO-OFDM with partial feedback, would be obvious.

Combination 2: For Precoding Claims (e.g., Claims 25, 28, 42, 43)

1. MIMO-OFDM Systems with Precoding and Full Feedback: Similar to beamforming, MIMO-OFDM systems employing linear precoding for spatial multiplexing were known. Precoding matrices are chosen based on channel information to improve spectral efficiency and robustness. In non-reciprocal channels, feedback of precoding matrices or related CSI from the receiver to the transmitter was necessary. [cite: Source: Patent Detailed Description, "Overall System Summary" and "FIG. 6 is a block diagram illustrating a multiple-input multiple-output—orthogonal frequency division multiplex (MIMO-OFDM) communication system 600 with precoding."]
2. Reducing Feedback via Subsampling and Interpolation for Precoding: Just as with beamforming, the overhead of feeding back precoding matrices for all subcarriers would motivate a POSITA to reduce feedback. Exploiting the correlation of precoding matrices on adjacent subcarriers by sending a subset and interpolating at the transmitter would be an obvious extension of the feedback reduction techniques known for beamforming. [cite: Source: Patent Detailed Description, "Overall System Summary"]
3. Addressing Non-Uniqueness/Derotation in Precoding Matrices: The patent explicitly points out that optimal precoding matrices exhibit a "performance invariant to the right multiplication of the orthogonal matrix" (e.g., MSE(W(k), H(k))=MSE(W(k)Q, H(k))). [cite: Source: Patent Detailed Description, "The system 600 has a tone model"] This is directly analogous to the phase invariance issue for beamforming vectors. A POSITA would recognize that such non-uniqueness would cause similar interpolation problems to phase ambiguity. Therefore, it would be obvious to introduce a unitary derotation matrix (Q) to align the precoding matrices (as described in the patent's equation (22) for "derotated interpolation" and the concept of "a unitary derotation matrix Q is associated with precoder interpolation" [cite: Source: Patent Detailed Description, "Overall System Summary"]) before or during interpolation to ensure consistent and accurate reconstruction. Optimizing this derotation matrix Q from a finite codebook (as per equations (23) and (24) in the patent) based on performance measures like MMSE or capacity maximization would be a routine optimization for a skilled artisan, and then feeding this Q as part of the limited feedback is a logical implementation.

Therefore, the claimed features of using derotation (via matrix multiplication or unitary matrices) as interpolation information for precoding vectors/matrices to reduce distortion during interpolation, particularly in a MIMO-OFDM system with partial feedback, would be obvious.

In both beamforming and precoding scenarios, the explicit recognition of the phase/unitary ambiguity problem by the patent itself, combined with the known techniques for feedback reduction, interpolation, and optimization of wireless system parameters, would have led a POSITA to the claimed solution. The various mathematical approaches to finding the optimal phase or derotation matrix (e.g., grid search or closed-form solutions as detailed in the patent) are standard optimization methodologies.