DETAILED ACTION
Remarks
Claims 1-20 have been examined and rejected. This Office Action is responsive to the continued examination request.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-20 are presented for examination.
Response to Amendment
Applicant’s amendment filed on 12/30/2025 has been entered. Claims 1, 10, 12, 15 and 20 are amended. Claims 1-20 are pending in the application.
Specification
The disclosure is objected to because of the following informalities:
Applicant should clarify the term “RF” in the entire Specification. There is no definition of RF included in the Specification.
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the Specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
With respect to claim 1 [line 5], claim 15 [line 5] and claim 20 [line 8], the limitation “a radio-frequency (RF) signal” is not disclosed in the Applicant’s Specification and is considered new matter. The Applicant’s Specification does not contain any description of a radio-frequency (RF) signal. At best, Applicant’s Specification indicates that the transmitter system 110 further comprises for each sensor node and each associated high-dimensional encoder a corresponding RF-transmitter 113, which may be also denoted as a RFmod. More particularly, the transmitter system 110 comprises a number S of RF-transmitters 1131 to II3s corresponding to the S sensor nodes 1111 to 111s and the S high-dimensional encoder nodes 1121 to 112s [par. 0034]. The RF-transmitters 113 are configured to transmit the high-dimensional vectors simultaneously via the transmission channel 120 over a respective link between the respective sensor node and the receiver system. The RF-transmitters 1131 to 113s may perform a physical modulation of the high-dimensional vectors V1, V2, ...VS onto a high- frequency carrier signal. According to embodiments, amplitude modulation (AM), frequency modulation (FM) or phase modulation (PM) may be used. In some embodiments, a link may be a physical link, an optical link, a wireless link, etc. [par. 0035]. For the purposes of examination, Examiner suggests Applicant to clearly recite the meaning of radio-frequency and provide examples of radio-frequency both in the claims and the Specification. As in the Specification, there is no paragraph includes a definition of RF, such that a person skilled in the art would not know what RF means.
With respect to claims 2-14 and 16-19, they are also rejected based on their virtual dependency of claims 1 and 15.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-6, 11-20 are rejected under 35 U.S.C. 103 as being unpatentable over Karunaratne et al (“In-memory hyperdimensional computing”) hereafter Karunaratne, in view of Olwal et al (US 20230230597 A1) hereafter Olwal, further in view of Chang et al (US 20130070677 A1) hereafter Chang, and further in view of Foukas et al (US 20220159785 A1) hereafter Foukas.
Karunaratne was cited in the IDS filed on 09/21/2021.
With respect to claim 1, Karunaratne teaches a sensor system for performing distributed sensing and classification of sensor data (hyperdimensional computing (HDC) has been employed in a wide range of applications, including machine learning (ML), cognitive computing, robotics and traditional computing, wherein some applications involve temporal patterns such as text classification, biomedical signal processing, and distributed sensors [page 1, I. Introduction]), the sensor system performing a method comprising:
encode the sensor data of each sensor node of a set of distributed sensor nodes for sensing the sensor data as high-dimensional vectors (the encoder receives an input text associated with a known language and generates a prototype hypervector corresponding to that language, wherein high-dimensional vectors are also referred as hypervectors [page 2, II. The Concept of In-memory HDC]);
superpose the high-dimensional vectors of the sensor data from the set of distributed sensor nodes by (in an example, the encoder would receive an input text associated with a language and would generate a prototype hypervector corresponding to that language. When the encoder receives n consecutive symbols, it produces an n-gram hypervector through a binding operation. The encoder then bundles several such n-gram hypervectors from the training data to produce a prototype hypervector [page 2, II. The Concept of In-memory HDC]).
However, Karunaratne does not particularly disclose transmit the high-dimensional vectors as a radio-frequency (RF) signal over a respective link between a sensor node and a receiver system, wherein the RF signal comprises high-frequency carrier signals modulated with the high-dimensional vectors; transmitting the vectors simultaneously over a transmission channel, such that the transmission channel physically adds the vectors together; and classify the superposed high-dimensional vectors by demodulating the RF-signal and comparing it to stored prototype vectors using a trained classifier to determine a classification label, wherein the classification is performed without transmission channel decoding.
In the same field of endeavor, Olwal teaches transmit the high-dimensional vectors as a radio-frequency (RF) signal over a respective link between a sensor node and a receiver system, wherein the RF signal comprises high-frequency carrier signals modulated with the high-dimensional vectors (the method for distributed sound recognition includes detecting audio data and transmitting, via a wireless connection, the audio data to a computing device in response to a sound of interest being detected. A radio-frequency (RF) transceiver configured to transmit the audio data to a computing device via a wireless connection. A carrier signal is a high-frequency electromagnetic wave generated at a specific frequency and amplitude. In some examples, the ML model is a sound classifier that can evaluate incoming sound for specific criteria (frequency, amplitude, feature detection, etc.) [par. 0004, 0005, 0007, 0046, 0047, 0066]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of transmitting via a wireless connection the sensor data to a computing device as suggested by Olwal into the concept of hyperdimensional computing that involves manipulation and comparison of large patterns within memory as suggested by Karunaratne because both of these systems addressing the process of classifying input data, such as sound, text or image, in machine learning applications in the memory computing environment. Doing so would be desirable because the system of Karunaratne would be more efficient by transmitting via a wireless connection the audio data to a computing device in response to the sound of interest being detected within the audio data, wherein the audio data is used by a ML model for further sound classification (Olwal, [par. 0004-0012]).
However, the combination of Karunaratne and Olwal does not specifically disclose transmitting the vectors simultaneously over a transmission channel, such that the transmission channel physically adds the vectors together; and classify the superposed high-dimensional vectors by demodulating the RF-signal and comparing it to stored prototype vectors using a trained classifier to determine a classification label, wherein the classification is performed without transmission channel decoding.
In the same field of endeavor, Chang teaches transmitting the vectors simultaneously over a transmission channel, such that the transmission channel physically adds the vectors together (Wavefront multiplexing technology is used to perform coherent power combining of the radiated signals to receive a strong and/or encoded signal. An aggregated vector may be generated with many components, wherein each component has a weight and coefficients associated with a vector. Many vectors (signals) may be transmitted simultaneously by multiplexing individual vectors and may be transmitted on a single channel [par. 0071 and FIG. 7]); and
demodulating the RF signal and comparing it to stored prototype vectors (the techniques of coherent power combining may also relates to demodulation of the signals in a receiver. The waveforms may be converted by performing frequency conversion or demodulation of the beam signals into data strings. These data strings are delivered to the mobile hubs 413 and terrestrial network 480. The GBBF facility receives the data or the information from these mobile hubs and/or terrestrial network after they perform modulation and channel formatting [par. 0045, 0054, 0055, 0066, 0071 and FIG. 5]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of using ground-based beam forming and wavefront multiplexing enhance the ability to coherently combine the power of the communication signals and to improve the signal-to-noise ratio as suggested by Chang into the combination of Karunaratne and Olwal because all of these systems addressing the process of transmitting signals relating to a prototype vector that is combined from multiple vectors. Doing so would be desirable because the combination of Karunaratne and Olwal would be more efficient by simultaneously transmitting multiple vectors (signals) over a single channel to get an aggregated vector that is represented as a wavefront using a technique called wavefront multiplexing (Chang, [par. 0071]).
However, the combination of Karunaratne, Olwal and Chang does not explicitly disclose classify the superposed high-dimensional vectors using a trained classifier to determine a classification label, wherein the classification is performed without transmission channel decoding.
In the same field if endeavor, Foukas teaches classify the superposed high-dimensional vectors using a trained classifier to determine a classification label, wherein the classification is performed without transmission channel decoding (a MAC UL dispatcher component may solve the timing related issues relating to uplink retransmission. The component may predict HARQ outcome without decoding the data. A demodulator component may send in-phase and quadrature (IQ) samples of the codeblocks to the MAC UL component. The MAC UL component may use the IQ samples of the codeblocks to make a prediction of whether the outcome for decoding the codeblocks without decoding the codeblocks using an error vector magnitude (EVM) classifier. The EVM classifier is a metric used in the physical layer uses the IQ samples to indicate the signal quality received from the base station [par. 0092-0094, 0132]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of offloading a signal processing task of a physical layer from a vRAN server located at the far edge of a network nearby a base station to a remote location further away from the base station as suggested by Foukas into the combination of Karunaratne, Olwal and Chang because all of these systems addressing the process of transmitting signals relating to a specific vector. Doing so would be desirable because the combination of Karunaratne, Olwal and Chang would be more efficient by making a prediction on the outcome for decoding codeblocks without decoding the codeblocks using a particular vector classifier such as the error vector magnitude (EVM) classifier (Foukas, [par. 0092-0094, 0132]).
With respect to claim 2, the combination of Karunaratne, Olwal, Chang and Foukas teaches the sensor system encodes the sensor data of each sensor node of the set of distributed sensor nodes as a unique quasi- orthogonal high-dimensional vector (Karunaratne, a set of quasi-orthogonal hypervectors are first selected to represent each symbol associated with a dataset, in the learning phase and the classification phase. During the learning phase of HDC, the basis hypervectors in the item memory (IM) are combined with the component-wise operations inside an encoder to compute a quasi-orthogonal n-gram hypervector representing an object of interest [page 2, II. The Concept of In-memory HDC]).
With respect to claim 3, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein the receiver system comprises an associative memory, the associative memory directly classifies the superposed high-dimensional vectors (Karunaratne, HDC represents the symbols with hypervectors those combined with binding, bundling and permutation, and then stored in associative memory (AM). The overall encoding operation results in c, d-dimensional prototype hypervectors (referred as associative memory) assuming there are c classes [page 1, I. Introduction & page 2, II. The Concept of In-memory HDC]).
With respect to claim 4, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein each sensor node comprises a corresponding high-dimensional encoder, the corresponding high-dimensional encoder encodes the sensor data by assigning a unique quasi-orthogonal high- dimensional vector to possible combinations of the sensor data (Karunaratne, hypervectors are defined as d-dimensional (pseudo)random vectors with independent and identically distributed components. If the dimensionality is in thousands, a large number of quasi-orthogonal hypervectors exist, this allow HDC converts these hypervectors into new hypervectors using well-defined vector space oprations, such that the resulting hypervector is unique with the same dimension [page 2, II. The Concept of In-memory HDC]).
With respect to claim 5, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein each high-dimensional encoder is randomly initialized such that its corresponding encoded high- dimensional vectors are quasi-orthogonal to the encoded high-dimensional vectors of one or more other sensor nodes (Karunaratne, HDC is remarkably robust to random variability and device failures. However, given the holographic nature of the hypervectors, this can be address easily by a random partitioning approach. A coarse-grained randomization strategy is used to segment the prototype hypervector and to place the resulting segments spatially distributed across the crossbar array [page 5, III. The Associative Memory Search Module]).
With respect to claim 6, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein;
each high-dimensional encoder generates a D-bit high- dimensional vector (Karunaratne, the operations used on hypervectors produce a d-bit hypervectors that is resulted in a closed system [page 2, II. The Concept of In-memory HDC]); and
the sensor system transmits the D-bit high-dimensional vector directly without any further encoding, parity, and transformation (Karunaratne, an encoder performs dimensionality, preserving mathematical manipulations on the basis hypervectors to produce c, d-dimensional prototype hypervectors those are stored in AM. The prototypes are transmitted directly to the AM directly without any further encoding, or the query hypervector to the distance computation [page 2, II. The Concept of In-memory HDC & Fig. 1]).
With respect to claim 11, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein the links are selected from a group consisting of wireless links, optical links and electrical links (Olwal, the device is connected to the computing device via a wireless connection, wherein the wireless connection is a short-range communication link such as near-field communication connection or Bluetooth connection. The device may include some sensors such as a microphone to capture audio data, imaging sensors to capture image data. The microphone is a transducer device that converts sound into an electrical signal [par. 0047, 0057]).
With respect to claim 12, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein the associative memory comprises for each sensor node a separate associative sub-memory (Karunaratne, assuming there are c classes inside the associative memory, wherein each class associated with a prototype hypervector, a class is associated with a sub-memory in the associative memory [page 2, II. The Concept of In-memory HDC & Fig. 1]).
With respect to claim 13, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein the associative memory classifies the high-dimensional vectors by a single pass (Karunaratne, the training algorithm of HDC works in a few ways, such as object categories are learned from one or few examples, and in a single pass over the training data as opposed to many iterations [page 1, I. Introduction]).
With respect to claim 14, the combination of Karunaratne, Olwal, Chang and Foukas teaches wherein the associative memory is trained without considering noise on the links as well as with considering noise on the links between the sensor nodes and the receiver system (Karunaratne, HDC those are combined with preserved operations to generate prototype hypervectors are stored in the AM. This chain implies that failure in a component of a hypervector is not contagious and forms a computational framework that is robust and immune to defects, variations and noise [page 1, I. Introduction]).
With respect to claim 15, it is a computer-implemented method that is corresponding to the sensor system of claim 1. Therefore, it is rejected for the same reason as claimed in claim 1 above.
With respect to claim 16, it is a computer-implemented method that is corresponding to the sensor system of claim 2. Therefore, it is rejected for the same reason as claimed in claim 2 above.
With respect to claim 17, it is a computer-implemented method that is corresponding to the sensor system of claim 3. Therefore, it is rejected for the same reason as claimed in claim 3 above.
With respect to claim 18, it is a computer-implemented method that is corresponding to the sensor system of claim 6. Therefore, it is rejected for the same reason as claimed in claim 6 above.
With respect to claim 19, it is a computer-implemented method that is corresponding to the sensor system of claim 14. Therefore, it is rejected for the same reason as claimed in claim 14 above.
With respect to claim 20, it is a computer program product comprising a computer readable storage medium that is corresponding to the sensor system of claim 1. Therefore, it is rejected for the same reason as claimed in claim 1 above.
Claims 7 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Karunaratne et al (“In-memory hyperdimensional computing”) hereafter Karunaratne, in view of Olwal et al (US 20230230597 A1) hereafter Olwal, further in view of Chang et al (US 20130070677 A1) hereafter Chang, as applied in claim 4 above, further in view of Salman et al (US 10678511 B1) hereafter Salman, and further in view of Foukas et al (US 20220159785 A1) hereafter Foukas.
Karunaratne was cited in the IDS filed on 09/21/2021.
With respect to claim 7, the combination of Karunaratne, Olwal, Chang and Foukas teaches all limitations as claimed in claim 4 above.
However, the combination of Karunaratne, Olwal, Chang and Foukas does not disclose wherein; each high-dimensional encoder is embodied as a cellular automaton.
In the same field of endeavor, Salman teaches wherein; each high-dimensional encoder is embodied as a cellular automaton (a method for using cellular automata to generate quality pseudo-random numbers that may be used in cryptographic and other applications. A cellular automaton is a decentralized computing model that enables the performance of complex computations with local information [col. 1, lines 20-45]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of generating quality pseudo-random numbers that is used in cryptographic and other applications as suggested by Salman into the combination of Karunaratne, Olwal, Chang and Foukas because all of these systems addressing the process of generating pseudo-random variables (numbers or vectors) in a specific machine learning application. Doing so would be desirable because the combination of Karunaratne, Olwal, Chang and Foukas would be more efficient by using cellular automata in generating pseudo-random numbers, which is a decentralized computing model that enables the performance of complex computations that is useful in some applications, such as Monte Carlo simulations, communications, cryptography and network security (Salman, [col. 1, lines 20-45]).
With respect to claim 8, the combination of Karunaratne, Olwal, Chang, Foukas and Salman teaches wherein; the cellular automaton is a rule 30 automaton (Salman, chaotic rule 30 is combined with a one-dimensional cellular automaton at a present state of neighborhood of time step to produce the next state cell of time step [col. 2, lines 30-50 & FIG. 2]).
Claims 9 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Karunaratne et al (“In-memory hyperdimensional computing”) hereafter Karunaratne, further in view of Olwal et al (US 20230230597 A1) hereafter Olwal, further in view of Chang et al (US 20130070677 A1) hereafter Chang, as applied in claim 1 above, further in view of Jiang et al (“High-dimensional Channel Estimation for Simultaneous Localization and Communications”) hereafter Jiang, and further in view of Foukas et al (US 20220159785 A1) hereafter Foukas.
Karunaratne was cited in the IDS filed on 09/21/2021.
With respect to claim 9, the combination of Karunaratne, Olwal, Chang and Foukas teaches all limitations as claimed in claim 1 above.
However, the combination of Karunaratne, Olwal, Chang and Foukas does not disclose wherein the sensor system transmits the high-dimensional vectors via single path propagation.
In the same field of endeavor, Jiang teaches wherein the sensor system transmits the high-dimensional vectors via single path propagation (3-dimensional is considered for localization and communications. Both transmitter and receiver are equipped with uniform rectangular arrays (URA). Although multi-path is used in this multidimensional channel model, single path is another scenario in the model, wherein each propagation path is associated with angles-of-departure, angles-of-arrival, propagation delay, and complex gain [page 2, II. System Model]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have incorporated the concept of low-complexity high-dimensional channel estimation approach as suggested by Jiang into the combination of Karunaratne, Olwal, Chang and Foukas because all of these systems addressing the process of generating and estimating the signals from a transmitter to a receiver in a high-dimensional system. Doing so would be desirable because the combination of Karunaratne, Olwal, Chang and Foukas would be more efficient by using both the spatial smoothing and forward-backward averaging techniques to explore data samples to extract multipath components (Jiang, [page 1, I. Introduction]).
With respect to claim 10, the combination of Karunaratne, Olwal, Chang, Foukas and Jiang teaches wherein the sensor system transmits the high-dimensional vectors via multi-path propagation, wherein the receiver system performs a permutation operation on the high- dimensional vectors received via the multi-path propagation to align the received high- dimensional vectors (Jiang, multi-path propagation scenario is used in the multidimensional channel model, wherein each propagation path is associated with angles-of-departure, angles-of-arrival, propagation delay, and complex gain [page 2, II. System Model]).
Response to Arguments
The examiner respectfully acknowledges the applicant’s amendments to claims 1, 10, 12, 15 and 20.
Applicant’s amendments filed on 12/30/2025 regarding the objections to claims 10 and 12 have been considered and are consequently withdrawn.
Applicant’s amendments filed on 12/30/2025 regarding the claim rejections to claims 1-20 under 35 USC 112(b) have been considered and are withdrawn.
Applicant’s arguments filed on 12/30/2025 regarding the claim rejections to claims 1-20 under 35 USC 103 have been fully considered and moot in view of new ground of rejection (see rejection above).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Sugano et al (US 8264616 B2) disclosed a scene classification apparatus for classifying uncompressed or compressed video into various types of scenes at low cost and with high accuracy using characteristics of a video and audio characteristics accompanied by the video. When video are compressed data, their motion intensity, spatial distribution of motion and histogram of motion direction are detected by using values of motion vectors of predictive coding images existing in respective shots, and the respective shots of the video are classified into a dynamic scene, a static scene, a slow scene, a highlight scene, a zooming scene, a panning scene, a commercial scene and the like based on the motion intensity, the spatial distribution of motion, the histogram of motion direction and shot density.
Sutherland et al (US 12393677 B2) disclosed systems and methods for metadata processing. In some embodiments, one or more metadata inputs may be processed to determine whether to allow an instruction. For instance, one or more classification bits may be identified from a metadata input of the one or more metadata inputs, and the metadata input may be processed based on the one or more classification bits.
Nammi et al (US 20210111740 A1) disclosed subject, for example, obtaining a received channel-encoded data block having information bits, a transmitted error-check value, and redundant code bits. The redundant code bits correspond to a channel code applied to the received channel-encoded data block prior to transmission via a communication channel. A channel code type is identified and responsive to it being systematic, the information bits and the transmitted error-check value are obtained without decoding according to the channel code. The received channel-encoded data block is checked according to the transmitted error-check value to obtain a result. Responsive to the result not indicating an error, extracting the information bits without decoding the received channel-encoded data block according to the channel code. Responsive to the result indicating an error, decoding the received channel-encoded data block according to the channel code to obtain decoded information bits.
Argyropoulos et al (US 20130271668 A1) disclosed a method for assessing the quality of a transmitted video signal sequence at a receiver side includes: capturing the input video bit stream; extracting at least one feature or a set of features; determining the continuous probability of visibility for each packet loss event which occurred within a specific time interval; and employing the continuous probability of packet loss visibility as a weighting factor of the at least one feature or set of features extracted from the video bit stream to calculate an estimate of the overall quality, Q, of the transmitted video sequence.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Quoc Phung whose telephone number is (703) 756 1330. The examiner can normally be reached on Monday through Friday from 9am to 5pm PT.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) athttp://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jennifer Welch can be reached on 571-272-7212. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Q.L.P./Examiner, Art Unit 2143
/JENNIFER N WELCH/Supervisory Patent Examiner, Art Unit 2143
Prosecution Insights