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 .
DETAILED ACTION
This Office Action is in response to the Applicants’ communication filed on 07/17/2024. In virtue of this communication, claims 1-20 are currently pending in the instant application.
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 of this title, 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 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Parichehrehteroujeni et al. (US 2023/0115368 A1), hereinafter Ericsson in view of DOMINGOS PEDRO PEDROD@CS WASHINGTON EDU: "A few useful things to know about machine learning", ARXIV:2003.08934V1, UNITED STATES, vol. 55, no. 10, 1 October 2012 (2012-10-01), pages 78-87, XP058627688,
DOI: 10.1145/2347736.2347755 (provided by Applicant and hereinafter Domingos).
Regarding Claim 1 Ericsson teaches the limitations "A terminal device, comprising: a processor and a memory, wherein the memory is configured to store a computer program, the processor is configured to call the computer program stored in the memory and run the computer program, so as to enable the terminal device to perform:
switching a network model for implementing a first function from a first model to a second model in a case where a computing resource of the terminal device is insufficient, wherein an input of the first model is a first information set, and an input of the second model is a second information set; or, switching an algorithm for implementing a first function from a first algorithm to a second algorithm in a case where a computing resource of the terminal device is insufficient, wherein an input of the first algorithm is a first information set, and an input of the second algorithm is a second information set; (see fig. 20 and par. 0310 and 0312, where the UE changes the AI/ML model when a computing resource is insufficient (“block 2020, where the network node can select the AI/ML predictive model from a plurality of available model types based on one or more of the following criteria: wireless network capabilities, UE capabilities, model size and/or complexity, severity of random access problems in the cell, available inputs, necessary and/or desirable outputs, need for retraining the model” and the UE sends “an indication that the provided AI/ML predictive model needs to be retrained, and a request for a further training dataset for retraining the model. In such embodiments, the exemplary method can also include the operations of either block 2070 or block 2080. In block 2070, the network node can send a second portion of the training dataset to the particular UE. In block 2080, the network node can retrain the AI/ML predictive model based a second portion of the training dataset and send the retrained AI/ML predictive model to the particular UE.” Here the second model is equated to the updated model and the computing resource is equated to the “UE capabilities, model size and/or complexity, severity of random access problems in the cell, available inputs, necessary and/or desirable outputs.”);
and the first function comprises at least one of: downlink channel estimation, downlink beam management, downlink positioning, channel state information (CSI) feedback, or mobility management" (see par. 0291 “FIG. 20 shows a flow diagram of an exemplary method (e.g., procedure) for a network node to configure random access by one or more UEs in a cell of the wireless network,” and par. 0292 where “an AI/ML predictive model that includes one or more input parameters and corresponding one or more output parameters that are associated with random-access configurations for the cell.”
However, Ericsson does not explicitly disclose the limitation “wherein complexity of the second model is lower than complexity of the first model, complexity of the second algorithm is lower than complexity of the first algorithm, an amount of information comprised in the second information set is greater than an amount of information comprised in the first information set.”
Ericsson does show that the predicative models have varying model size and complexity (see par. 0048 and 0289 “The type of model to be used can be based on the network capabilities, UE capabilities, model size and/or complexity, severity of random access problems in the cell, available inputs, necessary and/or desirable outputs, need for retraining, etc. In case the model is provided to UEs for execution, the network needs to trade-off the overhead in providing the model to UEs in the cell versus the benefits of having the model at the UE. As another example, the necessary size and/or complexity of the model may be dependent on the severity of RACH collisions in cells of the network. As another example, if the UE should re-train the model, neural-networks are able to do updates without requiring the entire training dataset, in contrast to tree-based methods such as random forest.” (Emphasis added). Here, this shows Ericsson teaches models of various complexity and also second input data (i.e. training data) that can be variable as well depending on the model and UE needs.
In the same field of endeavor Domingos discloses simple or less complex models with large amounts of input data can outperform more complex models based on less input data (see last par. On pg. 84-first par. On pg. 85 “More Data Beats a Cleverer Algorithm”: “As a rule of thumb, a dumb algorithm with lots and lots of data beats a clever one with modest amounts of it.”.
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to use a less complex model with more data as taught by Domingos in the system of Ericsson who shows selecting varying complexity models with varying input training data, in order to provide the most efficient way to achieve desire output results where “the quickest path to success is often to just get more data….after all, machine learning is all about letting data do the heavy lifting.”
Claims 10 and 19 are rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 2 Ericsson teaches the limitations "The terminal device according to claim 1, wherein the terminal device further performs:
receiving first information and second information respectively, wherein the first information is determined based on an available computing resource of the terminal device within a first duration, and the first information is used to configure first downlink reference signal resource information; and (see par. 0151 “the NR PHY includes various reference signals (RS) such as synchronization signal/PBCH block (SSB), channel state information RS (CSI-RS), tertiary RS, positioning RS (PRS), demodulation reference signals (DM-RS), phase-tracking RS (PTRS), etc. In general, SSB is available to all UEs regardless of RRC state, while other RS (e.g., CSI-RS, DM-RS, PTRS) are associated with specific UEs that have a network connection, i.e., in RRC_CONNECTED state.” This shows downlink reference signals.);
the second information is determined based on an available computing resource of the terminal device within a second duration, and the second information is used to configure second downlink reference signal resource information; and (see fig. 20 (2050 and 2070) showing first and second information at first and second durations and par. 0288 “the network can determine that RACH configurations for a cell need to be improved and/or optimized, such as by monitoring the number of RACH collisions reported by UEs. For example, if more than N failed RACH attempts have occurred in a predetermined duration, T, the network can train an AI/ML model based on any of the inputs discussed above. The newly trained model can then be executed by a RAN node or provided to UEs operating in the cell for execution by the UEs.”);
However, Ericsson does not explicitly disclose “the available computing resource within the first duration is more than the available computing resource within the second duration; and acquiring the first information set according to the first downlink reference signal resource information, and acquiring the second information set according to the second downlink reference signal resource information"
Ericsson does show models are based on UE capabilities (i.e. computing resources) (see par. 0289 “Various types of AI/ML models can be used according to different embodiments of the present disclosure. Some exemplary model types include decision trees, random forest, feed forward neural networks, and convolutional neural networks. The type of model to be used can be based on the network capabilities, UE capabilities, model size and/or complexity, severity of random access problems in the cell, available inputs, necessary and/or desirable outputs, need for retraining, etc.”).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention that the computing resources or UE capabilities can decrease (i.e. less resources in a second duration) in the system of Ericsson who shows selecting varying complexity models with varying input training data, in order to compensate for various network conditions such as increase RACH collision and varying UE capabilities (see e.g. par. 0289 of Ericsson).
Claims 11 and 20 are rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 3 Ericsson teaches the limitations "The terminal device according to claim 2, wherein the first information is further used to configure the first model, and/or, the second information is further used to configure the second model; or, the first information is further used to configure the first algorithm, and/or, the second information is further used to configure the second algorithm" (see fig. 20 steps 2040, 2050 and 2070 and 2080, showing the first and second information are used to train first and second models).
Claim 12 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 4 Ericsson teaches the limitations "The terminal device according to claim 2, wherein the terminal device further performs:
determining the first model according to the available computing resource of the terminal device within the first duration, and/or, determining the second model according to the available computing resource of the terminal device within the second duration; or determining the first algorithm according to the available computing resource of the terminal device within the first duration, and/or, determining the second algorithm according to the available computing resource of the terminal device within the second duration" (see par. 0048 and 0289 “The type of model to be used can be based on the network capabilities, UE capabilities, model size and/or complexity, severity of random access problems in the cell, available inputs, necessary and/or desirable outputs, need for retraining, etc. In case the model is provided to UEs for execution, the network needs to trade-off the overhead in providing the model to UEs in the cell versus the benefits of having the model at the UE. As another example, the necessary size and/or complexity of the model may be dependent on the severity of RACH collisions in cells of the network. As another example, if the UE should re-train the model, neural-networks are able to do updates without requiring the entire training dataset, in contrast to tree-based methods such as random forest.” (Emphasis added). Here, this shows Ericsson teaches models of various complexity and also second input data (i.e. training data) that can be variable as well depending on the model and UE needs.
Claim 13 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 5 Ericsson teaches the limitations "The terminal device according to claim 2, wherein the available computing resource of the terminal device is determined by at least one of: computing power of the terminal device, or a computing power level of the terminal device" (see par. 0263 “the UE can retrain the AI/ML model if the initial transmission power level produced by the model did not produce a desirable result. This could occur, for example, if the UE failed in a first RACH attempt without detecting any congestion, presumably due to inadequate initial preamble transmission power.”).
Claim 14 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 6 Ericsson teaches the limitations "The terminal device according to claim 5, wherein a correspondence between the available computing resource of the terminal device and the computing power of the terminal device is determined by at least one of: protocol agreement, pre-definition, pre-configuration of an operator, configuration of a network device, or configuration of the terminal device; and/or, a correspondence between the available computing resource of the terminal device and the computing power level of the terminal device is determined by at least one of: protocol agreement, pre-definition, pre-configuration of an operator, configuration of a network device, or configuration of the terminal device" (see abstract “a network node to configure random access by one or more user equipment, UEs, in a cell of the wireless network. Such methods include providing (2040) one of the following to one or more UEs operating in the cell: an artificial intelligence/machine learning, AI/ML, predictive model that includes one or more input parameters and corresponding one or more output parameters that are associated with random-access configurations for the cell; or one or more random-access configurations for the cell, each random-access configuration associated with one or more values of output parameters of the AI/ML predictive model. Such methods include detecting (2090) a random access to the cell, by a particular UE, according to a particular random-access configuration associated with particular values of the output parameters.” This shows the preconfiguration by network node).
Claim 15 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 7 Ericsson teaches the limitations "The terminal device according to claim 5, wherein the computing power of the terminal device includes fixed computing power and real-time computing power" (see par. 0023 “When such collisions occur, the network may not correctly receive a UE's random-access preamble transmissions, causing the UE to attempt retransmission at a higher power level, referred to as “power ramping”.” This shows fixed power level at initial transmission and real-time power level at retransmission during power ramping.
Claim 16 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 8 Ericsson teaches the limitations "The terminal device according to claim 2, wherein the terminal device further performs:
sending third information and fourth information respectively; wherein the third information comprises at least one of: the available computing resource of the terminal device within the first duration, computing power of the terminal device within the first duration, or a computing power level of the terminal device within the first duration; and the fourth information comprises at least one of: the available computing resource of the terminal device within the second duration, computing power of the terminal device within the second duration, or a computing power level of the terminal device within the second duration" (see par. 0150 “UL physical channels Physical Uplink Shared Channel (PUSCH), Physical Uplink Control Channel (PUCCH), and Physical Random Access Channel (PRACH). PUSCH is the UL counterpart to the PDSCH, used by UEs to transmit UL control information (UCI) including HARQ feedback for DL transmissions, channel quality feedback (e.g., CSI) for the DL channel, scheduling requests (SRs), etc. PRACH is used for random access preamble transmission.” Also see par. 0166 “The msg 3 is the UE's first scheduled uplink transmission on the PUSCH. It conveys an actual RRC procedural message, such as an RRCConnectionRequest, and RRCResumeRequest, etc. It is addressed to the temporary C-RNTI allocated in RAR during step 2 and carries the C-RNTI or an initial UE identity.”). Also (see par. 0263 “the UE can retrain the AI/ML model if the initial transmission power level produced by the model did not produce a desirable result. This could occur, for example, if the UE failed in a first RACH attempt without detecting any congestion, presumably due to inadequate initial preamble transmission power.” This shows the UE requesting updated model parameters is based upon insufficient power level and this is informed with request.
Claim 17 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Regarding Claim 9 Ericsson teaches the limitations "The terminal device according to claim 8, wherein the third information is uplink control information (UCI), or the third information is a radio resource control (RRC) message, wherein the third information is carried by at least one of: an uplink data channel, an uplink control channel, an uplink artificial intelligence data transmission channel, an uplink multicast channel, or an uplink broadcast channel; and/or, the fourth information is UCI, or the fourth information is an RRC message, wherein the fourth information is carried by at least one of: an uplink data channel, an uplink control channel, an uplink artificial intelligence data transmission channel, an uplink multicast channel, or an uplink broadcast channel" (see par. 0150 “UL physical channels Physical Uplink Shared Channel (PUSCH), Physical Uplink Control Channel (PUCCH), and Physical Random Access Channel (PRACH). PUSCH is the UL counterpart to the PDSCH, used by UEs to transmit UL control information (UCI) including HARQ feedback for DL transmissions, channel quality feedback (e.g., CSI) for the DL channel, scheduling requests (SRs), etc. PRACH is used for random access preamble transmission.” Also see par. 0166 “The msg 3 is the UE's first scheduled uplink transmission on the PUSCH. It conveys an actual RRC procedural message, such as an RRCConnectionRequest, and RRCResumeRequest, etc. It is addressed to the temporary C-RNTI allocated in RAR during step 2 and carries the C-RNTI or an initial UE identity.”).
Claim 18 is rejected for the same reasons set forth above because the claims have similar limitations or have been addressed.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DAVID BILODEAU whose telephone number is (571)270-3192. The examiner can normally be reached Monday-Thursday 6:00am-4:00pm Eastern Standard Time. 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) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Wesley Kim can be reached at (571) 272-7867. 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.
/David Bilodeau/
Primary Examiner, Art Unit 2648
Prosecution Insights