Abstract: Building such optimized DNN models model for resource constrained devices require huge amount of workflow setup, engineering skills and research skills. There is a need to automate the process for generation of optimized DNN models for Green analytics. A method and system providing integrated platform for rapid automated generation of Tiny ML models for edge devices by integrating multiple optimization techniques and recommending the mist appropriate technique based on available input parameters is provided. Further one of the optimization technique the Fast-NAS can be generalized across multiple applications and consume 95% less computational power (GPU hours). The enhancement is achieved using a performance evaluation technique for generated models using new metric, a new reward function with adaptive parameters, an early-exit strategy to further expedite the optimization process, and a new NAS flow enhanced with AutoML (Hyper-Parameter Optimization) to minimize human intervention.

Abstract: A method for determining Listen Before Talk (LBT) type and Channel Access Priority Class (CAPC) for signaling message transmission over signaling radio bearers (SRBs) in 5G New Radio-Unlicensed (NR-U) is proposed. A user equipment (UE) generates an SRB message to be transmitted to a serving base station over a control channel. Before the uplink (UL) transmission of an UL PDU containing the SRB message, the UE performs an LBT procedure using a set of LBT parameters associated with a CAPC value. In one embodiment, the CAPC is provided by the base station via an UL grant contained in a downlink control information (DCI). In another embodiment, the CAPC is determined by the UE based on an SRB type of the SRB message.

Abstract: A method of manufacturing a hollow fiber carbon membrane, the method includes heating a polymeric precursor to a pyrolysis temperature that is greater than or equal to 900° C. and less than or equal to 1200° C., and pyrolyzing the polymeric precursor at the pyrolysis temperature in a pyrolysis atmosphere that comprises oxygen in an amount that is greater than 0 ppm and less than 200 ppm.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Various solutions for capability indication of repetition for Message 4 (Msg4) hybrid automatic repeat request (HARQ)-acknowledgement (ACK) or Msg4 HARQ-negative ACK (NACK) are described. An apparatus may receive a system information block (SIB) from a network node of a wireless network, wherein the SIB comprises one or more physical uplink control channel (PUCCH) repetition numbers for Msg4. The apparatus may report a capability indication of PUCCH repetition for Msg4 HARQ-ACK or Msg4 HARQ-NACK to the network node through a Message 1 (Msg1) or a Message 3 (Msg3). The apparatus may transmit with repetitions a PUCCH with HARQ-ACK information or HARQ-NACK information for Msg4 to the network node according to one of the one or more PUCCH repetition numbers for Msg4.

Abstract: The present invention relates to a method and device for distributing idle UE by a carrier in eNB of a multi-carrier based mobile communication system. The method of distributing idle UE in a multi-carrier based mobile communication system according to the present invention includes a process of determining a search rate by a carrier on the basis of information representing load on the carrier, a step of determining a cell reselection priority on the idle UE on the basis of the determined search rate, and a process of transmitting the determined cell reselection priority to the idle UE.

Abstract: Various solutions for global navigation satellite system (GNSS) operations with respect to user equipment and network node in mobile communications are described. An apparatus may transmit a GNSS position fix time duration and a GNSS validity duration to a network node during an initial access. The apparatus may receive a configuration based on the GNSS position fix time duration and the GNSS validity duration from the network node in a connected mode. The apparatus may perform a GNSS measurement based on the configuration. The apparatus may transmit a new GNSS validity duration to the network node in the connected mode in an event that the apparatus completes the GNSS measurement.

Abstract: The cloud-based query workload optimization system disclosed herein the cloud-based query workloads optimization system receives query logs from various query engines to a cloud data service, extracts various query entities from the query logs, parses query entities to generate a set of common workload features, generates intermediate representations of the query workloads, wherein the intermediate representations are agnostic to the language of the plurality of the queries, identifies a plurality of workload patterns based on the intermediate representations of the query workloads, categorizes the workloads in one or more workload type categories based on the workload patterns and the workload features, and selects an optimization scheme based on the category of workload pattern.

Abstract: The present invention relates to a method and device for distributing idle UE by a carrier in eNB of a multi-carrier based mobile communication system. The method of distributing idle UE in a multi-carrier based mobile communication system according to the present invention includes a process of determining a search rate by a carrier on the basis of information representing load on the carrier, a step of determining a cell reselection priority on the idle UE on the basis of the determined search rate, and a process of transmitting the determined cell reselection priority to the idle UE.

Abstract: In an aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The apparatus may be wireless equipment. The wireless equipment selects a first subset of beams to be utilized for beam management. The beams are from a set of first type of beams used for communication with a base station or a UE. The wireless equipment measures signals transmitted on a second subset of beams. The beams are from the set of first type of beams or from a set of second type of beams. The wireless equipment measures the signals over a time window. The wireless equipment inputs the measurements to a computational model. The wireless equipment receives predictions of channel measurements on the first subset of beams from the computational model.

Abstract: This invention presents methods leveraging artificial intelligence and machine learning (AI/ML) models to enhance wireless communications efficiency in 5G/6G networks. The processes involve storing, configuring, and transferring AI/ML models within base stations and user equipment devices (UE), allowing for localized decision-making and improved network performance. Features include dynamic model activation/deactivation, model compression/decompression, and encoding/decoding method negotiation. Periodic or condition-driven model updates ensure responsiveness to network changes, while model replacements enable upgrades and iterations. The system facilitates seamless handovers between base stations, with information sharing about model capabilities and UE specifics. Model storage and configuration can also occur in the UE, empowering it for local decision-making in variable or challenging network conditions.

Abstract: Apparatus and methods are provided for enhanced AM RLC for NB-IoT UE in the NTN. In one novel aspect, the UE is configured with HARQ disabled, and configured with fast polling and/or fast status report for AM RLC enhancement. In one embodiment, the UE is configured with shortened AM window size, and wherein the shortened window size is smaller than an AM window size for an NB-IoT device. In other embodiments, the fast polling is triggered based on PDU threshold or byte threshold, or triggered based on an NTN polling timer, or triggered based on a combination of an NTN polling timer and one or more threshold triggers. In one embodiment, UE is configured to autonomously send a status report when a reception failure of an RLC PDU is detected, and wherein the status report is sent by a semi-persistent scheduling (SPS) uplink resource.

Abstract: Solutions pertaining to spectrum sharing in coordination between a non-terrestrial network (NTN) and a terrestrial network (TN) are proposed. An apparatus implemented in a UE measures a signal strength of each of one or more TN cells and one or more NTN cells. The apparatus reports a result of the measuring to a TN. The apparatus also receives an indication from the TN configuring and activating a bandwidth part (BWP) that separates transmission by a base station of the TN from NTN downlink (DL) transmissions.

Abstract: Apparatuses, methods, and systems for optimizing vehicle route overlap in vehicle routing plans are disclosed. A method comprises selecting a first vehicle route and a second vehicle route of a plurality of vehicle routes using a plurality of overlap metrics, reallocating destination nodes of the first vehicle route and the second vehicle route to a first new route and a second new route using an ordered projection of the destination nodes onto a line passing through the first vehicle route and the second vehicle route, and determining an overlap-optimized vehicle routing plan including the first new route and the second new route.

Abstract: This disclosure relates generally to systems and methods for autonomous task composition of vision pipelines using an algorithm selection framework. The framework leverages transformer architecture along with deep reinforcement learning techniques to search an algorithmic space for unseen solution templates. In an embodiment, the present disclosure describes a two stage process of identifying the vision pipeline for a particular task. At first stage, a high-level sequence of the vision pipeline is provided by a symbolic planner to create the vision workflow. At second stage, suitable algorithms for each high-level task are selected. This is achieved by performing a graph search using a transformer architecture over an algorithmic space on each component of generated workflow. In order to make the system more robust, weights of embedding, key and query networks of a visual transformer are updated with a Deep Reinforcement Learning framework that uses Proximal Policy Optimization (PPO) as underlying algorithm.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Examples pertaining to selection of a tracking area identifier (TAI) as a last visited registered TAI in a geographical location in mobile communications are described. A user equipment (UE) receives broadcast information from a network indicating a plurality of tracking area codes (TACs) for the network. The UE then determines one or more candidate TACs, including at least a first TAC, best representing a current geographical location of the UE from the plurality of TACs received by removing zero, one or more TACs of the plurality of TACs from consideration based on the broadcast information received over time.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Methods for selectively enabling and disabling UL HARQ retransmission in NTN communications and communication apparatus utilizing the same. A method for selectively enabling and disabling uplink Hybrid Automatic Repeat request (HARQ) retransmission in Non-Terrestrial Network (NTN) communications includes: determining a set of logical channels that are allowed to transmit or a set of MAC CEs that are allowed to be transmitted using one or more specific uplink (UL) HARQ processes. The one or more specific UL HARQ processes include one or more UL HARQ processes for which the UL retransmission has been disabled, or one or more UL HARQ processes for which the UL retransmission has been enabled.

Abstract: Low Earth Orbit (LEO) satellites orbit around the earth at a high speed (mobility), but over a predictable or deterministic orbit. In this invention, an efficient mechanism to include a list of multiple Tracking Area Codes (TAC List) in LEO-NTN is proposed for efficient tracking area update (TAU). Each cell broadcasts a TAC list corresponding to all the tracking areas it covers as it moves, dynamically. A UE selects a cell and is registered with a TA during initial registration and cell selection. As the satellite moves, the UE reselects another cell during cell reselection. The UE triggers a TAU only if its current TA does not match with any tracking area code of the TAC List broadcasted by the reselected cell. By using the concept of TAC list, it significantly reduces the TAU signaling overhead in LEO-NTN.