Abstract: Input data for inference processes are often highly correlated and interdependent. These correlations provide an inherent redundancy in the input data which can be used to refine inference results and thus improve inference performance. In some embodiments, probability information is used to refine results from an inference process. The probability information indicates, for each of a plurality of potential results obtainable from the inference process, a probability of obtaining the respective potential result and another potential result from the plurality of potential results.

Abstract: Methods and apparatus for scheduling uplink transmissions based on contributiveness to a downstream task are provided. Multiple devices to schedule for uplink transmission are selected from a set of candidate devices based on a contributiveness metric for each device. The contributiveness metric for each device is related to a downstream task in the wireless communication network and is indicative of how well the device is able to successfully transmit information to the network for the downstream task and how informative the information provided by the device is for the downstream task. The contributiveness metric of a candidate device may be learned via machine learning using a deep neural network.

Abstract: Aspects of the present disclosure relate to inference and, in particular, to distributed inference representative of a machine learning process. It is expected that inferencing will be a service in wireless networks. Aspects of the present application relate to applying aspects of coding theory to distributed inference to introduce redundancy. Methods of decoding outputs from a distributed inference process are also provided.

Abstract: Aspects of the present disclosure relate to a reference signal assignment, in which reference signals for a first apparatus may be assigned based on second channel estimates for one or more second apparatus having a same location and network resource configuration as the first apparatus.

Abstract: Some embodiments of the present disclosure provide certification for the handling of an inference request that is transmitted to a DNN hosted on a remote computing system. Output data, received responsive to the inference request, may be certified as being appropriately generated by the DNN, rather than being tampered with or generated by a malicious DNN. Output data from the DNN may be also certified as appropriately corresponding to input data included in the inference request. Linear block coding may be used on transmissions to guard against eavesdropping and tampering. Through the use of a certification DNN, a degree of comfort may be gained that given output data appropriately corresponds to input data included in a given inference request.

Abstract: Some embodiments of the present application provide a forward-propagation-only (FP-only) method of training a DNN model. Such methods result in a trained DNN model whose performance comparable to a DNN model trained using bidirectional training methods. The FP-only method for training a DNN model may operate without employing the known chain rule. The chain rule is employed when computing gradients for a backward propagation in a bidirectional method. The FP-only method may allow for the computations and updates to the parameters for each layer of the DNN model to be performed in parallel. The FP-only methods for training a DNN model use stochastic gradient descent and the FP-only method for training a DNN model still involves computing gradients. However, the FP-only methods of training a DNN model allow for computing of gradients without the chain rule.

Abstract: Systems and methods of reporting wireless channel state information are provided. With the provided system and method, in a situation where there are multiple UEs which are close to each other, such that channel conditions may be similar for the multiple UEs, one of the UEs is configured to report interference information on a time pattern that has at least two measurement time durations for which interference is to be measured, for example, only for a subset of N consecutive measurement time durations. Other UEs may be configured to report interference information for different time patterns, for example different subsets of the N time slots.

Abstract: This application provides example communication methods and a related apparatuses. An example communication method may be used for a first processing entity of a first communication device. The example method includes the first processing entity sending a first registration request to a second entity, where the first registration request includes a first identifier which indicates one or more of the following information of the first processing entity: model information or data information. A model parameter of the second entity can be adjusted based on a real-time environment parameter. The first processing entity receives a first registration response sent by the second entity, where the first registration response indicates an input parameter of the second entity. The first processing entity sends a first response acknowledgment, where the first response acknowledgment indicates whether an output parameter of the first processing entity matches the input parameter of the second entity.

Abstract: Current online training procedures for artificial intelligence/machine learning (AI/ML) models for wireless communication generally suffer from high communication overhead and/or significant delays in training, particularly when training data is exchanged over an unreliable/hostile communication channel. An example method includes communicating, in accordance with a first communication mode, data or control information with a second device in the wireless communication network. The first communication mode is one of a plurality of communication modes comprising at least the first communication mode and a second communication mode, the second communication mode differing from the first communication mode in terms of at least one of a quantization level used to quantize values of variables in the data or control information, or a Hybrid Automatic Repeat reQuest (HARQ) feedback and retransmission mode for selective retransmission of one or more portions of the data or control information.

Abstract: A wireless network can generate candidate signal configurations for physical transmissions to or from a user equipment (UE) in a radio environment. The generation of candidate signal configurations can be performed using a first neural network that is associated with the UE. These signal configurations can then be evaluated using a second neural network that is associated with the radio environment. The second neural network can be trained using measurements from previous physical transmissions in the radio environment. The trained second neural network generates a reward value that is associated with the candidate signal configurations. The first neural network is then trained using the reward values from the second neural network to produce improved candidate signal configurations. When a signal configuration that produces a suitable reward value is generated, this signal configuration can be used for the physical transmission in the radio environment.

Abstract: Embodiments of this application disclose a communication method and a communication apparatus. The method includes: A transmit end performs first encoding processing on first data by using an encoder neural network, to obtain a first sent feature, where the first sent feature is related to a channel distribution dimension of an environment in which the transmit end is located. The transmit end performs second encoding processing on the first sent feature by using a matching layer, to obtain a first feature, where the encoder neural network and the matching layer are obtained through independent training. The transmit end sends the first feature to a receive end, where the first feature is used by the receive end to obtain the first data.

Abstract: Some embodiments of the present disclosure relate to inferencing using a trained deep neural network. Inferencing may, reasonably, be expected to be a mainstream application of 6G wireless networks. Agile, robust and accurate inferencing is important for the success of AI applications. Aspects of the present application relate to introducing coding theory into inferencing in a distributed manner. It may be shown that redundant wireless bandwidths and edge units help to ensure agility, robustness and accuracy in coded inferencing networks.

Abstract: A wireless network can generate candidate signal configurations for physical transmissions to or from a user equipment (UE) in a radio environment. The generation of candidate signal configurations can be performed using a first neural network that is associated with the UE. These signal configurations can then be evaluated using a second neural network that is associated with the radio environment. The second neural network can be trained using measurements from previous physical transmissions in the radio environment. The trained second neural network generates a reward value that is associated with the candidate signal configurations. The first neural network is then trained using the reward values from the second neural network to produce improved candidate signal configurations. When a signal configuration that produces a suitable reward value is generated, this signal configuration can be used for the physical transmission in the radio environment.

Abstract: A resource mapping method and apparatus are described that reduce resource mapping complexity. The method includes an access network device obtaining resource mapping information of at least one terminal, and sending the resource mapping information to the at least one terminal respectively. The resource mapping information indicates a resource element RE to which each of the at least one terminal performs mapping in each RMB in a first resource mapping block RMB set. An RE to which a first terminal performs mapping in each RMB in the first RMB set is determined based on a first interleaved sequence, and the first interleaved sequence is determined based on a base sequence. The first terminal is any one of the at least one terminal, and the first RMB set includes at least one RMB.

Abstract: Embodiment techniques map parity bits to sub-channels based on their row weights. In one example, an embodiment technique includes polar encoding, with an encoder of the device, information bits and at least one parity bit using the polar code to obtain encoded data, and transmitting the encoded data to another device. The polar code comprises a plurality of sub-channels. The at least one parity bit being placed in at least one of the plurality of sub-channels. The at least one sub-channel is selected from the plurality of sub-channels based on a weight parameter.

Abstract: According to the present disclosure, there are provided methods and devices for estimating a high-dimensional MIMO channel by expressing the MIMO channel in the form of a flow function to determine a change from an input state to an output state. This may be considered similar to a manner in which fluid dynamic problems are solved. Instead of using linearization to estimate all of the coefficients of a matrix representing a MIMO channel, a receiver may determine the MIMO channel rank by detecting leading modes in a subspace of the entire MIMO channel matrix.

Abstract: A broadcast signaling method performed by a network device having a protocol stack of with first and second protocol layers where the second protocol layer is below the first protocol layer, the method including generating, by the network device, first information at the first protocol layer, generating, by the network device, second information at the second protocol layer, where the second information is used to determine a time-frequency resource corresponding to one or more synchronization signal blocks (SSBs), processing, by the network device, the first information and the second information at the second protocol layer, and sending, by the network device to a terminal device by using a physical broadcast channel (PBCH) in the one or more SSBs, data obtained after second protocol layer processing.

Abstract: Methods and apparatuses for feature-driven communications are described. A set of features describing an observed subject is transmitted by a transmitting electronic device (ED) to a base station (BS). The BS translates the received features to another set of transmission features to be transmitted to a receiving ED. The receiving ED recovers information about the subject from the features received from the BS.

Abstract: A resource mapping method and apparatus and a resource mapping indication method and apparatus are provided to adapt to an interference cancellation function of a receiver. According to the method, a network device obtains a mapping mode used for resource mapping during uplink transmission, and sends, to a terminal, information indicating the mapping mode. The mapping mode is used to indicate mapping locations of a plurality of modulation symbols in a resource mapping block (RMB), the RMB comprises a plurality of resource elements (REs), and at least one of the plurality of REs carries at least two modulation symbols.

Abstract: A wireless network can generate candidate signal configurations for physical transmissions to or from a user equipment (UE) in a radio environment. The generation of candidate signal configurations can be performed using a first neural network that is associated with the UE. These signal configurations can then be evaluated using a second neural network that is associated with the radio environment. The second neural network can be trained using measurements from previous physical transmissions in the radio environment. The trained second neural network generates a reward value that is associated with the candidate signal configurations. The first neural network is then trained using the reward values from the second neural network to produce improved candidate signal configurations. When a signal configuration that produces a suitable reward value is generated, this signal configuration can be used for the physical transmission in the radio environment.