Abstract: It is provided a federated learning system for aggregating gradient information representing a result of training an AI model in an edge device, the federated learning system comprising the edge device and a server apparatus, the training module in the edge device being configured to generate an edge switch share in which the encrypted aggregated gradient is encrypted, and to transmit the generated edge switch share to the server apparatus, the encryption/decryption module in the server apparatus being configured to generate an encrypted aggregated gradient for decryption by adding edge switch shares received from the plurality of the edge device, generate an aggregated gradient by decrypting the generated encrypted aggregated gradient for decryption, and to transmit the generated aggregated gradient to the edge device, the training module in the edge device being configured to train the AI model by using the aggregated gradient received from the server apparatus.

Abstract: In various implementations, a method includes determining a sequence of source packets. In some implementations, the sequence of source packets satisfies a windowing condition. In various implementations, the method includes synthesizing a first set of one or more parity packets as a function of a first set of source packets in the sequence. In some implementations, the first set of source packets satisfies a first encoding pattern. In various implementations, the method includes synthesizing a second set of parity packets as a function of a second set of source packets in the sequence. In some implementations, the second set of source packets satisfies a second encoding pattern that is different from the first encoding pattern. In some implementations, the first and second encoding patterns characterize an encoding structure determined as a function of a channel characterization vector.

Abstract: In one embodiment, a device in a network assigns packets from a communication transmitted via the network to time windows over a period of time. The device determines a transmission performance metric for each of the packets in a particular time window and calculates, for each of the time windows, local disturbance scores, which are based on the transmission performance metrics for the packet in the time windows. A particular local disturbance score for a particular time window maps the transmission performance metrics for the packets in the time window to a perceived quality metric. The device determines a distortion score for the communication by aggregating the local disturbance scores for the time windows over the period of time.

Abstract: In various implementations, a method includes determining a sequence of source packets. In some implementations, the sequence of source packets satisfies a windowing condition. In various implementations, the method includes synthesizing a first set of one or more parity packets as a function of a first set of source packets in the sequence. In some implementations, the first set of source packets satisfies a first encoding pattern. In various implementations, the method includes synthesizing a second set of parity packets as a function of a second set of source packets in the sequence. In some implementations, the second set of source packets satisfies a second encoding pattern that is different from the first encoding pattern. In some implementations, the first and second encoding patterns characterize an encoding structure determined as a function of a channel characterization vector.

Abstract: In one embodiment, a device in a network identifies delay requirements of each of a plurality of media streams. The device selects a joint forward error correction (FEC) encoding strategy for the plurality of media streams based on the identified delay requirements of the streams and on a burst loss length of a communication channel. The device applies the selected joint FEC encoding strategy to the plurality of media streams, to form a multiplexed packet stream. The device sends the multiplexed packet stream to one or more nodes in the network via the communication channel.

Abstract: An error correction code includes a separate error code portion for each of two or more separate burst erasure durations (or burst error durations). For each burst erasure duration, the code can be employed to recover from the burst erasure with a different delay time. Each error code portion has a particular parameter for burst duration (B) and delay (T), meaning that the code can be used to recover from a burst erasure of duration B with delay T. Each error code portion is based on separating the source symbols into sub-symbols and diagonally interleaving the sub-symbols based on the (B,T) parameters for the error code portion. Accordingly, different burst erasures are recovered from with different delays.

Abstract: An error correction code includes a separate error code portion for each of two or more separate burst erasure durations (or burst error durations). For each burst erasure duration, the code can be employed to recover from the burst erasure with a different delay time. Each error code portion has a particular parameter for burst duration (B) and delay (T), meaning that the code can be used to recover from a burst erasure of duration B with delay T. Each error code portion is based on separating the source symbols into sub-symbols and diagonally interleaving the sub-symbols based on the (B,T) parameters for the error code portion. Accordingly, different burst erasures are recovered from with different delays.

Abstract: Biometric parameters acquired from human forces, voices, fingerprints, and irises are used for user authentication and access control. Because the biometric parameters are continuous and vary from one reading to the next, syndrome codes are applied to determine biometric syndrome vectors. The biometric syndrome vectors can be stored securely while tolerating an inherent variability of biometric data. The stored biometric syndrome vector is decoded during user authentication using biometric parameters acquired at that time. The syndrome codes can also be used to encrypt and decrypt data.

Abstract: Biometric parameters acquired from human forces, voices, fingerprints, and irises are used for user authentication and access control. Because the biometric parameters are continuous and vary from one reading to the next, syndrome codes are applied to determine biometric syndrome vectors. The biometric syndrome vectors can be stored securely while tolerating an inherent variability of biometric data. The stored biometric syndrome vector is decoded during user authentication using biometric parameters acquired at that time. The syndrome codes can also be used to encrypt and decrypt data.

Abstract: Appropriate determinations in a series of tests, which transition from more passive tests to more active tests, control the admission of data streams onto a network data path. More passive tests can include promiscuous mode and packet-pair tests. When the results of more passive tests indicate a reduce likelihood of a data stream causing a network data path to transition into a congested state, the network data path can be actively probed to make a more precise determination on the likelihood of congestion. A train of diagnostic data packets is transmitted at a diagnostic data transfer rate having a significantly reduced chance of causing congestion. A train of target data packets is transmitted at a requested application transfer data rate. The number of target data packets received within a specified delay threshold is compared to the number of diagnostic data packets receive with the delay threshold.

Abstract: A method compresses a set of correlated signals by first converting each signal to a sequence of integers, which are further organized as a set of bit-planes. This can be done by signal transformation and quantization. An inverse accumulator is applied to each bit-plane to produce a bit-plane of shifted bits, which are permuted according to a predetermined permutation to produce bit-planes of permuted bits. Each bit-plane of permuted bits is partitioned into a set of blocks of bits. Syndrome bits are generated for each block of bits according to a rate-adaptive base code. Subsequently, the syndrome bits can be decompressed in a decoder to recover the original correlated signals. For each bit-plane of the corresponding signal, a bit probability estimate is generated. Then, the bit-plane is reconstructed using the syndrome bits and the bit probability estimate.

Abstract: Appropriate determinations in a series of tests, which transition from more passive tests to more active tests, control the admission of data streams onto a network data path. More passive tests can include promiscuous mode and packet-pair tests. When the results of more passive tests indicate a reduce likelihood of a data stream causing a network data path to transition into a congested state, the network data path can be actively probed to make a more precise determination on the likelihood of congestion. A train of diagnostic data packets is transmitted at a diagnostic data transfer rate having a significantly reduced chance of causing congestion. A train of target data packets is transmitted at a requested application transfer data rate. The number of target data packets received within a specified delay threshold is compared to the number of diagnostic data packets receive with the delay threshold.