An over-the-air computation (AirComp) scheme is proposed for federated edge learning (FEEL) without channel state information (CSI) at the edge devices (EDs) or edge server (ES). The proposed scheme adopts the majority vote (MV) principle and uses pulse-position modulation (PPM) symbols constructed with discrete Fourier transform (DFT)-spread orthogonal frequency division multiplexing (OFDM) (DFT-s-OFDM) as votes from EDs. By taking the delay spread and synchronization errors into account, we show how to eliminate the need for truncated-channel inversion (TCI) at the EDs and detect MV at the ED with a non-coherent detector. The proposed method naturally reduces the peak-to-mean envelope power ratio (PMEPR) of the signal as it inherits the properties of the single-carrier (SC) waveform. An alternative proposed scheme also adopts the majority vote (MV) principle but further defines multiple subcarriers and orthogonal frequency division multiplexing (OFDM) symbols for voting options, which reduces to frequency-shift keying (FSK) over OFDM subcarriers as a special case. Since the votes from EDs are separated on orthogonal resources, the proposed scheme eliminates the need for truncated-channel inversion (TCI) at the EDs and allows the ES to detect MV with a non-coherent detector. We also mitigate the peak-to-mean envelope power ratio (PMEPR) of the synthesized signals by using randomization symbols. Through simulations, we show that the proposed schemes provide high test accuracy in fading channels for both independent and identically distributed (IID) and non-IID data while resulting in lower PMEPR symbols as compared to one-bit broadband digital aggregation (OBDA) with quadrature amplitude modulation (QAM).

The present invention concerns a real-time method for the blind demodulation of digital telecommunication signals, based on the observation of a sampled version of this signal. The method comprises the following steps: —acquisition, by a sampling, of a first plurality of signals in order to each constitute an input of a network of L processing blocks (G, F, H), also referred to here as “specialized neurons”, each neuron being simulated by the outputs of the preceding block, the first plurality of signals being input into the first block simulating a first neuron of the network in order to generate a plurality of outputs of the first block; each neuron F being simulated by the outputs of an upstream chain G and stimulating a downstream chain H; each set of samples passes through the same processing chain; —the outputs of the last blocks of the network ideally correspond to the demodulated symbols; —addition of a nonlinearity to each of the outputs of the last block of the network making it possible to calculate an error signal and propagation of this error in the reverse direction of the processing chain (“retropropagation”); —estimation, upon receipt of the error by each neuron (i), of a corrective term δθ.sub.i and updating, in each block, of the value of the parameter θ.sub.i according to θ.sub.i+=δθ.sub.i.

A method by which a terminal demodulates a signal in a wireless access system according to one embodiment of the present invention comprises the steps of: receiving a modulation signal having a modulation order of 2.sup.m (m is a natural number); determining a first demodulation constellation arrangement corresponding to a k.sup.th (k is an even number among natural numbers of m or less) bit among m numbers of bits; determining a second demodulation constellation arrangement corresponding to an nth (n is an odd number among natural numbers of m or less) bit among the m numbers of bits; and demodulating the received modulation signal by using the first and second demodulation constellation arrangements, wherein the first constellation arrangement can have a pattern in which a square matrix having a size of 2.sup.(m/2+1)−(k/2) is repeated.

Improvements to demodulators in receivers of radios used for train communications t to receive a radio frequency (RF) signal transmitting a packet of data. A demodulator is configured to generate a soft decision for a decoder, the soft decision including reliability information determined by calculating for the bit position a logarithmic likelihood ratio (LLR). The demodulator is configured to correct a bias in the LLR calculation for any one of the bit positions resulting from a difference in the number symbols in the set of all possible symbols that could have a 0 value in the bit position and the set of all possible symbols that could have a 1 in the bit position.

Embodiments of this application provide a non-coherent data transmission method and a communication apparatus. In the method, a transmit end device determines, in a first constellation, a first constellation point corresponding to first to-be-modulated bits, where the first constellation point corresponds to P first symbols, P=M*N, M is a positive integer, and N is an integer greater than 1; and sends the P first symbols on N resource units by using M antenna ports, or sends P second symbols determined based on the P first symbols, and foregoes sending demodulation reference signals of the P first symbols or the P second symbols. In the method, a constellation is designed, and each constellation point in the constellation corresponds to a plurality of resource units so that data can be transmitted with no need to transmit a reference signal, and only the data needs to be transmitted.

Methods (C) for converting a data signal (U). The methods may comprise (i) providing an input symbol stream (IB) of input symbols (Bj), the input symbol stream (IB) being representative for the data signal (U) to be converted and (ii) applying to consecutive disjunct partial input symbol sequences (IB.sup.p) of a number of p consecutive input symbols (IBj) covering said input symbol stream (IB), a distribution matching process (DM) to generate and output a final output symbol stream (OB) or a preform thereof, wherein the distribution matching process (DM) may be formed by a preceding shell mapping process (SM) and a succeeding amplitude mapping process (AM), wherein said shell mapping process (SM) may be configured to form and output to said amplitude mapping process (AM) for each of said consecutive partial input symbol sequences (IB.sup.p) a sequence (s.sup.q) of a number of q shell indices (s), and wherein said amplitude mapping process (AM) may be configured to assign to each shell index (s) a tuple of amplitude values.

The present application provides a non-orthogonal demodulation module, receiving a received signal and the received signal is related to a summation of a plurality of transmitted signals. The plurality of transmitted signals are corresponding to a plurality of frequencies, and the plurality of transmitted signals are not orthogonal to each other. The non-orthogonal demodulation module comprises a mixing-and-integrating unit, configured to perform mixing operations and integrating operations on the received signal respectively at the plurality of frequencies, to generate a plurality of in-phase components and a plurality of quadrature components corresponding to the plurality of frequencies; and a decoding unit, configured to generate at least a decoding matrix, and compute a plurality of energies corresponding to the plurality of transmitted signals according to the at least a decoding matrix, the plurality of in-phase components and the plurality of quadrature components.

The present invention concerns a real-time method for the blind demodulation of digital telecommunication signals, based on the observation of a sampled version of this signal. The method comprises the following steps: acquisition, by a sampling, of a first plurality of signals in order to each constitute an input of a network of L processing blocks (G, F, H), also referred to here as specialized neurons, each neuron being simulated by the outputs of the preceding block, the first plurality of signals being input into the first block simulating a first neuron of the network in order to generate a plurality of outputs of the first block; each neuron F being simulated by the outputs of an upstream chain G and stimulating a downstream chain H; each set of samples passes through the same processing chain; the outputs of the last blocks of the network ideally correspond to the demodulated symbols; addition of a nonlinearity to each of the outputs of the last block of the network making it possible to calculate an error signal and propagation of this error in the reverse direction of the processing chain (retropropagation); estimation, upon receipt of the error by each neuron (i), of a corrective term .sub.i and updating, in each block, of the value of the parameter .sub.i according to .sub.i+=.sub.i.

Methods and systems for waking up a wireless receiving device having a wake-up radio (WUR) circuit. A low-power wake-up signal is transmitted, comprising a wake-up frame (WUF) including having a portion that is subjected to spectrum spreading and transmitted using a single carrier. The wake-up signal is processed by the receiving using non-coherent detection.

Improvements to demodulators in receivers of radios used for train communications t to receive a radio frequency (RF) signal transmitting a packet of data. A demodulator is configured to generate a soft decision for a decoder, the soft decision including reliability information determined by calculating for the bit position a logarithmic likelihood ratio (LLR). The demodulator is configured to correct a bias in the LLR calculation for any one of the bit positions resulting from a difference in the number symbols in the set of all possible symbols that could have a 0 value in the bit position and the set of all possible symbols that could have a 1 in the bit position.