Terminal and communication method
The terminal's neural network-based modulation method addresses performance degradation in wireless communication systems by mapping bit sequences to symbols considering power amplifier saturation points, resulting in improved SNR and throughput gains.
Patent Information
- Application Number
- PCT/JP2023/017346
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-08
- Publication Date
- 2024-11-14
AI Technical Summary
Conventional modulation methods in wireless communication systems, such as QAM and APSK, experience performance degradation due to phase noise and nonlinear distortions caused by high power amplifiers, which are not effectively addressed by existing technologies.
A terminal equipped with a control unit that learns a neural network to map bit sequences to modulation symbols, considering power amplifier saturation points, and transmits these symbols to a base station, thereby reducing performance degradation caused by phase noise and nonlinear distortions.
The solution achieves an SNR gain of 2 dB and a throughput gain of 10%-20% relative to existing constellations, improving resistance to phase noise and nonlinear PA effects, leading to enhanced transmission performance.
Smart Images

Figure JP2023017346_14112024_PF_FP_ABST
Abstract
Description
Terminal and communication method
[0001] The present invention relates to a terminal and a communication method in a wireless communication system.
[0002] For NR (New Radio) (also known as "5G"), the successor system to LTE (Long Term Evolution), technologies are being considered that satisfy the requirements of a large-capacity system, high-speed data transmission speed, low latency, simultaneous connection of a large number of terminals, low cost, and low power consumption (for example, Non-Patent Document 1).
[0003] In conventional transmissions using higher frequencies and wider bandwidths, the phase noise (PN) generated by the RF oscillator increases with increasing carrier frequency, and while a high power amplifier (HPA) is used to obtain a large signal-to-noise ratio (SNR), nonlinear distortion may occur.
[0004] 3GPP TS 38.300 V17.3.0(2022-12)C. Rapp, "Effects of HPA-Nonlinearity on a 4-DPSK / OFDM-Signal for a Digital Sound Broadcasting System", in Proceedings of the Second European Conference on Satellite Communications, Liege, Belgium, Oct. 22-24, 1991, pp. 179-184.A. Ugolini, A. Piemontese and T. Eriksson, "Spiral Constellations for Phase Noise Channels," in IEEE Transactions on Communications, vol. 67, no. 11, pp. 7799-7810, Nov. 2019, doi: 10.1109 / TCOMM.2019.2937293.Digital Video Broadcasting (DVB); Second generation framing structure, channel coding and modulation systems for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications; Part 1: DVB-S2.( ETSI EN 302 307-1 V1.4.1 (2014-11))Y. Ji, G. Zhang, J. Huang, J. Yang, G. Gui and H. Sari, "Deep Learning for Adaptive Modulation and Coding with Payload Length in Vehicle-to-Vehicle Communications Systems," 2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall), 2021, pp. 1-5, doi: 10.1109 / VTC2021-Fall52928.2021.9625436.L. -S. Chen, W. -H. Chung, I. -Y. Chen and S. -Y. Kuo, "AMC With a BP-ANN Scheme for 5G Enhanced Mobile Broadband," in IEEE Access, doi: 10.1109 / ACCESS.2020.3024726.J. P. Twarayisenze et al., "Artificial Neural Network Based Adaptive Spatial Modulation," 2021 IEEE 6th International Conference on Computer and Communication Systems (ICCCS), 2021, pp. 553-557, doi: 10.1109 / ICCCS52626.2021.9449310.N. Kurniawati and Y. K. Ningsih, "Reinforcement Learning-Based Adaptive Modulation for Vehicular Communication," 2021 18th International Conference on Electrical Engineering / Electronics, Computer, Telecommunications and Information Technology (ECTI-CON), 2021, pp. 224-227, doi: 10.1109 / ECTI-CON51831.2021.9454740.W. Su, J. Lin, K. Chen, L. Xiao and C. En, "Reinforcement Learning-Based Adaptive Modulation and Coding for Efficient Underwater Communications," in IEEE Access, vol. 7, pp. 67539-67550, 2019, doi: 10.1109 / ACCESS.2019.2918506.J. P. Leite, P. H. P.de Carvalho and R. D. Vieira, "A flexible framework based on reinforcement learning for adaptive modulation and coding in OFDM wireless systems," 2012 IEEE Wireless Communications and Networking Conference (WCNC), 2012, pp. 809-814, doi: 10.1109 / WCNC.2012.6214482.D. Lee et al., "DQN-Based Adaptive Modulation Scheme Over Wireless Communication Channels," in IEEE Communications Letters, vol. 24, no. 6, pp. 1289-1293, June 2020, doi: 10.1109 / LCOMM.2020.2978390.L. Zhang, J. Tan, Y. Liang, G. Feng and D. Niyato, "Deep Reinforcement Learning-Based Modulation and Coding Scheme Selection in Cognitive Heterogeneous Networks," in IEEE Transactions on Wireless Communications, vol. 18, no. 6, pp. 3281-3294, June 2019, doi: 10.1109 / TWC.2019.2912754.S. Bicais, J. -B. Dore and J. Luis Gonzalez Jimenez, "Adaptive PSK Modulation Scheme in the Presence of Phase Noise," 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), Kalamata, Greece, 2018, pp.1-5, doi: 10.1109 / SPAWC.2018.8445933.3GPP TS 38.211 V17.4.0(2022-12)"RF impairment models for 60GHz-band SYS / PHY simulation", doc.: IEEE 802.15-06-0477-01-003c3GPP TS 38.214 V17.4.0(2022-12).
[0005] In conventional modulation methods, such as quadrature amplitude modulation (QAM) and amplitude phase shift keying (APSK), performance degradation due to phase noise is significant. Furthermore, conventional adaptive modulation and coding (AMC) does not take into account nonlinear factors such as the saturation point of a power amplifier (PA), which significantly impacts performance.
[0006] The present invention has been made in view of the above-mentioned points, and has as its object to reduce performance degradation caused by a PA (Power Amplifier) saturation point in a wireless communication system.
[0007] According to the disclosed technology, there is provided a terminal having a control unit that pre-trains a neural network (NN) that maps a bit sequence to an index, outputs a one-hot encoder vector corresponding to the index, and outputs modulation symbols using the one-hot encoder vector and a PA (Power Amplifier) saturation point as input, and a transmission unit that transmits to a base station using modulation symbols determined based on a constellation generated by the trained NN.
[0008] According to the disclosed technology, it is possible to reduce performance degradation caused by a PA (Power Amplifier) saturation point in a wireless communication system.
[0009] 1 is a diagram illustrating an example of the configuration of a wireless communication system. FIG. 2 is a diagram illustrating performance differences due to phase noise variance. FIG. 3 is a diagram illustrating performance differences due to phase noise variance and modulation schemes. FIG. 4 is a diagram illustrating an example (1) of a spiral constellation. FIG. 5 is a diagram illustrating an example (2) of a spiral constellation. FIG. 6 is a diagram illustrating performance differences due to PA characteristics of 64QAM. FIG. 7 is a diagram illustrating changes in 64QAM constellations due to PA characteristics. FIG. 8 is a diagram illustrating changes in 64APSK constellations due to PA characteristics. FIG. 9 is a diagram for explaining an example (1) of a system according to an embodiment of the present invention. FIG. 10 is a diagram illustrating an example of a constellation generated by a system according to an embodiment of the present invention. FIG. 11 is a diagram illustrating the performance of example (1) of a constellation generated by a system according to an embodiment of the present invention. FIG. 12 is a diagram illustrating the performance of example (2) of a constellation generated by a system according to an embodiment of the present invention. FIG. 13 is a diagram illustrating the performance of example (3) of a constellation generated by a system according to an embodiment of the present invention. FIG. 14 is a diagram illustrating the performance of example (4) of a constellation generated by a system according to an embodiment of the present invention. FIG. 15 is a diagram for explaining an example (2) of a system according to an embodiment of the present invention. FIG. 1 is a diagram showing the performance of an example constellation (5) generated by a system according to an embodiment of the present invention. FIG. 2 is a diagram showing the performance of an example constellation (6) generated by a system according to an embodiment of the present invention. FIG. 3 is a diagram showing the performance of an example constellation (7) generated by a system according to an embodiment of the present invention. FIG. 4 is a diagram showing the performance of an example constellation (8) generated by a system according to an embodiment of the present invention. FIG. 5 is a diagram for explaining an example of a system for reducing the effect of phase noise according to an embodiment of the present invention. FIG. 6 is a diagram showing an example constellation (1) generated by a system for reducing the effect of phase noise according to an embodiment of the present invention. FIG. 7 is a diagram showing an example constellation (2) generated by a system for reducing the effect of phase noise according to an embodiment of the present invention. FIG. 8 is a diagram showing an example constellation (3) generated by a system for reducing the effect of phase noise according to an embodiment of the present invention.FIG. 1 is a diagram showing an example (4) of a constellation generated by a system that reduces the influence of phase noise according to an embodiment of the present invention. FIG. 2 is a diagram showing an example (5) of a constellation generated by a system that reduces the influence of phase noise according to an embodiment of the present invention. FIG. 3 is a diagram showing an example of PA characteristics. FIG. 4 is a diagram for explaining an example of a system that reduces the influence of a nonlinear PA according to an embodiment of the present invention. FIG. 5 is a diagram showing an example (1) of a constellation generated by a system that reduces the influence of a nonlinear PA according to an embodiment of the present invention. FIG. 6 is a diagram showing an example (2) of a constellation generated by a system that reduces the influence of a nonlinear PA according to an embodiment of the present invention. FIG. 7 is a diagram showing an example (3) of a constellation generated by a system that reduces the influence of a nonlinear PA according to an embodiment of the present invention. FIG. 8 is a diagram showing an example (4) of a constellation generated by a system that reduces the influence of a nonlinear PA according to an embodiment of the present invention. FIG. 9 is a diagram showing an example (9) of a constellation generated by a system according to an embodiment of the present invention. FIG. 10 is a diagram for explaining an example of a system that reduces the influence of phase noise and a nonlinear PA according to an embodiment of the present invention. FIG. 1 is a diagram showing an example (1) of a constellation generated by a system for reducing the influence of phase noise and nonlinear PA according to an embodiment of the present invention. FIG. 2 is a diagram showing an example (2) of a constellation generated by a system for reducing the influence of phase noise and nonlinear PA according to an embodiment of the present invention. FIG. 3 is a diagram showing an example (3) of a constellation generated by a system for reducing the influence of phase noise and nonlinear PA according to an embodiment of the present invention. FIG. 4 is a diagram showing an example of the functional configuration of a base station 10 according to an embodiment of the present invention. FIG. 5 is a diagram showing an example of the functional configuration of a terminal 20 according to an embodiment of the present invention. FIG. 6 is a diagram showing an example of the hardware configuration of a base station 10 or a terminal 20 according to an embodiment of the present invention. FIG. 7 is a diagram showing an example of the configuration of a vehicle 2001 according to an embodiment of the present invention.
[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Note that the embodiment described below is an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.
[0011] In the operation of the wireless communication system according to the embodiment of the present invention, existing technology is used as appropriate. However, the existing technology is, for example, the existing LTE, but is not limited to the existing LTE. Furthermore, the term "LTE" used in this specification has a broad meaning including LTE-Advanced and systems subsequent to LTE-Advanced (e.g., NR), unless otherwise specified.
[0012] In addition, in the embodiments of the present invention described below, terms such as SS (Synchronization signal), PSS (Primary SS), SSS (Secondary SS), PBCH (Physical broadcast channel), PRACH (Physical random access channel), PDCCH (Physical Downlink Control Channel), PDSCH (Physical Downlink Shared Channel), PUCCH (Physical Uplink Control Channel), and PUSCH (Physical Uplink Shared Channel) used in existing LTE are used. This is for convenience of description, and similar signals, functions, etc. may be called by other names. In addition, the above-mentioned terms in NR correspond to NR-SS, NR-PSS, NR-SSS, NR-PBCH, NR-PRACH, etc. However, even if a signal is used in NR, it is not necessarily stated as "NR-".
[0013] Furthermore, in the embodiment of the present invention, the duplex method may be a TDD (Time Division Duplex) method, an FDD (Frequency Division Duplex) method, or other methods (for example, Flexible Duplex, etc.).
[0014] Furthermore, in the embodiments of the present invention, "configuring" radio parameters etc. may mean that predetermined values are pre-configured, or that radio parameters notified from the base station 10 or the terminal 20 are set.
[0015] Fig. 1 is a diagram showing a configuration example (1) of a wireless communication system according to an embodiment of the present invention. As shown in Fig. 1, the wireless communication system according to the embodiment of the present invention includes a base station 10 and a terminal 20. Although Fig. 1 shows one base station 10 and one terminal 20, this is an example, and there may be a plurality of each.
[0016] The base station 10 is a communication device that provides one or more cells and performs wireless communication with the terminal 20. The physical resources of a wireless signal are defined in the time domain and the frequency domain. The time domain may be defined by the number of Orthogonal Frequency Division Multiplexing (OFDM) symbols, and the frequency domain may be defined by the number of subcarriers or the number of resource blocks. The base station 10 transmits a synchronization signal and system information to the terminal 20. The synchronization signal is, for example, NR-PSS and NR-SSS. The system information is, for example, transmitted via NR-PBCH and is also referred to as broadcast information. The synchronization signal and system information may also be referred to as SSB (SS / PBCH block). As shown in FIG. 1 , the base station 10 transmits control signals or data to the terminal 20 via DL (Downlink) and receives control signals or data from the terminal 20 via UL (Uplink). Both the base station 10 and the terminal 20 are capable of transmitting and receiving signals by performing beamforming. Furthermore, both the base station 10 and the terminal 20 can apply MIMO (Multiple Input Multiple Output) communication to DL or UL. Furthermore, both the base station 10 and the terminal 20 may communicate via a secondary cell (SCell: Secondary Cell) and a primary cell (PCell: Primary Cell) using CA (Carrier Aggregation). Furthermore, the terminal 20 may communicate via a primary cell of the base station 10 and a primary secondary cell group cell (PSCell: Primary SCG Cell) of another base station 10 using DC (Dual Connectivity).
[0017] The terminal 20 is a communication device equipped with a wireless communication function, such as a smartphone, a mobile phone, a tablet, a wearable terminal, or an M2M (Machine-to-Machine) communication module. As shown in Fig. 1, the terminal 20 receives control signals or data from the base station 10 via DL and transmits control signals or data to the base station 10 via UL, thereby utilizing various communication services provided by the wireless communication system. The terminal 20 also receives various reference signals transmitted from the base station 10 and measures propagation path quality based on the reception results of the reference signals.
[0018] The terminal 20 is capable of performing carrier aggregation, which aggregates multiple cells (multiple CCs (Component Carriers)) to communicate with the base station 10. In carrier aggregation, one PCell (Primary cell) and one or more SCells (Secondary cells) are used. Also, a PUCCH-SCell having a PUCCH may be used.
[0019] In conventional transmissions using higher frequencies and wider bandwidths, the phase noise (PN) generated by the RF oscillator increases with increasing carrier frequency. Also, while a high power amplifier (HPA) is used to obtain a high signal-to-noise ratio (SNR), nonlinear distortion may occur. The impact of this on existing modulation schemes is explained below.
[0020] The spiral constellation is a constellation that is resistant to phase noise. However, it has a variable parameter f based on an analytical formula under the condition of fixed SNR and PN variance. s Performance depends on the degree of optimization.
[0021] APSK (Amplitude Phase Shift Keying) constellations have low PAPR (Peak-to-Average Power Ratio) characteristics and limited tolerance to nonlinear PA distortion. However, the APSK parameters need to be further optimized for various PA nonlinearities in sub-THz systems.
[0022] Adaptive modulation and coding (AMC) using quadrature amplitude modulation (QAM) or phase shift keying (PSK) considers signal-to-interference and noise ratio (SINR) as the most important factor when selecting modulation and coding schemes (MCS) with different modulation orders. However, the limited constellations of QAM or PSK do not allow for continuous optimization depending on channel conditions and / or impairments.
[0023] Therefore, a new constellation design based on APSK that is resistant to phase noise and AMC that takes PN dispersion into consideration may be introduced. Also, a new constellation design based on APSK that is resistant to the nonlinearity of PA and AMC that takes the saturated power of PA into consideration may be introduced.
[0024] Furthermore, to achieve more precise AMC, a new constellation may be automatically generated based on a neural network (NN) that inputs various PN variances and / or PA saturation powers.
[0025] For example, the new constellations achieve a 2 dB SNR gain and a 10%-20% throughput gain over spiral constellations. Also, for example, the new constellations achieve a 2 dB-3 dB SNR gain and a 10%-20% throughput gain over the existing DVB (Digital Video Broadcasting) standard APSK constellations. More details are provided below.
[0026] Conventional modulation schemes, such as QAM, have constellations that increase the minimum Euclidean distance. However, under the influence of PN and nonlinear PA (see Non-Patent Document 2), the minimum Euclidean distance is no longer the most important factor.
[0027] For example, in a spiral constellation (see Non-Patent Document 3), the parameter f s is optimized for one PN distribution, so in an environment where multiple PN distributions are expected, the tolerance decreases.
[0028] For example, APSK (see Non-Patent Document 4) has a small number of amplitudes to reduce PAPR. It has limited tolerance to PA nonlinearity so that performance degradation is acceptable. However, parameters such as the number of rings, the number of points in each ring, and the initial phase of each ring in existing APSK do not take into account nonlinear PA and PN.
[0029] Figure 2 shows the performance difference due to the variance of phase noise. Figure 2 shows the SER performance versus the PN variance of 64QAM. As shown in Figure 2, the SER (Symbol Error Rate) of 64QAM increases as the PN variance increases. When the PN variance is 0.01, the SER almost reaches the error floor.
[0030] FIG. 3 is a diagram showing the performance difference depending on the variance of phase noise and the modulation method. s 3 shows the SER performance for a PN dispersion of 0.015 (0.00413). As shown in Fig. 3, the spiral constellation has a higher SER performance than the QAM constellation at the same PN dispersion. On the other hand, for example, a spiral constellation configured for a PN dispersion of 0.015 exhibits a degraded SER performance at other PN dispersion values.
[0031] Figure 4 shows an example of a spiral constellation (1). Figure 4 shows a 256 spiral, f s0. In the following description, the horizontal axis of a graph showing a constellation corresponds to the in-phase component, and the vertical axis corresponds to the quadrature-phase component.
[0032] Figure 5 shows an example of a spiral constellation (2). Figure 5 shows a 256 spiral, f s 0.00413.
[0033] Figure 6 shows the performance difference due to PA characteristics for 64QAM. Figure 6 shows the SER performance of 64QAM using a nonlinear PA with different saturation point (PaSat) values. As shown in Figure 6, distortion of QAM after passing through a nonlinear PA increases the SER.
[0034] 7 is a diagram showing changes in the 64QAM constellation due to PA characteristics. As shown in FIG. 7, the 64QAM signal after passing through a nonlinear PA with PaSat of 0.58 has a distorted constellation, which causes an increase in SER.
[0035] Figure 8 shows the performance difference due to PA characteristics of 64APSK. It shows the SER performance of 64APSK using a nonlinear PA with different PaSat values. As shown in Figure 8, the distortion of APSK after passing through a nonlinear PA increases the SER.
[0036] 9 is a diagram showing a 64APSK constellation, which is the 64APSK constellation of DVB-S2 (see Non-Patent Document 4).
[0037] Fig. 10 is a diagram showing changes in the 64APSK constellation due to PA characteristics. As shown in Fig. 10, the constellation of 64APSK after passing through a nonlinear PA with PaSat of 0.58 is distorted compared to Fig. 9, which causes an increase in SER.
[0038] Existing AMC adjusts MCS based on SNR and CQI (Channel quality information) and uses binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), 16QAM, 64QAM, etc. Table 1 is an example of an existing AMC table (see Non-Patent Document 12).
[0039]
[0040] As shown in Table 1, the MCS index specifies the modulation order, target code rate, and spectral efficiency (SE).
[0041] In AI (Artificial Intelligence)-based AMC (see Non-Patent Documents 5 to 11), MCS is adjusted based on better estimated SNR, SINR, and CSI using AI technology, and PSK, QAM, etc. Table 2 shows an example of an existing AI-based AMC design.
[0042]
[0043] As shown in Table 2, supervised learning uses a convolutional neural network (CNN, see Non-Patent Document 5), uses estimated channel and noise as feedback information, and uses BPSK, QPSK, 16QAM, and 64QAM as constellations.
[0044] As shown in Table 2, in supervised learning, an artificial neural network (ANN, see Non-Patent Documents 6 and 7) is used, and when estimated SINR is used as feedback information, QAM is used as the constellation, and when CSI is used as feedback information, BPSK, 4QAM, 8QAM, and 16QAM are used.
[0045] As shown in Table 2, the reinforcement learning uses the Epsilon-greedy algorithm (see Non-Patent Document 8), and the feedback information is b / N 0 [dB] is used, and QPSK, 8PSK, and 16QAM are used as constellations.
[0046] As shown in Table 2, in reinforcement learning, the Q-learning algorithm (see Non-Patent Documents 9 and 10) is used, and when CIR (Channel Impulse Response) is used as feedback information, BPSK, QPSK, 8PSK, and 16QAM are used as constellations, and when the average SNR in all subcarriers of an OFDM symbol is used as feedback information, QPSK, 16QAM, and 64QAM are used as constellations.
[0047] As shown in Table 2, the reinforcement learning uses a deep Q-network, uses SNR and SINR as feedback information, and uses Gray-coded QPSK, 8PSK, 16PSK, and 32PSK as constellations, or BPSK, QPSK, 16QAM, and 64QAM as constellations.
[0048] The above-mentioned existing AI-based AMC schemes do not consider the effects of non-ideal factors such as phase noise and nonlinear PA on transmission performance, and the existing AMC schemes only consider phase noise in the case of low-order PSK modulation and low spectral efficiency.
[0049] As mentioned above, in the constellations of conventional modulation schemes, such as QAM and APSK, the phase noise generated by the RF oscillator and the nonlinearity of the PA have a significant impact on the transmission system. In addition, the conventional constellation design is limited and complicated. The conventional constellation design cannot continuously incorporate channel feedback information.
[0050] In addition, conventional AI-based AMC does not take into account nonlinear factors such as phase noise, which significantly impacts transmission performance. Furthermore, existing AMC that reduces the effects of phase noise only considers low-order PSK, which has low spectral efficiency.
[0051] Therefore, new modulation schemes and AMC may be designed as follows: For example, a constellation may be designed that takes nonlinear factors into account; For example, a set of adaptive modulation schemes may be designed based on PA and / or PN parameters; For example, a modulation with high spectral efficiency may be used.
[0052] Method 1) A modulation scheme that takes PN into account may be used. A constellation that reduces the effect of phase noise based on APSK may be designed, and a series of constellations may be designed for adaptive modulation that reduces the effect of phase noise based on NN.
[0053] Fig. 11 is a diagram for explaining an example (1) of a system according to an embodiment of the present invention. Fig. 11 shows an example of an end-to-end AI method related to optimization. For example, as shown in Fig. 11, a three-layer DNN may be arranged for nonlinear mapping on the transmitting side, and a three-layer DNN may be arranged for de-mapping on the receiving side. As shown in Fig. 11, a none-hot vector and a PN variance are input to an encoder, and the generated 2D constellation is input to a decoder via a channel together with the PN variance to obtain an output.
[0054] In the system shown in FIG. 11, to achieve finer AMC, the constellation may be automatically generated based on a neural network (NN) that assumes various PN distributions.
[0055] For offline training, the PN variance is input and learned. The NN is trained with a few limited values of the PN variance. The NN is generalized by inputting a large number of PN variances to generate a set of constellations that reduce the effects of phase noise.
[0056] For online deployment, an AI network is deployed at both the transmitter and receiver. Based on a given modulation order and an arbitrary PN distribution, a constellation suitable for that PN distribution is automatically generated.
[0057] Based on a set of trained NNs, constellations can be generated for specific modulation orders that mitigate the effects of changing PN variance.
[0058] Based on the constellation generated by the trained AI, the constellation points may be normalized to use 64 APSK with PN tolerance. To obtain an APSK constellation that reduces the effects of PN dispersion and generate a more normalized constellation, the optimized results generated by the trained AI may be used to determine the number of rings, the number of points, the radius or relative radius, and the initial phase of each ring.
[0059] Below we describe a series of constellation designs and adaptive modulations that mitigate the effects of phase noise.
[0060] 12 shows an example of a constellation generated by a system according to an embodiment of the present invention, where the PN variances are 0, 0.02, and 0.1.
[0061] Fig. 13 shows the performance of an example (1) of a constellation generated by a system according to an embodiment of the present invention. The constellation generated by a trained NN based on the PN variance is more resistant to phase noise than QAM. For example, as shown in Fig. 13, when the PN variance is 0.02, the SER is 10 -2 In this case, the SNR gain of the constellation generated by the trained NN is about 2 dB.
[0062] Note that QAM has lower performance than spiral constellations, as shown in Figure 13. The parameter f s is 0.00588.
[0063] Figure 14 shows the performance of example constellations (2) generated by a system according to an embodiment of the present invention, which are normalized constellations based on 64APSK to mitigate the effects of PN.
[0064] The normalization principle is as follows: By referring to the optimized results generated by the AI, the number of rings, number of points, radius or relative radius, and initial phase of each ring can be determined to obtain a PN-resistant APSK constellation and generate a more normalized constellation. The proposed 64-bit APSK has 16 rings, and each ring is arranged with four equally divided points. The radius and initial phase are determined by the constellation generated by the trained AI.
[0065] As shown in Figure 14, even the normalized constellation is more tolerant to phase noise than QAM. As shown in Figure 14, when the PN dispersion is 0.02, the SER is 10 -2 When , the SNR gain of the normalized constellation is about 2 dB.
[0066] Figure 15 shows the performance of example constellations (3) generated by a system according to an embodiment of the present invention, which are normalized constellations based on 64APSK to mitigate the effects of PN.
[0067] As shown in Figure 15, even the normalized constellation is more tolerant to phase noise than QAM. As shown in Figure 15, when the PN dispersion is 0.1, the SER is 5x10 compared to the spiral constellation. -2 When , the SNR gain of the normalized constellation is about 2 dB.
[0068] Fig. 16 shows the performance of an example constellation (4) generated by a system according to an embodiment of the present invention. Fig. 16 shows an example of a normalized constellation based on 256APSK that reduces the effects of PN. The proposed 256APSK has 32 rings, each of which has eight equally spaced points. The radius and initial phase are determined by the constellation generated by the trained AI.
[0069] Even the normalized constellation is more tolerant to phase noise than QAM, as shown in Figure 16. As shown in Figure 16, when the PN dispersion is 0.1, the SER is 4x10 compared to the spiral constellation. -2 When , the SNR gain of the normalized constellation is about 4 dB.
[0070] Method 2) A modulation scheme that takes nonlinear PA into consideration may be used. A constellation that reduces the impact of nonlinearity in the APSK-based PA may be designed, and a series of constellations may be designed for adaptive modulation that reduces the impact of nonlinearity in the NN-based PA.
[0071] Fig. 17 is a diagram for explaining an example (2) of a system according to an embodiment of the present invention. Fig. 17 shows an example of an end-to-end AI method related to optimization. For example, as shown in Fig. 17, a three-layer DNN may be arranged for nonlinear mapping on the transmitting side, and a three-layer DNN may be arranged for de-mapping on the receiving side. As shown in Fig. 17, the none-hot vector and PaSat are input to an encoder, and the generated 2D constellation is input to a decoder together with PaSat via a channel to obtain an output.
[0072] The parameter PaSat indicating the PA saturation point is the saturated output voltage of the PA Rapp model (see Non-Patent Document 15), and is defined by the position where the slope of the PA input / output characteristic curve becomes 0. If the amplitude of the signal input to the PA is larger than the threshold PaSat, nonlinear distortion occurs in the output signal.
[0073] Considering distortion caused by operation in the nonlinear power region of the PA, in the system shown in Fig. 17, to achieve more precise AMC, the constellation may be automatically generated based on a neural network that assumes various PA saturation point parameters PaSat.
[0074] For offline training, the NN is trained by inputting PaSat. The NN is trained by a few limited values of PaSat. The NN is generalized by inputting a large number of PaSats to generate a set of constellations that mitigate the effects of the nonlinearity of the PA.
[0075] For online deployment, an AI network is deployed at both the transmitter and receiver. Based on a given modulation order and an arbitrary PaSat, a constellation suitable for the nonlinearity of the PA is automatically generated.
[0076] Based on a set of trained NNs, constellations of specific modulation orders can be generated that mitigate the effects of varying PA nonlinearities.
[0077] Based on the constellation generated by the trained AI, the constellation points may be normalized to use 64 APSK, which is resistant to the nonlinearity of the PA. To obtain an APSK constellation that reduces the impact of PA nonlinearity and generate a more normalized constellation, the optimized results generated by the trained AI may be used to determine the number of rings, the number of points, the radius or relative radius, and the initial phase of each ring. The radius of each ring may be adjusted to obtain an APSK that suits different PA saturation points.
[0078] Below we describe a series of constellation designs and adaptive modulations that mitigate the effects of PA nonlinearities.
[0079] 18 shows the performance of example constellations (5) generated by a system according to an embodiment of the present invention. Examples of constellations are shown for PaSat of 0.45, 0.68, and 0.85. Furthermore, an example of a constellation after nonlinear PA is shown for PaSat of 0.68.
[0080] 19 shows the performance of an example (6) of a constellation generated by a system according to an embodiment of the present invention. The constellation generated by the trained NN based on PaSat is more tolerant to PA nonlinearity than QAM and APSK. For example, as shown in FIG. 19, when PaSat is 0.68, the SER is 2.2×10 compared to the 64APSK constellation. -3 In this case, the SNR gain of the constellation generated by the trained NN is about 3 dB. Note that, as shown in Figure 19, QAM has lower performance than the 64APSK constellation.
[0081] Figure 20 shows the performance of an example constellation (7) generated by a system according to an embodiment of the present invention. Figure 20 shows the proposed 64APSK before PA and the proposed 64APSK after PA when PaSat is 0.58. Note that 64APSK is composed of (6+12+20+26)APSK.
[0082] The proposed APSK constellation is more tolerant to PA nonlinearities than QAM and DVB standard APSK. For example, as shown in Figure 20, when PaSat is 0.58, the SER is 10 -3 The SNR gain of the proposed 64APSK constellation is about 2 dB when QAM is used. Note that QAM performs worse than the DVB standard 64APSK constellation, as shown in Figure 20.
[0083] Fig. 21 shows the performance of the example constellation (8) generated by the system according to the embodiment of the present invention. In Fig. 20, the proposed hexagonal constellation before PA and the proposed hexagonal constellation after PA are shown for PaSat = 0.58.
[0084] The proposed hexagonal constellation is more tolerant to PA nonlinearities than QAM and DVB standard APSK. For example, as shown in Figure 21, when PaSat is 0.58, the SER is 10-3 The SNR gain of the proposed hexagonal constellation is about 2 dB when . Note that QAM performs worse than the DVB standard 64APSK constellation, as shown in Figure 21.
[0085] A hexagonal constellation consists of multiple concentric hexagonal rings, divided into an outer ring and several smaller inner rings. This constellation is tolerant to PA nonlinearities and exhibits good performance. Constellations that match different PA saturation points can be obtained by setting the radius or relative radius of each ring.
[0086] Constellation design and adaptive modulation under the influence of non-ideal factors may be performed as follows:
[0087] Operation 1) Constellation design and adaptive modulation to mitigate the effects of PN Operation 1-1) Non-regular constellation design to mitigate the effects of PN Operation 1-1-1) Definition of NN parameters Operation 1-1-2) Bit-to-symbol (hereinafter also referred to as "B2S") mapping table design for non-regular constellations Operation 1-2) Normalized APSK constellation design to mitigate the effects of PN Operation 1-2-1) Definition of APSK Operation 1-2-2) B2S mapping table design for APSK Operation 1-3) MCS table design to mitigate the effects of PN and related signaling for adaptive modulation Operation 1-3-1) MCS table design including PN-resistant constellations Operation 1-3-2) Related signaling suitable for using multiple PN-resistant MCS tables
[0088] Operation 2) Constellation design and adaptive modulation to mitigate the effects of PA nonlinearity Operation 2-1) Non-regular constellation design to mitigate the effects of PA nonlinearity Operation 2-1-1) Definition of NN parameters Operation 2-1-2) B2S mapping table design for non-regular constellations Operation 2-2) Normalized constellation design to mitigate the effects of PA nonlinearity Operation 2-2-1) Hexagonal constellation design Operation 2-2-2) APSK design Operation 2-3) MCS table design to mitigate the effects of PA nonlinearity and related signaling for adaptive modulation
[0089] Operation 3) Constellation design and adaptive modulation to mitigate the effects of PN and PA nonlinearities Operation 3-1) Non-regular constellation design to mitigate the effects of PN and PA nonlinearities Operation 3-1-1) Definition of NN parameters Operation 3-1-2) B2S mapping table design for non-regular constellations Operation 3-2) Normalized constellation design to mitigate the effects of PN and PA nonlinearities Operation 3-2-1) Definition of APSK Operation 3-2-2) B2S mapping table design for APSK Operation 3-3) MCS table design to mitigate the effects of PN and PA nonlinearities and related signaling for adaptive modulation
[0090] Operation 4) UE capability definition
[0091] Operation 1) Constellation design and adaptive modulation to mitigate the effects of PN will be described below.
[0092] Below, operation 1-1) irregular constellation design to reduce the influence of PN will be explained.
[0093] A non-normal constellation generated by an end-to-end (hereinafter also referred to as "E2E") learning neural network may be used in a system that reduces the influence of PN, such as the system that performs adaptive modulation based on the input of PN variance shown in Fig. 11. The PN variance is defined as one of the input parameters of the neural network.
[0094] The basic principle is as follows: the NN performs offline learning under limited channel conditions, such as a given SNR range and some PN variance, and then the NN is generalized based on more channel conditions (continuous or fine PN variance) to obtain a set of modulation constellations that mitigate the effects of PN.
[0095] For example, offline learning can be performed as follows: 2 N, and construct a NN architecture. 2 PN,1 , σ 2 PN,2 , ..., σ 2 PN,L}, the NN is trained to learn the constellation.
[0096] PN variance σ 2 PN,l and N one-hot encoder vectors are used as inputs to perform E2E learning. j , j=1,...,N is a vector of length N where the jth value is 1 and the other values are 0. For example, e j may be associated with the j-th constellation point.
[0097] The L PN variances are used for training so that the network has the ability to generalize to acquire constellations that vary with the PN. Note that the PN may be considered at both the transmitter and receiver sides.
[0098] The above operation can improve performance in an environment where multiple PN distributions exist.
[0099] For example, the system may train a NN to obtain constellations corresponding to PN variances of 0, 0.02, 0.06, and 0.1, and then generalize the constellations to obtain constellations corresponding to PN variances of 0, 0.01, and 0.1. Note that as the PN variance increases, the arrangement of constellation points tends to become narrower on the inside and wider on the outside. The constellation diagram gradually changes from a circle to several branches and then to several wings.
[0100] Operation 1-1-1) Definition of NN parameters will be explained below.
[0101] FIG. 22 is a diagram illustrating an example of a system for mitigating the effects of phase noise according to an embodiment of the present invention. The parameters of the offline trained neural network may include weights, biases, and activation functions. These parameters are defined and used in the transmitter and receiver. A pre-stored neural network generates constellation points based on the modulation order and PN variance. As shown in FIG. 22, bit-to-index mapping is added before the neural network module converts the modulated bit sequence into a neural network one-hot encoder vector.
[0102] Let module 1 be an offline trained neural network that generates constellation points for a given modulation order and PN distribution.
[0103] Module 2 is the bit-to-index mapping. Module 2 realizes the function of mapping bits to symbols for online deployment. During the offline training phase, the input parameters include the none-hot encoder vector and the PN variance of the channel. In the online deployment phase of the communication system, the bit sequence may be mapped to the corresponding constellation points that were deviated during the offline training phase. The bit-to-index mapping is added in front of the predefined NN to generate the corresponding one-hot encoder vector (i.e., the input of the NN).
[0104] Through the combination of bit-to-index mapping and NN, the bit sequence is modulated onto a constellation point that is PN-resistant.
[0105] By generalizing a NN trained offline with a coarse PN distribution, a constellation that is resistant to any PN distribution can be dynamically generated based on the dynamic PN distribution in a real system.
[0106] The system shown in FIG. 22 uses a modulation order log 2 N, PN variance σ 2 PN and has length log 2 Bit-to-index mapping of N and offline trained NN (N and σ 2 PN ) by the bit-to-symbol mapping and variance σ 2 PN This generates a constellation that is resistant to PN noise.
[0107] As shown in Figure 22, on the transmitter side, a bit sequence is input to bit-to-index mapping, and a one-hot encoder vector is output. The one-hot encoder vector is input to an offline trained NN, and a modulation symbol is output. Symbols received at the receiver side from the transmitter side via a channel are input to the offline trained NN, and a one-hot encoder vector is output. The one-hot encoder vector is input to bit-to-index demapping, and a bit sequence is output.
[0108] The parameters of the NN to be trained offline may be defined by at least {the number of layers of the NN, the weight of each layer, the bias of each layer, and the activation function of each layer}.
[0109] Table 3 is an example of a table for parameters of an offline training NN for all PN distributions.
[0110]
[0111] A neural network mainly has three basic elements: weights, biases, and activation functions. Weights are the strength of connections between neurons, which are represented by weights. The weight values indicate the probability or proportion of each input element. Bias is a setting to accurately classify samples. Bias is an important parameter for the model and is equivalent to adding a constant. Activation functions act as nonlinear mappings, generally restricting the output values from neurons to a predetermined range, such as -1 to 1 or 0 to 1.
[0112] The weight and bias values are results obtained from the trained network and may be stored, for example, in a text file format.
[0113] The common activation function may be, for example, a sigmoid, a hyperbolic tangent (tanh), a rectified linear unit (ReLU), or the like.
[0114] A normalization layer is essential in the encoder to normalize the power of the constellations produced by the system.
[0115] For a given modulation order, the input parameters of the NN include at least one of 1) and 2) below.
[0116] 1) One-hot encoder vector e j , j=1,...,N is a vector of length N, where the j-th value is 1 and the others are 0. 2) PN variance
[0117] Table 4 shows examples of NN parameters for PN-resistant 16-point and 64-point constellations.
[0118]
[0119] In the example shown in Table 4, the weights, biases, and activation functions each consist of three layers. As shown in Table 4, the parameters may be specified as a matrix for the weights, a 1D array for the biases, and a sigmoid or tanh function for the activation function.
[0120] The bit-to-index mapping is explained below. 2 The N bit sequences are transformed into one-hot encoder vectors of length N for input to the NN. Option 1)-Option 4) below may be adopted.
[0121] Option 1) A binary number, which is a bit sequence, may be converted to a decimal number, which is an index of a constellation point in the neural network. This design is suitable for constellations generated online for any PN distribution. For example, the PN distribution during offline training does not need to be considered. Table 5 is an example of a 16-point constellation for option 1.
[0122]
[0123] Option 2) Bit-to-index mapping, which satisfies gray-mapping, may be performed for the constellation points in the neural network. Gray-mapping is expected to improve performance. Table 6 shows an example of a 64-point constellation for option 2.
[0124]
[0125] Option 3) An optimized mapping may be performed to minimize the error rate or maximize the throughput / SE, which can improve the decoding performance.
[0126] Option 4) The bit-to-index mapping module may be separated or joined by a neural network or other AI network to learn about the constellation points.
[0127] The transformed index j is used to generate the one-hot encoder vector e j is generated and input to the NN.
[0128] Operation 1-1-2) Bit-to-symbol mapping table design for non-regular constellations will be described below.
[0129] A bit-to-symbol mapping table may be designed to define non-normal constellations with PN resistance. Similar to the conventional definition, a bit-to-symbol mapping table may be defined to modulate a bit sequence onto a constellation point. This allows the constellation point and bit-to-symbol mapping to be combined and defined in one table. Either option 1) or option 2) shown below may be adopted.
[0130] Option 1) A bit-to-symbol mapping table may be generated based on the branches of Gray mapping, where the difference between bits of adjacent constellation points in each branch is only one bit.
[0131] Fig. 23 is a diagram showing an example (1) of a constellation generated by a system for reducing the influence of phase noise according to an embodiment of the present invention. Fig. 23 shows an example of a trained 16-bit constellation for a PN variance of 0.02 and option 1) of a trained 64-bit constellation for a PN variance of 0.02. Bit-to-symbol mapping is performed in the order of the arrows shown in Fig. 23. Table 7 shows an example of a bit-to-symbol mapping table for a trained 16-bit constellation for a PN variance of 0.02. Note that, similarly, the real part of the symbols in the table corresponds to the in-phase component and the imaginary part corresponds to the quadrature component.
[0132]
[0133] Option 2) A Bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of one ring of the same amplitude or adjacent constellation points of similar amplitude is 1 bit.
[0134] Fig. 24 is a diagram showing an example (2) of a constellation generated by a system for reducing the influence of phase noise according to an embodiment of the present invention. Fig. 24 shows an example of option 2) of a trained 16-constellation for a PN variance of 0.06, a trained 64-constellation for a PN variance of 0.1, and a trained 256-constellation for a PN variance of 0.02. Bit-to-symbol mapping is performed in the order of the clockwise arrows shown in Fig. 24. Table 8 is an example of a bit-to-symbol mapping table for a trained 16-constellation for a PN variance of 0.06.
[0135]
[0136] Below, operation 1-2) normalized APSK constellation design that reduces the influence of PN will be described.
[0137] The normalized constellation based on APSK is designed based on the PN dispersion, and the parameters of APSK are adjusted according to the PN dispersion to improve the robustness.
[0138] APSK is composed of multiple concentric rings as shown in equation (1).
[0139]
[0140] APSK is a signal with parameters {L, R k , n k , φ k} and optimized based on requirements and scenarios.
[0141] Operation 1-2-1) Definition of APSK will be explained below.
[0142] The normalized APSK constellation is given by the parameters {L, R k , n k , φ k, k = 1, ..., L}. Based on the constellation generated by the learned AI in the above operation 1-1), the constellation points may be normalized to design a 64-APSK with PN resistance. The result of the optimization by the AI is the parameters of the APSK with PN resistance, the number of rings L, the number of points in each ring n k , k=1,...,L, radius or relative radius R k , k=1, . . . , L and the initial phase φ of each ring k , k=1,...,L. By normalizing the non-normal constellation according to operation 1-1) and converting it to APSK, performance and complexity can be optimized. By the above operation, the normalized constellation points reduce the demodulation complexity.
[0143] Fig. 25 is a diagram showing an example (3) of a constellation generated by a system for reducing the influence of phase noise according to an embodiment of the present invention. Fig. 25 is an example of 64APSK for PN variance 0.02, in which the learned constellation obtained by operation 1-2-1) is normalized. Table 9 shows the radius, phase, and bit labeling of each ring of the 64APSK in Fig. 25.
[0144]
[0145] As shown in FIG. 25 and Table 9, the number of rings is 16, the number of points in each ring is 4, and adjacent constellation points in each ring differ in phase by 90 degrees.
[0146] FIG. 26 is a diagram showing an example (4) of a constellation generated by a system for reducing the influence of phase noise according to an embodiment of the present invention. FIG. 26 shows an example of 64APSK for PN dispersion 0.02 obtained by normalizing the learned constellation according to operation 1-2-1), and an example of 64APSK for PN dispersion 0.1 obtained by normalizing the learned constellation. Table 10 is a table showing the radius and initial phase of each ring of the normalized 64APSK for PN dispersion 0.02 shown in FIG. 26. Table 11 is a table showing the radius and initial phase of each ring of the normalized 64APSK for PN dispersion 0.02 shown in FIG.
[0147]
[0148]
[0149] In the example of 64APSK for a normalized PN dispersion of 0.02 in Figure 26, the number of rings is 16, the number of points in each ring is 4, and the points in each ring are equally spaced. The initial phase of two adjacent rings is 20 degrees or 60 degrees as shown in Table 10.
[0150] In the example of 64APSK for normalized PN dispersion 0.1 in Figure 26, the number of rings is 16, the number of points in each ring is 4, and the points in each ring are equally spaced. The minimum phase from the original four points may be used as the initial phase of the ring.
[0151] Operation 1-2-2) B2S mapping table design for APSK will be described below.
[0152] A bit-to-symbol mapping table may be designed to define a normalized constellation that is PN-resistant. Either option 1) or option 2) shown below may be adopted.
[0153] Option 1) A bit-to-symbol mapping table may be generated based on the branches of Gray mapping, where the difference between bits of adjacent constellation points in each branch is only one bit.
[0154] Option 2) A Bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of one ring of the same amplitude or adjacent constellation points of similar amplitude is 1 bit.
[0155] Figure 27 shows an example (5) of a constellation generated by a system for mitigating the effects of phase noise according to an embodiment of the present invention. Figure 27 shows an example of option 1) of a 64-bit constellation for a normalized PN dispersion of 0.02, and an example of option 2) of a 64-bit constellation for a normalized PN dispersion of 0.1. Bit-to-symbol mapping is performed in the order of the arrows shown in Figure 27. Table 12 shows an example of a bit-to-symbol mapping table for a 64-bit constellation for a normalized PN dispersion of 0.02.
[0156]
[0157] Table 13 shows an example of a bit-to-symbol mapping table for 64 constellations for a normalized PN variance of 0.1.
[0158]
[0159] Below, we will explain operations 1-3) MCS table design to mitigate the effects of PN and related signaling for adaptive modulation.
[0160] To better suit application scenarios and requirements, an MCS table may be designed that includes constellations with PN resistance and high SE.
[0161] Hereinafter, operation 1-3-1) MCS table design including PN-resistant constellations will be described.
[0162] The constellation generated by the trained NN in operation 1-1) or the APSK normalized in operation 1-2) may be included in the PN-resistant MCS table. Note that, hereinafter, the constellation generated by the trained NN in operation 1-1) or the APSK normalized in operation 1-2) will be referred to as "newPN" without loss of generality.
[0163] To avoid signaling overhead, constellations designed for several PN distributions may be included in the MCS table. For example, the MCS table may be configured as shown in Option 1) to Option 3) below.
[0164] Also, for example, a constellation designed for a PN dispersion of 0.02 may be defined in the MCS table of option 1) below. Also, for example, constellations designed for PN dispersions of 0.02 and 0.06 may be defined in the MCS table of option 2) below. Also, for example, constellations designed for PN dispersions of 0.02, 0.04, and 0.08 may be defined in the MCS table of option 3) below.
[0165] Option 1) Some QAM MCSs in the 5-bit NR MCS table may be replaced with PN-resistant constellations. Some QAM MCSs in the NR MCS table may be replaced with MCSs corresponding to the constellations generated by operation 1-1) or the APSK normalized by operation 1-2). Replacements may be performed as shown in Examples 1) to 3) below.
[0166] Example 1) To reduce the number of MCSs replaced by new constellations, a constellation for one PN distribution may be added to the MCS table. The MCS index associated with 16QAM may be replaced with the constellation "16newPN" for a 16-point MCS with a PN distribution of 0.06, which is a modulation order 4 MCS. The MCS index associated with 64QAM may be replaced with the constellation "64newPN" for a 64-point MCS with a PN distribution of 0.04, which is a modulation order 6 MCS. The MCS index associated with 256QAM may be replaced with the constellation "256newPN" for a 256-point MCS with a PN distribution of 0.02, which is a modulation order 8 MCS. Table 14 is an example of an MCS table for Example 1).
[0167]
[0168] Example 2) Taking into account the PN resistance for different modulation orders, different PN distribution numbers for the constellation may be applied to different modulation orders. For example, for a 64-point constellation with a modulation order of 6, the MCS indexes associated with 64QAM may be partially replaced with the constellation "64newPN-mid" for a PN distribution of 0.04 and the constellation "64newPN-high" for a PN distribution of 0.08. On the other hand, for a 256-point constellation with a modulation order of 8, the MCS indexes associated with 256QAM may be partially replaced with the constellation "256newPN-low" for a PN distribution of 0.02.
[0169] Example 3) A partial index of BPSK, QPSK, or QAM in a conventional MCS table may be replaced with a newPN for which the PN dispersion is not explicitly defined. That is, the PN dispersion may be obtained from explicit UE feedback instead of MCS notification. The constellation of the newPN may be determined based on a combination of the MCS notification and PN dispersion feedback. For example, if MCS16newPN is signaled and PN dispersion 0.02 is fed back, 16newPN may be configured for PN dispersion 0.02 and used for transmission.
[0170] Option 1) above provides an MCS table containing PN-resistant MCSs, and the total number of MCSs and MCS signaling overhead remains unchanged.
[0171] Option 2) A new MCS corresponding to the PN-resistant constellation generated by operation 1-1) or the APSK normalized by operation 1-2) may be added to the 5-bit NR MCS table to generate a new MCS table exceeding 5 bits. An MCS corresponding to the constellation generated by operation 1-1) or the APSK normalized by operation 1-2) may be added to the NR MCS table. Addition may be performed as shown in Examples 1) to 3) below.
[0172] Example 1) To increase the number of MCSs that can be replaced by new constellations, constellations for PN variances of 2 or 3 may be added to the MCS table. The constellation "16newPN-low" for a 16-point MCS with a modulation order of 4 and a PN variance of 0.02 and the constellation "16newPN-high" for a PN variance of 0.08 may be added.
[0173] Alternatively, a constellation "16newPN-mid" for a 16-point PN variance of 0.04, which is an MCS with a modulation order of 4, or a constellation "16newPN-high" for a PN variance of 0.08 may be added. Alternatively, a constellation "16newPN-low" for a 16-point PN variance of 0.02, a constellation "16newPN-mid" for a PN variance of 0.04, or a constellation "16newPN-high" for a PN variance of 0.08 may be added.
[0174] It is also possible to add a constellation "64newPN-low" for a 64-point MCS with a PN variance of 0.02 and a constellation "64newPN-high" for a PN variance of 0.08, which is a modulation order 6 MCS. Alternatively, it is also possible to add a constellation "64newPN-mid" for a 64-point MCS with a PN variance of 0.04 and a constellation "64newPN-high" for a 64-point MCS with a PN variance of 0.08. Alternatively, it is also possible to add a constellation "64newPN-low" for a 64-point MCS with a PN variance of 0.02, a constellation "64newPN-mid" for a PN variance of 0.04, and a constellation "64newPN-high" for a PN variance of 0.08, which is a modulation order 6 MCS.
[0175] It is also possible to add a constellation "256newPN-low" for a 256-point PN variance of 0.02, which is an MCS for modulation order 8, and a constellation "256newPN-high" for a PN variance of 0.08. Alternatively, it is also possible to add a constellation "256newPN-mid" for a 256-point PN variance of 0.04, which is an MCS for modulation order 8, and a constellation "256newPN-high" for a PN variance of 0.08. Alternatively, it is also possible to add a constellation "256newPN-low" for a 256-point PN variance of 0.02, a constellation "256newPN-mid" for a PN variance of 0.04, and a constellation "256newPN-high" for a PN variance of 0.08, which are MCS for modulation order 8.
[0176] Tables 15 and 16 are examples of MCS tables for Example 1).
[0177]
[0178]
[0179] Example 2) High modulation orders and / or high code rates are susceptible to phase noise. Different PN distributions may be applied to different modulation orders and code rates. For example, "16newPN-high" may be added for a 16-point constellation with a PN distribution of 0.08. For example, "64newPN-low" may be added for a 64-point constellation with a PN distribution of 0.02, "64newPN-mid" may be added for a 64-point constellation with a PN distribution of 0.04, and "64newPN-high" may be added for a 256-point constellation with a PN distribution of 0.02.
[0180] Example 3) A partial index of BPSK, QPSK, or QAM in a conventional MCS table may be replaced with a newPN for which the PN dispersion is not explicitly defined. That is, the PN dispersion may be obtained from explicit UE feedback instead of MCS notification. The constellation of the newPN may be determined based on a combination of the MCS notification and PN dispersion feedback. For example, if MCS16newPN is signaled and PN dispersion 0.02 is fed back, 16newPN may be configured for PN dispersion 0.02 and used for transmission.
[0181] Option 2) above allows the addition of a PN-resistant MCS while maintaining the traditional MCS.
[0182] Option 3) A new MCS table may be configured that includes only PN-resistant constellations. The MCS table may be configured based on existing PN-resistant constellations, such as BPSK, pi-2BPSK, pi-4BPSK, and pi-4QPSK, as well as newly designed constellations. The MCS table may include MCS indexes including 16newPN, 64newPN, and 256newPN. Table 17 is an example of an MCS table for option 3).
[0183]
[0184] Option 3) above allows for the provision of an MCS table containing a PN-resistant MCS. Multiple sets of MCS tables can be configured, allowing for flexible selection of different MCS tables and adjusting the cost required for MCS indication.
[0185] Below, operation 1-3-2) related signaling suitable for using multiple MCS tables with PN tolerance will be described.
[0186] Signaling actions may be performed as shown below in options 1)-3).
[0187] Option 1) When the MCS table is completely replaced with a newly designed MCS table (e.g., Option 1) or Option 2) in Operation 1-3-1), the conventional MCS notification method and field (e.g., I in NR) are replaced. MCS ) or signaling related to MCS notification in future systems may be reused.
[0188] Option 2) If the NN parameters are pre-stored in the transmitter and receiver, a combination of MCS and PN distribution may be used to generate the constellation. The PN distribution may be fed back via CSI signaling or new signaling at the physical layer or higher layers (e.g. RRC, MAC-CE, UCI, etc.).
[0189] Option 3) When a newly designed MCS table (e.g., Option 1), Option 2) or Option 3) of Operation 1-3-1) is added to the specification, the following notification method may be adopted.
[0190] Option 3-1) New Signaling I flag may signal whether to use the MCS table containing the constellation of the newPN. flag may be transmitted via RRC, MAC-CE, UCI or DCI signaling.
[0191] Alt. 1 (DL): The UE may feed back the MCS index for DL transmission to the BS. The UE may also provide PN-tolerant MCS notification (I flag ) and MCS index (I MCS ) may be sent as additional feedback to the BS. If PN-resistant MCS notification is required, flag is set to 1, otherwise I flag may be set to 0. flag may be transmitted via RRC, MAC-CE or UCI signaling. MCS The feedback may be sent via existing signaling.
[0192] The BS determines the required MCS table and I based on the UE feedback. MCS First, I flag If Q is 1, the MCS table including newPN may be selected, otherwise the conventional MCS table may be selected. m , the target coding rate R is I MCS and may be determined based on the MCS table.
[0193] Alt. 2 (DL): The BS may actively select the MCS for DL transmission. New RRC, MAC-CE, DCI, or UCI signaling may be designed to feed back the PN variance from the UE. The BS selects whether to use a PN-tolerant MCS table based on the PN feedback from the UE. If the PN variance is near zero, the BS uses the conventional MCS table, and if the PN variance is near zero, the BS uses the I flag may be set to 0. If the PN variance is in the range of 0.01 to 0.1, select the MCS table containing the newPN constellation, and flag may be set to 1. The BS may flag and I MCS may be sent to the UE via RRC, MAC-CE or DCI signaling.
[0194] Alt. 1 (UL): The BS may signal the UL MCS to the UE. The BS may signal the PN-tolerant MCS (I flag ) and MCS index (I MCS If PN-resistant MCS notification is required, I flag is set to 1, otherwise I flag may be set to 0. flag may be transmitted via RRC, MAC-CE or UCI signaling. MCS may be transmitted via existing signaling.
[0195] The UE acquires the required MCS table and I based on the BS instruction. MCS First, I flagIf Q is 1, the MCS table including newPN may be selected, otherwise the conventional MCS table may be selected. m , the target coding rate R is I MCS and may be determined based on the MCS table.
[0196] Alt. 2 (UL): The UE may actively select an MCS for UL transmission. The UE selects whether to use a PN-resistant MCS table based on the estimated PN variance. If the PN variance is near zero, the UE uses a conventional MCS table; if the PN variance is near zero, the UE uses a conventional MCS table; flag may be set to 0. If the PN variance is in the range of 0.01 to 0.1, select the MCS table containing the newPN constellation, and flag The UE may set I flag and I MCS may be sent to the BS via RRC, MAC-CE or UCI signaling.
[0197] Option 3-2) The existing MCS selection method and signaling definition method in 3GPP TS 36544 or future systems may be reused. New signaling may be added to indicate whether to use an MCS table containing PN-resistant constellations. This signaling may be transmitted via higher or lower layer signaling, such as RRC, MAC-CE, UCI, or DCI signaling.
[0198] Alt. 1: The signaling may be explicitly associated with a PN. For example, in the case of DL transmission, the process of 3GPP TS 36.11.1 may be reused. If mcs-Table DCI-1-2 is set to PhaseNoise or PN, the UE may use an MCS table containing the N-newPN constellation. MCS and an MCS table including N-new PN constellations, m and may be used to determine the target coding rate R.
[0199] For example, for UL transmission, the higher layer parameters ifPhaseNoise or ifPN may be added. If ifPhaseNoise or ifPN is 1, an MCS table containing the N-newPN constellation may be used for PUSCH.
[0200] Alt. 2: The signaling may be implicitly associated with the PN. For example, in systems configured with sub-THz bands, an MCS table including a newly designed PN-resistant constellation may be used. In other systems, an MCS table independent of the PN may be used.
[0201] Here, conventional reference signals and estimation methods may be considered to estimate PN variance and / or CSI-related information (e.g., SNR, SINR, CQI, CSI, etc.) In addition, AI may be used to estimate PN variance and / or CSI-related information for selecting MCS.
[0202] Operation 2) Constellation design and adaptive modulation to mitigate the effects of PA nonlinearity will be described below.
[0203] Below, operation 2-1) non-regular constellation design to mitigate the influence of PA nonlinearity will be described.
[0204] A non-normal constellation generated by an end-to-end (hereinafter also referred to as "E2E") learning neural network may be used in a system operating in the non-linear region of a non-linear PA, such as the system performing adaptive modulation based on the input of PaSat shown in Fig. 17. The PA saturation point is defined as one of the input parameters of the neural network.
[0205] The basic principle is as follows: the NN performs offline learning under limited channel conditions, such as a predetermined SNR range and some PA saturation points, and then the NN is generalized based on more channel conditions (continuous or fine PA saturation points) to obtain a set of modulation constellations that mitigate the impact of PA nonlinearity.
[0206] For example, offline learning can be performed as follows: 2 Let N be the number of PA saturation points, and construct a NN architecture. Train the NN to learn the constellation under the condition of L PA saturation points.
[0207] E2E learning is performed using the PA saturation point PaSat and the N one-hot encoder vector as input. j , j=1,...,N is a vector of length N where the jth value is 1 and the other values are 0. For example, e j may be associated with the j-th constellation point.
[0208] The L PA saturation points are used for training so that the network has the ability to generalize to acquire constellations that vary with the PA saturation point. Note that the PA saturation points may be considered both at the transmitter and receiver sides.
[0209] The above operation can improve performance in an environment where multiple PA saturation points exist.
[0210] The outline of the PA model is as follows (see Non-Patent Document 15).
[0211] FIG. 28 is a diagram showing an example of PA characteristics. The modeling of the PA nonlinearity taken into consideration is the Rapp model, which targets amplitude distortion and a solid-state power amplifier (SSPA). Phase distortion caused by the PA is large and is directly pre-distorted and mapped back. The PA saturation point parameter PaSat is the saturated output voltage of the PARapp model and is defined as the point where the slope of the PA input / output characteristic curve drops to zero. When the input signal amplitude of the PA is greater than the threshold PaSat, nonlinear distortion occurs. The amplitude distortion (AM-AM) in the Rapp model is expressed by Equation 2.
[0212]
[0213] A is the amplitude of the input signal, g is the small gain signal, s is the smoothness factor, Asat is the saturation level, or PaSat.
[0214] For example, constellations may be trained at PA saturation points of 0.45, 0.58, 0.68, 0.85, etc. If the PA saturation point is small, the nonlinearity becomes stronger, resulting in increased density in the center of the constellation and sparseness at the outside.
[0215] Operation 2-1-1) Definition of NN parameters will be explained below.
[0216] FIG. 29 is a diagram illustrating an example of a system for mitigating the effects of nonlinear PA according to an embodiment of the present invention. The parameters of the offline trained neural network may include weights, biases, and activation functions. These parameters are defined and used in the transmitter and receiver. The pre-stored neural network generates constellation points based on the modulation order and PN variance. As shown in FIG. 29, bit-to-index mapping is added before the neural network module converts the modulated bit sequence into the neural network one-hot encoder vector.
[0217] Let Module 1 be an offline trained NN. Module 1 generates constellation points for a given modulation order and PA saturation point.
[0218] Module 2 is the bit-to-index mapping. Module 2 realizes the function of mapping bits to symbols for online deployment. During the offline training phase, the input parameters include the none-hot encoder vector and the PN variance of the channel. In the online deployment phase of the communication system, the bit sequence may be mapped to the corresponding constellation points that were deviated during the offline training phase. The bit-to-index mapping is added in front of the predefined NN to generate the corresponding one-hot encoder vector (i.e., the input of the NN).
[0219] Through the combination of bit-to-index mapping and NN, the bit sequence is modulated onto constellation points that are tolerant to the nonlinearity of the PA.
[0220] By generalizing the NN trained offline with a coarse PA saturation point, constellations that are tolerant to any PA saturation point can be dynamically generated based on the dynamic PA saturation point in a real system.
[0221] As shown in FIG. 29, on the transmitter side, a bit sequence is input to bit-to-index mapping, and a one-hot encoder vector is output. The one-hot encoder vector is input to an offline trained NN, and a modulated symbol is output. The modulated symbol becomes a transmitted symbol starting from PA, which is the saturation point PaSat. The symbol received on the receiver side from the transmitter side via the channel is input to the offline trained NN, and a one-hot encoder vector is output. The one-hot encoder vector is input to bit-to-index demapping, and a bit sequence is output.
[0222] The parameters of the NN to be trained offline may be defined by at least {the number of layers of the NN, the weight of each layer, the bias of each layer, and the activation function of each layer}.
[0223] Table 3 is an example of a table for parameters of a NN for offline training for all PA saturation points.
[0224] A neural network mainly has three basic elements: weights, biases, and activation functions. Weights are the strength of connections between neurons, which are represented by weights. The weight values indicate the probability or proportion of each input element. Bias is a setting to accurately classify samples. Bias is an important parameter for the model and is equivalent to adding a constant. Activation functions act as nonlinear mappings, generally restricting the output values from neurons to a predetermined range, such as -1 to 1 or 0 to 1.
[0225] The weight and bias values are results obtained from the trained network and may be stored, for example, in a text file format.
[0226] The common activation function may be, for example, a sigmoid, a hyperbolic tangent (tanh), a rectified linear unit (ReLU), or the like.
[0227] A normalization layer is essential in the encoder to normalize the power of the constellations produced by the system.
[0228] For a given modulation order, the input parameters of the NN include at least one of 1) and 2) below.
[0229] 1) One-hot encoder vector e j , j=1,...,N is a vector of length N, where the j-th value is 1 and the others are 0. 2) PA saturation point
[0230] Table 4 shows examples of NN parameters for 16-point and 64-point constellations that are tolerant to PA nonlinearities.
[0231] In the example shown in Table 4, the weights, biases, and activation functions each consist of three layers. As shown in Table 4, the parameters may be specified as a matrix for the weights, a 1D array for the biases, and a sigmoid or tanh function for the activation function.
[0232] The bit-to-index mapping is explained below. 2 The N bit sequences are transformed into one-hot encoder vectors of length N for input to the NN. Option 1)-Option 4) below may be adopted.
[0233] Option 1) A binary number, which is a bit sequence, may be converted to a decimal number, which is an index of a constellation point in the neural network. This design is suitable for constellations generated online for any PA saturation point. For example, the PA saturation point during offline training does not need to be considered. Table 5 shows an example of a 16-point constellation for option 1.
[0234] Option 2) Bit-to-index mapping, which satisfies gray-mapping, may be performed for the constellation points in the neural network. Gray-mapping is expected to improve performance. Table 6 shows an example of a 64-point constellation for option 2.
[0235] Option 3) An optimized mapping may be performed to minimize the error rate or maximize the throughput / SE, which can improve the decoding performance.
[0236] Option 4) The bit-to-index mapping module may be separated or joined by a neural network or other AI network to learn about the constellation points.
[0237] The transformed index j is used to generate the one-hot encoder vector e j is generated and input to the NN.
[0238] Below, operation 2-1-2) B2S mapping table design for irregular constellations will be explained.
[0239] Bit-to-symbol mapping tables may be designed to define non-normal constellations with PN immunity. Similar to conventional definitions, bit-to-symbol mapping tables may be defined to map bit sequences onto constellation points. This allows the constellation points and bit-to-symbol mapping to be combined and defined in a single table.
[0240] The bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of one ring of the same amplitude or adjacent constellation points of similar amplitude is 1 bit.
[0241] Fig. 30 is a diagram showing an example (1) of a constellation generated by a system for mitigating the effects of nonlinear PA according to an embodiment of the present invention. Fig. 30 shows examples of a trained 16-constellation for PaSat0.58, a trained 64-constellation for PaSat0.85, and a trained 256-constellation for PaSat0.58. Bit-to-symbol mapping is performed in the order of the clockwise arrows shown in Fig. 30. Table 18 is an example of a bit-to-symbol mapping table for a trained 16-constellation for PaSat0.58.
[0242]
[0243] Below, operation 2-2) normalized constellation design that reduces the influence of PA nonlinearity will be described.
[0244] The normalized constellation is designed based on the PA saturation point and the constellation learned in operation 2-1).
[0245] Operation 2-2-1) Hexagonal Constellation Design will be described below.
[0246] The normalized constellation based on the hexagonal constellation is designed based on the PA saturation point. k} may be adjusted based on the PA saturation point to increase tolerance to nonlinear PAs.
[0247] The hexagonal constellation is composed of L concentric hexagons. The generation method may be steps 1 to 4 shown below.
[0248] FIG. 31 is a diagram showing an example (2) of a constellation generated by a system for mitigating the influence of a nonlinear PA according to an embodiment of the present invention.
[0249] In step 1, a 1 / 6 constellation is generated as follows: The i-th constellation point located at the edge of the k-th layer is expressed as Equation 3.
[0250]
[0251] For example, the absolute radius of the L-th hexagon {R 1 , R 2 , ..., R L} is the L-1th associated radius {r 1 , r 2 , ..., r L-1} in r l =R l+1 / R 1 This may be equivalent to:
[0252] As shown in Figure 31, in step 2, rotate the 1 / 6 constellation to obtain the entire constellation. In general, the number of constellation points is 2 n For example, suppose we rotate a five-layer 1 / 6 constellation to obtain a perfect concentric hexagonal constellation. However, the resulting regular hexagonal constellation has 1 + 6 + 12 + 18 + 24 = 61 points.
[0253] As step 3, to obtain a log2N order hexagonal constellation that is tolerant to the nonlinearity of the PA, add the remaining constellation points to the edges of the hexagon as shown in Figure 31. Either option 1) or option 2) below may be performed.
[0254] Option 1) The AI-optimized constellation may be referenced to determine which edge to insert the constellation point. For example, nAssume there are three points remaining to make it 1. One is inserted on the edge of the fourth layer (k=3). The other two are inserted on the fifth layer (k=4). In particular, for k=3, 4, we can adjust number 1 in step 1 and replace it with number 4 shown below.
[0255]
[0256] Option 2) The edge to insert may be determined by taking into account the transmission power of the constellation point to be inserted.
[0257] As step 4, the radius of each hexagon may be adjusted to obtain a hexagonal constellation suitable for different PA saturation points.
[0258] Operation 2-2-1-1) The normalized hexagonal constellation is calculated using the parameters {L, R k , k=1, . . . , L}.
[0259] The constellation generated by the learned AI in operation 2-1) may be normalized to design a 64-hexagonal constellation that is tolerant to the nonlinearity of the PA. The optimized result by the AI may be used to determine the number L of hexagons, and the radius or relative radius Rk, k = 1, ..., L. By normalizing the irregular constellation in operation 2-1) and converting it into a hexagonal constellation, performance and complexity can be optimized. The normalized constellation points obtained by the above operation reduce the complexity of demodulation.
[0260] Figure 32 shows example (3) of a constellation generated by a system for mitigating the effects of nonlinear PA according to an embodiment of the present invention. Figure 32 shows examples of a hexagonal constellation with N=64 and PaSat=0.45, a hexagonal constellation with N=64 and PaSat=0.58, and a hexagonal constellation with N=64 and PaSat=0.85. Table 19 shows the radius of each hexagonal constellation.
[0261]
[0262] The AM-AM parameters are assumed to be CMOS (g = 4.65, s = 0.81, PaSat around 0.58). Furthermore, by adjusting the radius or relative radius, a constellation for GaAs (g = 19, s = 0.81, PaSat around 1.4) can be obtained.
[0263] Operation 2-2-1-2) A Bit-to-symbol mapping table may be designed to define a hexagonal constellation that is tolerant to nonlinear PA.
[0264] The bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of adjacent constellation points on a hexagonal constellation is 1 bit.
[0265] Figure 33 shows an example (4) of a constellation generated by a system for mitigating the effects of nonlinear PA according to an embodiment of the present invention. Figure 33 shows an example of a constellation learned for N=64 and PaSat=0.58 normalized to a hexagonal constellation. As shown in Figure 33, the bit-to-symbol mapping table is generated counterclockwise. Table 20 shows an example of a bit-to-symbol mapping table for a normalized 64-constellation for PaSat=0.58.
[0266]
[0267] Operation 2-2-2) APSK design will now be described.
[0268] The normalized constellation based on APSK is designed based on the PA saturation point, and the parameters of APSK are adjusted according to the PA saturation point to improve the robustness.
[0269] Operation 2-2-2-1) The normalized APSK constellation is calculated using the parameters {L, R k , n k , φ k, k=1, ..., L}. Either option 1) or option 2) below may be performed.
[0270] Option 1) Based on the hexagonal constellation according to operation 2-2-1), an APSK that is resistant to nonlinear PA may be designed by normalizing the constellation points based on the inscribed circle and / or circumscribed circle of the hexagonal constellation. Two different APSKs may be designed based on whether the inscribed circle or circumscribed circle of the hexagon is used. For the original hexagonal constellation with L+1 rings, points may be inserted at the centers of the other rings to design an APSK with L rings.
[0271] Option 2) Based on the constellation generated by the learned AI in operation 2-1), the constellation points may be normalized to design a 64-APSK that is resistant to nonlinear PA. The result of the AI optimization is the parameters of the APSK that is resistant to nonlinear PA, the number of rings L, the number of points in each ring n k , k=1,...,L, radius or relative radius R k , k=1, . . . , L and the initial phase φ of each ring k , k=1,...,L. By normalizing the non-normal constellation according to operation 1-1) and converting it to APSK, performance and complexity can be optimized. By the above operation, the normalized constellation points reduce the demodulation complexity.
[0272] 34 is a diagram showing an example (5) of a constellation generated by a system for mitigating the influence of nonlinear PA according to an embodiment of the present invention. Fig. 34 shows examples of APSK constellations based on circumscribed circles for PaSat=0.45 and PaSat=0.85, and APSK constellations based on inscribed circles for PaSat=0.45 and PaSat=0.85. Table 21 shows the radius of each APSK constellation.
[0273]
[0274] Operation 2-2-2-2) A Bit-to-symbol mapping table may be designed to define an APSK constellation that is tolerant to nonlinear PA.
[0275] The bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of adjacent constellation points in a ring is 1 bit.
[0276] Figure 35 shows an example (6) of a constellation generated by a system for mitigating the effects of nonlinear PA according to an embodiment of the present invention. Figure 35 shows an example in which a constellation learned for N=64 and PaSat=0.58 is converted into a hexagonal constellation, and then normalized to an APSK constellation based on the circumscribed circle. As shown in Figure 35, the bit-to-symbol mapping table is generated counterclockwise. Table 22 shows an example of a bit-to-symbol mapping table for a normalized 64-constellation for PaSat=0.58.
[0277]
[0278] Below, we will explain operation 2-3) MCS table design to mitigate the impact of PA nonlinearity and related signaling for adaptive modulation.
[0279] To better suit application scenarios and requirements, an MCS table may be designed that includes constellations with high SE and tolerance to the nonlinearity of the PA in the region where the PA operates nonlinearly.
[0280] Hereinafter, operation 2-3-1) MCS table design including a constellation that is tolerant to the nonlinearity of PA will be described.
[0281] The constellation generated by the trained NN in operation 2-1) or the constellation normalized in operation 2-2) may be included in an MCS table that is tolerant to the nonlinearity of the PA. Note that, hereinafter, the constellation generated by the trained NN in operation 2-1) or the constellation normalized in operation 2-2) (hexagonal constellation and / or APSK constellation) will be referred to as "newPA" without loss of generality.
[0282] To avoid signaling overhead, constellations designed for several PA saturation points may be included in the MCS table. For example, the MCS table may be configured as shown in Option 1) to Option 3) below.
[0283] Also, for example, a constellation designed for a PA saturation point of 0.58 may be defined in the MCS table of option 1) below. Also, for example, constellations designed for PA saturation points of 0.58 and 0.85 may be defined in the MCS table of option 2) below. Also, for example, constellations designed for PA saturation points of 0.45, 0.58, and 0.85 may be defined in the MCS table of option 3) below.
[0284] Option 1) Some QAM MCSs in the 5-bit NR MCS table may be replaced with constellations that are resistant to nonlinear PA. Some QAM MCSs in the NR MCS table may be replaced with MCSs corresponding to the constellations generated by operation 2-1) or the constellations (hexagonal and / or APSK) normalized by operation 2-2). Replacements may be performed as shown in Examples 1) to 3) below.
[0285] Example 1) To reduce the number of MCSs replaced by new constellations, a constellation for one PA saturation point may be added to the MCS table. The MCS index associated with 16QAM may be replaced with "16newPA," a constellation for a 16-point MCS with a PA saturation point of 0.45, which is a modulation order 4 MCS. The MCS index associated with 64QAM may be replaced with "64newPA," a constellation for a 64-point MCS with a PA saturation point of 0.58, which is a modulation order 6 MCS. The MCS index associated with 256QAM may be replaced with "256newPA," a constellation for a 256-point MCS with a PA saturation point of 0.85, which is a modulation order 8 MCS. Table 23 is an example of an MCS table for Example 1).
[0286]
[0287] Example 2) Taking into account the PN resistance for different modulation orders, different numbers of PA saturation points associated with the constellation may be applied to different modulation orders. For example, for a 64-point constellation with a modulation order of 6, the MCS indexes associated with 64QAM may be partially replaced with the constellation "64newPA-mid" for a PA saturation point of 0.58 and the constellation "64newPA-high" for a PA saturation point of 0.45. On the other hand, for a 256-point constellation with a modulation order of 8, the MCS indexes associated with 256QAM may be partially replaced with the constellation "256newPA-low" for a PA saturation point of 0.85.
[0288] The lower the PaSat value, the greater the effect of nonlinearity. Therefore, the PA saturation point 0.85 is used as low, 0.58 as mid, and 0.45 as high.
[0289] Example 3) A partial index of BPSK, QPSK, or QAM in a conventional MCS table may be replaced with a newPN for which no PA saturation point is explicitly defined. That is, the PA saturation point may be obtained from explicit UE feedback instead of MCS notification. The constellation of newPA may be determined based on a combination of MCS notification and PA saturation point feedback. For example, if MCS 16newPA is signaled and a PA saturation point of 0.58 is fed back, 16newPA may be configured for a PA saturation point of 0.58 and used for transmission.
[0290] Option 1) above provides an MCS table that includes MCSs that are tolerant to non-linear PA, and the total number of MCSs and MCS signaling overhead remains unchanged.
[0291] Option 2) A new MCS that is resistant to nonlinear PA and that corresponds to the constellation generated by operation 2-1) or the APSK normalized by operation 2-2) may be added to the 5-bit NR MCS table to generate a new MCS table that exceeds 5 bits. The MCS that corresponds to the constellation generated by operation 2-1) or the APSK normalized by operation 2-2) may be added to the NR MCS table. The addition may be as shown in Examples 1) to 3) below.
[0292] Example 1) To increase the number of MCSs that can be replaced by new constellations, constellations for PA saturation points of 2 or 3 may be added to the MCS table. A 16-point MCS with modulation order 4, "16newPA-low," for a PA saturation point of 0.85, and a constellation "16newPA-high" for a PA saturation point of 0.45 may also be added.
[0293] Alternatively, a constellation "16newPA-mid" for a 16-point PA saturation point of 0.58, which is an MCS with a modulation order of 4, or a constellation "16newPA-high" for a PA saturation point of 0.45 may be added. Alternatively, a constellation "16newPA-low" for a 16-point PA saturation point of 0.85, which is an MCS with a modulation order of 4, or a constellation "16newPA-mid" for a PA saturation point of 0.58, or a constellation "16newPA-high" for a PA saturation point of 0.45 may be added.
[0294] It is also possible to add a constellation "64newPA-low" for a 64-point PA saturation point of 0.85, which is an MCS with a modulation order of 6, and a constellation "64newPA-high" for a PA saturation point of 0.45. Alternatively, it is also possible to add a constellation "64newPA-mid" for a 64-point PA saturation point of 0.58, which is an MCS with a modulation order of 6, and a constellation "64newPA-high" for a PA saturation point of 0.45. Alternatively, it is also possible to add a constellation "64newPA-low" for a 64-point PA saturation point of 0.85, which is an MCS with a modulation order of 6, a constellation "64newPA-mid" for a PA saturation point of 0.58, and a constellation "64newPA-high" for a PA saturation point of 0.45.
[0295] It is also possible to add a constellation "256newPA-low" for a PA saturation point of 0.85, which is an MCS with a modulation order of 8, and a constellation "256newPA-high" for a PA saturation point of 0.45. Alternatively, it is also possible to add a constellation "256newPA-mid" for a PA saturation point of 0.58, which is an MCS with a modulation order of 8, and a constellation "256newPA-high" for a PA saturation point of 0.45. Alternatively, it is also possible to add a constellation "256newPA-low" for a PA saturation point of 0.85, which is an MCS with a modulation order of 8, a constellation "256newPA-mid" for a PA saturation point of 0.58, and a constellation "256newPA-high" for a PA saturation point of 0.45.
[0296] Tables 24 and 25 are examples of MCS tables for Example 1).
[0297]
[0298]
[0299] Example 2) High modulation orders and / or high coding rates are susceptible to phase noise. Different PA saturation points may be applied to different modulation orders and coding rates. For example, "16newPA-high" may be added for a PA saturation point of 0.45 in a 16-point constellation. For example, "64newPA-low" may be added for a PA saturation point of 0.85 in a 64-point constellation, "64newPA-mid" may be added for a PA saturation point of 0.58, and "64newPA-high" may be added for a PA saturation point of 0.45. For example, "256newPA-low" may be added for a PA saturation point of 0.85 in a 256-point constellation.
[0300] Example 3) A partial index of BPSK, QPSK, or QAM in a conventional MCS table may be replaced with a newPN for which the PN dispersion is not explicitly defined. That is, the PN dispersion may be obtained from explicit UE feedback instead of MCS notification. The constellation of the newPN may be determined based on a combination of the MCS notification and PN dispersion feedback. For example, if MCS16newPN is signaled and PN dispersion 0.02 is fed back, 16newPN may be configured for PN dispersion 0.02 and used for transmission.
[0301] Option 2) above allows the addition of a PN-resistant MCS while maintaining the traditional MCS.
[0302] Option 3) A new MCS table may be configured that includes only constellations that are resistant to nonlinear PA. The MCS table may be configured based on existing constellations that are resistant to nonlinear PA, such as BPSK, pi-2BPSK, pi-4BPSK, and pi-4QPSK, as well as newly designed constellations. The MCS table may include MCS indexes that include 16newPA, MCS indexes that include 64newPA, and MCS indexes that include 256newPA. Table 26 is an example of an MCS table for option 3).
[0303]
[0304] The above option 3) can provide an MCS table including an MCS that is resistant to non-linear PA. Multiple sets of MCS tables can be configured, allowing for flexible selection of different MCS tables and adjusting the cost required for MCS indication.
[0305] Below, operation 2-3-2) related signaling suitable for using multiple MCS tables with tolerance to non-linear PA will be described.
[0306] Signaling actions may be performed as shown below in options 1)-3).
[0307] Option 1) When the MCS table is completely replaced with a newly designed MCS table (e.g., Option 1) or Option 2) in Operation 1-3-1), the conventional MCS notification method and field (e.g., I in NR) are replaced. MCS ) or signaling related to MCS notification in future systems may be reused.
[0308] Option 2) If the NN parameters are pre-stored in the transmitter and receiver, a combination of the MCS and PA saturation point may be used to generate the constellation. The PA saturation point may be fed back via CSI signaling or new signaling at the physical layer or higher layers (e.g., RRC, MAC-CE, UCI, etc.).
[0309] Option 3) When a newly designed MCS table (e.g., Option 1), Option 2) or Option 3) of Operation 1-3-1) is added to the specification, the following notification method may be adopted.
[0310] Option 3-1) New Signaling I flag_PA may signal whether to use an MCS table containing newPA constellations. flag_PA may be transmitted via RRC, MAC-CE, UCI or DCI signaling.
[0311] Alt. 1 (DL): The UE may feed back the MCS index for DL transmission to the BS. The UE may also use non-linear PA-tolerant MCS indication (I flag_PA ) and MCS index (I MCS ) may be used or not, and may be transmitted to the BS as additional feedback. If a non-linear PA-tolerant MCS notification is required, I flag_PA is set to 1, otherwise I flag_PA may be set to 0. flag_PA may be transmitted via RRC, MAC-CE or UCI signaling. MCS The feedback may be sent via existing signaling.
[0312] The BS determines the required MCS table and I based on the UE feedback. MCS First, I flag_PA If Q is 1, the MCS table including newPA may be selected, otherwise the conventional MCS table may be selected. m , the target coding rate R is I MCS and may be determined based on the MCS table.
[0313] Alt. 2 (DL): The BS may actively select the MCS for DL transmission. New RRC, MAC-CE, DCI, or UCI signaling may be designed to feed back the PA saturation point from the UE. The BS selects whether to use an MCS table that is tolerant to nonlinear PA based on the PA saturation point feedback from the UE. If the PA saturation point is close to 1, the BS uses the conventional MCS table, and if the PA saturation point is close to 1, the BS uses the nonlinear PA-tolerant MCS table. flag_PA may be set to 0. If the PA saturation point is less than or equal to 0.85, select the MCS table containing the newPA constellation, and flag_PA may be set to 1. The BS may flag_PA and I MCS may be sent to the UE via RRC, MAC-CE or DCI signaling.
[0314] Alt. 1 (UL): The BS may signal the UL MCS to the UE. The BS may signal the nonlinear PA-tolerant MCS (I flag_PA ) and MCS index (I MCS If non-linear PA tolerant MCS notification is required, I flag_PA is set to 1, otherwise I flag_PA may be set to 0. flag_PA may be transmitted via RRC, MAC-CE or UCI signaling. MCS may be transmitted via existing signaling.
[0315] The UE acquires the required MCS table and I based on the BS instruction. MCS First, I flag_PA If Q is 1, the MCS table including newPA may be selected, otherwise the conventional MCS table may be selected. m , the target coding rate R is I MCS and may be determined based on the MCS table.
[0316] Alt. 2 (UL): The UE may actively select an MCS for UL transmission. The UE selects whether to use a nonlinear PA-tolerant MCS table based on the estimated PA saturation point. If the PA saturation point is near 1, the UE uses a conventional MCS table; if the PA saturation point is near 1, the UE uses a nonlinear PA-tolerant .... flag_PA may be set to 0. If the PA saturation point is less than or equal to 0.85, select the MCS table containing the newPA constellation, and flag_PA The UE may set I flag_PA and I MCS may be sent to the BS via RRC, MAC-CE or UCI signaling.
[0317] Option 3-2) The existing MCS selection method and signaling definition method in 3GPP TS 36544 or future systems may be reused. New signaling may be added to indicate whether to use an MCS table including constellations that are resistant to non-linear PA. This signaling may be transmitted via higher or lower layer signaling, such as RRC, MAC-CE, UCI, or DCI signaling.
[0318] Alt. 1: This signaling may be explicitly associated with nonlinear PA. For example, in the case of DL transmission, the process in section 5.1.3.1 of 3GPP TS 36.106.221 may be reused. If mcs-Table DCI-1-2 is set to nonlinearPA or PA, the UE may use an MCS table containing the N-newPA constellation. MCS and the MCS table including the N-newPA constellation, m and may be used to determine the target coding rate R.
[0319] For example, for UL transmission, the higher layer parameters ifnonlinearPA or ifPA may be added. If ifnonlinearPA or ifPA is 1, an MCS table containing the N-newPA constellation may be used for PUSCH.
[0320] Alt. 2: The signaling may be implicitly associated with the PN. For example, in a system configured with sub-THz bands, an MCS table including a newly designed constellation that is resistant to nonlinear PA may be used. In other systems, an MCS table that is independent of nonlinear PA may be used.
[0321] Here, conventional reference signals and estimation methods may be considered to estimate the PA saturation point and / or CSI-related information (e.g., SNR, SINR, CQI, CSI, etc.) In addition, AI may be used to estimate the PN variance and / or CSI-related information for selecting the MCS.
[0322] Below, we will explain operation 3) constellation design and adaptive modulation that reduces the effects of nonlinearities of PN and PA.
[0323] Below, operation 3-1) irregular constellation design to mitigate the effects of nonlinearities of PN and PA will be described.
[0324] Fig. 36 is a diagram for explaining an example (3) of a system according to an embodiment of the present invention. Like the system that performs adaptive modulation based on the input of PN variance and PA saturation point shown in Fig. 36, a non-normal constellation generated by a neural network that learns end-to-end (hereinafter also referred to as "E2E") may be used in a system that reduces the influence of PN and PA operating in a nonlinear region. The PN variance and PA saturation point are defined as one of the input parameters of the neural network.
[0325] The basic principle is as follows: the NN performs offline learning under limited channel conditions, such as a predetermined SNR range and some PN variances and PA saturation points. The NN is then generalized based on more channel conditions (continuous or finer PN variances and PA saturation points) to obtain a set of modulation constellations that mitigate the effects of PN.
[0326] For example, offline learning can be performed as follows: 2Let N be the number of PN variances and PA saturation points, and construct a NN architecture. Train the NN to learn the constellations under the conditions of L PN variances and PA saturation points.
[0327] PN variance σ 2 PN , the PA saturation point PaSat, and the N one-hot encoder vector are used as inputs to perform E2E learning. j , j=1,...,N is a vector of length N where the jth value is 1 and the other values are 0. For example, e j may be associated with the j-th constellation point.
[0328] The L sets {PN variance, PA saturation point} are used for training so that the network has the ability to generalize to acquire constellations that vary with PN. Note that PN may be considered at both the transmitter and receiver sides. PA saturation point may also be considered at the transmitter side.
[0329] The above operation can improve performance in environments where multiple PN variances and PA saturation points exist.
[0330] Fig. 37 shows an example (9) of a constellation generated by a system according to an embodiment of the present invention. Fig. 37 shows an example of a constellation generated by operation 3-1) trained with a PN variance of 0.02 and PaSat of 0.58, and an example of a constellation trained with a PN variance of 0.1 and PaSat of 0.58.
[0331] Operation 3-1-1) Definition of NN parameters will be explained below.
[0332] FIG. 38 is a diagram illustrating an example of a system for mitigating the effects of phase noise and nonlinear PA according to an embodiment of the present invention. The offline trained neural network parameters may include weights, biases, and activation functions. These parameters are defined and used in the transmitter and receiver. The pre-stored neural network generates constellation points based on the modulation order and PN variance. As shown in FIG. 22, bit-to-index mapping is added before the neural network module converts the modulated bit sequence into the neural network one-hot encoder vector.
[0333] Let module 1 be an offline trained NN. Module 1 generates constellation points for a given modulation order, PN variance and PA saturation point.
[0334] Module 2 is designated as bit-to-index mapping. Module 2 implements the function of mapping bits to symbols for online deployment. During the offline training phase, input parameters include the none-hot encoder vector, the PN variance of the channel, and the PA saturation point. During the online deployment phase of the communication system, the bit sequence may be mapped to the corresponding constellation points that were deviated during the offline training phase. The bit-to-index mapping is added before the predefined NN to generate the corresponding one-hot encoder vector (i.e., the input of the NN).
[0335] By combining bit-to-index mapping and NN, the bit sequence is modulated into constellation points that are resistant to PN and nonlinear PA.
[0336] By generalizing the NN learned offline with coarse PN variance and PA saturation point, a constellation that is tolerant to any PN variance and PA saturation point can be dynamically generated based on the dynamic PN variance and PA saturation point in a real system.
[0337] The system shown in FIG. 38 uses modulation order log 2 N, PN variance σ 2 PN , corresponding to the PA saturation point PaSat, and the length log 2 Bit-to-index mapping of N and offline trained NN(N, σ 2 PN and PaSat) for bit-to-symbol mapping and variance σ 2 PN A constellation that is resistant to PN and nonlinear PA is generated.
[0338] As shown in FIG. 38, on the transmitter side, a bit sequence is input to bit-to-index mapping, and a one-hot encoder vector is output. The one-hot encoder vector is input to an offline trained neural network, and a modulation symbol is output. The modulation symbol becomes a transmission symbol starting from PA, which is the saturation point PaSat. The symbol received on the receiver side from the transmitter side via the channel is input to the offline trained neural network, and a one-hot encoder vector is output. The one-hot encoder vector is input to bit-to-index demapping, and a bit sequence is output.
[0339] The parameters of the NN to be trained offline may be defined by at least {the number of layers of the NN, the weight of each layer, the bias of each layer, and the activation function of each layer}.
[0340] Table 3 is an example of a table for parameters of an offline training NN for all PN distributions.
[0341] A neural network mainly has three basic elements: weights, biases, and activation functions. Weights are the strength of connections between neurons, which are represented by weights. The weight values indicate the probability or proportion of each input element. Bias is a setting to accurately classify samples. Bias is an important parameter for the model and is equivalent to adding a constant. Activation functions act as nonlinear mappings, generally restricting the output values from neurons to a predetermined range, such as -1 to 1 or 0 to 1.
[0342] The weight and bias values are results obtained from the trained network and may be stored, for example, in a text file format.
[0343] The common activation function may be, for example, a sigmoid, a hyperbolic tangent (tanh), a rectified linear unit (ReLU), or the like.
[0344] A normalization layer is essential in the encoder to normalize the power of the constellations produced by the system.
[0345] For a given modulation order, the input parameters of the NN include at least one of 1) and 2) below.
[0346] 1) One-hot encoder vector e j , j=1,...,N is a vector of length N, where the j-th value is 1 and the others are 0. 2) PN variance 3) PA saturation point
[0347] The bit-to-index mapping is explained below. 2 The N bit sequences are transformed into one-hot encoder vectors of length N for input to the NN. Option 1)-Option 4) below may be adopted.
[0348] Option 1) A binary number, which is a bit sequence, may be converted into a decimal number, which is an index of a constellation point in the neural network. This design is suitable for constellations generated online for any PN distribution and / or PA saturation point. For example, the PN distribution and / or PA saturation point may not be taken into account during offline training.
[0349] Option 2) Bit-to-index mapping, which satisfies gray-mapping, may be performed for the constellation points in the neural network. Gray-mapping is expected to improve performance.
[0350] Option 3) An optimized mapping may be performed to minimize the error rate or maximize the throughput / SE, which can improve the decoding performance.
[0351] Option 4) The bit-to-index mapping module may be separated or joined by a neural network or other AI network to learn about the constellation points.
[0352] The transformed index j is used to generate the one-hot encoder vector e j is generated and input to the NN.
[0353] Below, operation 3-1-2) B2S mapping table design for irregular constellations will be described.
[0354] A bit-to-symbol mapping table may be designed to define irregular constellations that are resistant to PN and nonlinear PA. Similar to the conventional definition, a bit-to-symbol mapping table may be defined to modulate a bit sequence onto a constellation point. This allows the constellation point and bit-to-symbol mapping to be combined and defined in one table.
[0355] The bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of one ring of the same amplitude or adjacent constellation points of similar amplitude is 1 bit.
[0356] Figure 39 is a diagram showing an example (1) of a constellation generated by a system for mitigating the effects of phase noise and nonlinear PA according to an embodiment of the present invention. Figure 39 is an example of a trained 64-bit constellation for a PN variance of 0.02 and a PaSat of 0.58. Bit-to-symbol mapping is performed in the order of the clockwise arrows shown in Figure 39. Table 27 is an example of a bit-to-symbol mapping table for a trained 64-bit constellation for a PN variance of 0.02 and a PaSat of 0.58.
[0357]
[0358] Figure 39 is an example of a trained 64-bit constellation for PN variance 0.1 and PaSat 0.58. Bit-to-symbol mapping is performed in the clockwise direction indicated by the arrows in Figure 39. Table 28 is an example of a bit-to-symbol mapping table for a trained 64-bit constellation for PN variance 0.1 and PaSat 0.58.
[0359]
[0360] Below, we will explain operation 3-2) normalized constellation design that reduces the effects of nonlinearities of PN and PA.
[0361] The normalized constellation based on APSK is designed based on the PN dispersion and PA saturation point, and the parameters of APSK are adjusted according to the PN dispersion and / or PA saturation point to improve the robustness.
[0362] Operation 3-2-1) Definition of APSK will be explained below.
[0363] The normalized APSK constellations are given parameters {L, R k , n k , φ k, k = 1, ..., L}. Based on the constellation generated by the learned AI in the above operation 1-1), the constellation points may be normalized to design a 64APSK that is resistant to PN and nonlinear PA. The result of optimization by the AI is the parameters of the APSK that is resistant to PN and nonlinear PA, the number of rings L, the number of points in each ring n k , k=1,...,L, radius or relative radius R k , k=1, . . . , L and the initial phase φ of each ring k , k=1,...,L. By normalizing the non-normal constellation according to operation 3-1) and converting it to APSK, performance and complexity can be optimized. By the above operation, the normalized constellation points reduce the demodulation complexity.
[0364] 40 is a diagram showing an example (2) of a constellation generated by a system for mitigating the effects of phase noise and nonlinear PA according to an embodiment of the present invention. Fig. 40 shows an example of 64APSK for PN dispersion 0.02 and PaSat 0.58, in which the learned constellation obtained by operation 3-1) is normalized. Table 29 shows the radius and initial phase of each ring of 64APSK for PN dispersion 0.02 and PaSat 0.58 in Fig. 40.
[0365]
[0366] As shown in FIG. 40 and Table 29, the number of rings is 10, the number of points in each ring is {4, 4, 7, 7, 7, 7, 7, 7, 7, 7}, and the minimum phase from the original points may be set as the initial phase of the ring.
[0367] 40 is an example of 64APSK for PN dispersion 0.1 and PaSat 0.58, in which the learned constellation obtained by operation 3-1) is normalized. Table 30 shows the radius and initial phase of each ring of 64APSK for PN dispersion 0.1 and PaSat 0.58 in FIG.
[0368]
[0369] As shown in FIG. 40 and Table 30, the number of rings is 16, the number of points in each ring is 4, and the minimum phase from the original 4 points may be set as the initial phase of the ring.
[0370] Operation 3-2-2) B2S mapping table design for APSK will be described below.
[0371] A bit-to-symbol mapping table may be designed to define a normalized constellation that is tolerant to PN and nonlinear PA. Either option 1) or option 2) shown below may be adopted.
[0372] Option 1) A bit-to-symbol mapping table may be generated based on the branches of Gray mapping, where the difference between bits of adjacent constellation points in each branch is only one bit.
[0373] Option 2) A Bit-to-symbol mapping table may be generated based on Gray mapping from the center outward in a clockwise or counterclockwise direction, where the difference between bits of one ring of the same amplitude or adjacent constellation points of similar amplitude is 1 bit.
[0374] Figure 41 shows an example (3) of a constellation generated by a system for mitigating the effects of phase noise and nonlinear PA according to an embodiment of the present invention. Figure 41 shows an example of option 1) of a 64-bit constellation for normalized PN dispersion 0.02 and PaSat 0.58, and an example of option 2) of a 64-bit constellation for normalized PN dispersion 0.1 and PaSat 0.58. Bit-to-symbol mapping is performed in the clockwise order shown in Figure 41. Table 31 shows an example of a bit-to-symbol mapping table for a 64-bit constellation for normalized PN dispersion 0.02 and PaSat 0.58.
[0375]
[0376] Table 32 shows an example of a Bit-to-symbol mapping table for 64 constellations for normalized PN dispersion 0.1 and PaSat 0.58.
[0377]
[0378] Below, we will explain operation 3-3) MCS table design and related signaling for adaptive modulation that reduces the effects of PN and PA nonlinearities.
[0379] To better suit application scenarios and requirements, an MCS table may be designed that includes constellations with high SE and resistance to PN and nonlinear PA.
[0380] Hereinafter, operation 3-3-1) MCS table design including constellations resistant to PN and nonlinear PA will be described.
[0381] The constellation generated by the trained NN in operation 3-1) or the APSK normalized in operation 3-2) may be included in the PN-resistant MCS table. Note that, hereinafter, the constellation generated by the trained NN in operation 3-1) or the APSK normalized in operation 3-2) will be referred to as "newPNPA" without loss of generality.
[0382] To avoid signaling overhead, constellations designed for several PN distributions and PA saturation points may be included in the MCS table. For example, the MCS table may be configured as shown in Option 1) to Option 3) below.
[0383] Also, for example, a constellation designed for {PN variance, PA saturation point} of {0.04, 0.58} may be defined in the MCS table of option 1) below. Also, for example, a constellation designed for {PN variance, PA saturation point} of {0.04, 0.58} and {0.08, 0.45} may be defined in the MCS table of option 2) below. Also, for example, a constellation designed for {PN variance, PA saturation point} of {0.02, 0.85}, {0.04, 0.58}, and {0.08, 0.45} may be defined in the MCS table of option 3) below.
[0384] Option 1) Some QAM MCSs in the 5-bit NR MCS table may be replaced with constellations that are resistant to PN and nonlinear PA. Some QAM MCSs in the NR MCS table may be replaced with MCSs corresponding to the constellations generated by operation 1-1) or the APSK normalized by operation 1-2). Replacements may be performed as shown in Examples 1) to 3) below.
[0385] Example 1) To reduce the number of MCSs replaced by new constellations, a constellation for one PN distribution may be added to the MCS table. A 16-point {PN distribution, PA saturation point} MCS for modulation order 4 may replace the MCS index associated with 16QAM with the constellation "16newPNPA" for {0.08, 0.45}. A 64-point {PN distribution, PA saturation point} MCS for modulation order 6 may replace the MCS index associated with 64QAM with the constellation "64newPNPA" for {0.04, 0.58}. A 256-point {PN distribution, PA saturation point} MCS for modulation order 8 may replace the MCS index associated with 256QAM with the constellation "256newPNPA" for {0.02, 0.85}. Table 33 is an example of an MCS table for Example 1).
[0386]
[0387] Example 2) Considering the resistance to PN and nonlinear PA for different modulation orders, different PN variances and PA saturation points for the constellations may be applied to different modulation orders. For example, for a 64-point constellation with a modulation order of 6, the constellation "64newPNPA-mid" for {PN variance, PA saturation point} = {0.04, 0.58} and the constellation "64newPNPA-high" for {PN variance, PA saturation point} = {0.08, 0.45} may partially replace the MCS index associated with 64QAM. On the other hand, for a 256-point constellation with a modulation order of 8, the constellation "256newPNPA-low" for {PN variance, PA saturation point} = {0.02, 0.85} may partially replace the MCS index associated with 256QAM.
[0388] Example 3) A partial index of BPSK, QPSK, or QAM in the conventional MCS table may be replaced with newPNPA, in which the PN variance and PA saturation point are not explicitly defined. That is, the PN variance and PA saturation point may be obtained from explicit UE feedback instead of MCS notification. The constellation of newPNPA may be determined based on a combination of MCS notification and PN variance and PA saturation point feedback.
[0389] Option 1) above provides an MCS table containing PN-resistant MCSs, and the total number of MCSs and MCS signaling overhead remains unchanged.
[0390] Option 2) A new MCS table of more than 5 bits may be generated by adding a new MCS corresponding to the constellation generated by operation 3-1) that is resistant to PN and nonlinear PA or the APSK normalized by operation 3-2) to the 5-bit NR MCS table. An MCS corresponding to the constellation generated by operation 3-1) or the APSK normalized by operation 3-2) may be added to the NR MCS table. Addition may be performed as shown in Examples 1) to 3) below.
[0391] Example 1) To increase the number of MCSs that can be replaced by new constellations, constellations for PN variances and PA saturation points of 2 or 3 may be added to the MCS table. For a 16-point MCS with modulation order 4, the constellation "16newPNPA-low" may be added for {PN variance, PA saturation point} of {0.02, 0.85}, and the constellation "16newPNPA-high" may be added for {PN variance, PA saturation point} of {0.08, 0.45}.
[0392] Alternatively, a constellation "16newPNPA-mid" for a 16-point MCS with a modulation order of 4 and a {PN variance, PA saturation point} of {0.04, 0.58} and a constellation "16newPNPA-high" for a {PN variance, PA saturation point} of {0.08, 0.45} may be added. Alternatively, a constellation "16newPNPA-low" for a 16-point MCS with a modulation order of 4 and a {PN variance, PA saturation point} of {0.02, 0.85}, a constellation "16newPNPA-mid" for a {PN variance, PA saturation point} of {0.04, 0.58}, and a constellation "16newPNPA-high" for a {PN variance, PA saturation point} of {0.08, 0.45} may be added.
[0393] It is also possible to add a constellation "64newPNPA-low" for a 64-point MCS with a {PN variance, PA saturation point} of {0.02, 0.85} and a constellation "64newPNPA-high" for a {PN variance, PA saturation point} of {0.08, 0.45} for modulation order 6. Alternatively, it is also possible to add a constellation "64newPNPA-mid" for a 64-point MCS with a {PN variance, PA saturation point} of {0.04, 0.58} and a constellation "64newPNPA-high" for a {PN variance, PA saturation point} of {0.08, 0.45} for modulation order 6. Alternatively, the following may be added: a constellation "64newPNPA-low" for a 64-point MCS with modulation order 6, {PN variance, PA saturation point} of {0.02, 0.85}, a constellation "64newPNPA-mid" for {PN variance, PA saturation point} of {0.04, 0.58}, and a constellation "64newPNPA-high" for {PN variance, PA saturation point} of {0.08, 0.45}.
[0394] It is also possible to add a constellation "256newPNPA-low" for a 256-point MCS with a PN variance {PN variance, PA saturation point} of {0.02, 0.85} and a constellation "256newPNPA-high" for a 256-point MCS with a modulation order of 8, where {PN variance, PA saturation point} is {0.08, 0.45}. Alternatively, it is also possible to add a constellation "256newPNPA-mid" for a 256-point MCS with a modulation order of 8, where {PN variance, PA saturation point} is {0.04, 0.58} and a constellation "256newPNPA-high" for a 256-point MCS with a modulation order of 8, where {PN variance, PA saturation point} is {0.08, 0.45}. Alternatively, the following may be added: a constellation "256newPNPA-low" for a 256-point MCS with modulation order 8, {PN variance, PA saturation point} of {0.02, 0.85}, a constellation "256newPNPA-mid" for a {PN variance, PA saturation point} of {0.04, 0.58}, and a constellation "256newPNPA-high" for a {PN variance, PA saturation point} of {0.08, 0.45}.
[0395] Tables 34 and 35 are examples of MCS tables for Example 1).
[0396]
[0397]
[0398] Example 2) High modulation orders and / or high code rates are susceptible to phase noise and nonlinear PA. Different PN variances and PA saturation points may be applied to different modulation orders and code rates. For example, "16newPNPA-high" may be added for a 16-point constellation with {PN variance, PA saturation point} of {0.08, 0.45}. For example, "64newPNPA-low" may be added for a 64-point constellation with {PN variance, PA saturation point} of {0.02, 0.85}, "64newPNPA-mid" may be added for a 64-point constellation with {PN variance, PA saturation point} of {0.04, 0.58}, and "64newPNPA-high" may be added for a 64-point constellation with {PN variance, PA saturation point} of {0.08, 0.45}. For example, "256newPNPa-low" may be added for a 256-point constellation where {PN variance, PA saturation point} is {0.02, 0.85}.
[0399] Example 3) A partial index of BPSK, QPSK, or QAM in the conventional MCS table may be replaced with newPNPA, in which the PN variance and PA saturation point are not explicitly defined. That is, the PN variance and PA saturation point may be obtained from explicit UE feedback instead of MCS notification. The constellation of newPNPA may be determined based on a combination of MCS notification and PN variance and PA saturation point feedback.
[0400] Option 2) above allows the addition of an MCS that is resistant to PN and non-linear PA while maintaining the traditional MCS.
[0401] Option 3) A new MCS table may be configured that includes only constellations that are resistant to PN and nonlinear PA. The MCS table may be configured based on existing constellations that are resistant to PN and nonlinear PA, such as BPSK, pi-2BPSK, pi-4BPSK, pi-4QPSK, and newly designed constellations. The MCS table may include MCS indexes including 16newPNPA, MCS indexes including 64newPNPA, and MCS indexes including 256newPNPA. Table 36 is an example of an MCS table for option 3).
[0402]
[0403] Option 3) above allows for the provision of an MCS table containing a PN-resistant MCS. Multiple sets of MCS tables can be configured, allowing for flexible selection of different MCS tables and adjusting the cost required for MCS indication.
[0404] Below, operation 3-3-2) related signaling suitable for using multiple MCS tables with PN tolerance will be described.
[0405] Signaling actions may be performed as shown in options 1)-4) below.
[0406] Option 1) When the MCS table is completely replaced with a newly designed MCS table (e.g., Option 1) or Option 2) in Operation 3-3-1), the conventional MCS notification method and field (e.g., I in NR) are used. MCS ) or signaling related to MCS notification in future systems may be reused.
[0407] Option 2) If the NN parameters are pre-stored in the transmitter and receiver, a combination of the MCS, PN variance and PA saturation point may be used to generate the constellation. The PN variance and / or PA saturation point may be fed back via CSI signaling or new signaling at the physical layer or higher layers (e.g. RRC, MAC-CE, UCI, etc.).
[0408] Option 3) When a newly designed MCS table (e.g., Option 1), Option 2) or Option 3) of Operation 3-3-1) is added to the specification, the following notification method may be adopted.
[0409] Option 3-1) New Signaling I flag_PNPA may signal whether to use an MCS table containing the constellation of newPNPA. flag_PNPA may be transmitted via RRC, MAC-CE, UCI or DCI signaling.
[0410] Alt. 1 (DL): The UE may feed back the MCS index for DL transmission to the BS. The UE may also provide PN and non-linear PA tolerant MCS indication (I flag_PNPA ) and MCS index (I MCS ) may be used or not, and may be transmitted to the BS as additional feedback. If PN and non-linear PA tolerant MCS notification is required, I flag_PNPA is set to 1, otherwise I flag_PNPA may be set to 0. flag_PNPA may be transmitted via RRC, MAC-CE or UCI signaling. MCS The feedback may be sent via existing signaling.
[0411] The BS determines the required MCS table and I based on the UE feedback. MCS First, I flag_PNPA If Q is 1, the MCS table including newPNPA may be selected, otherwise the conventional MCS table may be selected. m , the target coding rate R is I MCSand may be determined based on the MCS table.
[0412] Alt. 2 (DL): The BS may actively select the MCS for DL transmission. New RRC, MAC-CE, DCI, or UCI signaling may be designed to feed back the PN variance and / or PA saturation point from the UE. The BS selects whether to use a PN and nonlinear PA-tolerant MCS table based on the feedback from the UE. If the PN variance is near 0 and / or PaSat is near 1, the BS uses the conventional MCS table and the I flag_PNPA may be set to 0. Otherwise, select the MCS table containing the newPNPA constellation and set I flag_PNPA may be set to 1. The BS may flag_PNPA and I MCS may be sent to the UE via RRC, MAC-CE or DCI signaling.
[0413] Alt. 1 (UL): The BS may signal the UL MCS to the UE. The BS may signal the PN and non-linear PA tolerant MCS (I flag_PNPA ) and MCS index (I MCS If PN and non-linear PA tolerant MCS notification is required, I flag_PNPA is set to 1, otherwise I flag_PNPA may be set to 0. flag_PNPA may be transmitted via RRC, MAC-CE or UCI signaling. MCS may be transmitted via existing signaling.
[0414] The UE acquires the required MCS table and I based on the BS instruction. MCS First, I flag_PNPA If Q is 1, the MCS table including newPNPA may be selected, otherwise the conventional MCS table may be selected. m , the target coding rate R is I MCS and may be determined based on the MCS table.
[0415] Alt. 2 (UL): The UE may actively select an MCS for UL transmission. The UE selects whether to use a PN- and nonlinear PA-tolerant MCS table based on the estimated PN variance and / or PA saturation point. If the PN variance is near 0 and / or the PA saturation point is near 1, the UE uses a conventional MCS table and the I flag_PNPA may be set to 0. Otherwise, select the MCS table containing the newPNPA constellation and set I flag_PNPA The UE may set I flag_PNPA and I MCS may be sent to the BS via RRC, MAC-CE or UCI signaling.
[0416] Option 3-2) The existing MCS selection method and signaling definition method in Non-Patent Document 16 or future systems may be reused. New signaling may be added to indicate whether to use an MCS table including PN and non-linear PA-resistant constellations. This signaling may be transmitted via higher or lower layer signaling, such as RRC, MAC-CE, UCI, or DCI signaling.
[0417] Alt. 1: The signaling may be explicitly associated with PN and non-linear PA. For example, in the case of DL transmission, the process in section 5.1.3.1 of 3GPP TS 36.106.221 may be reused. If mcs-Table DCI-1-2 is set to PNPA, the UE may use an MCS table containing the N-newPNPA constellation. MCS and an MCS table including N-newPNPA constellations, m and may be used to determine the target coding rate R.
[0418] For example, for UL transmission, a higher layer parameter ifPNPA may be added: If ifPNPA is 1, then an MCS table containing the N-newPNPA constellation may be used for PUSCH.
[0419] Alt. 2: The signaling may be implicitly associated with PN and / or nonlinear PA. For example, in a system configured with sub-THz bands, a newly designed MCS table including a constellation that is resistant to PN and nonlinear PA may be used. In other systems, an MCS table that is independent of PN and nonlinear PA may be used.
[0420] Option 4) Newly designed MCS tables for PN, nonlinear PA, and joint use of PN and nonlinear PA (e.g., MCS tables in operation 1-3-1), operation 2-3-1), and operation 3-3-1) may be added to the specification, and signaling to notify which MCS table to use may also be added to the specification.
[0421] For example, if the PN is large and the nonlinearity of the PA is negligible, the MCS table in operation 1-3-1) may be used.
[0422] For example, if the PN is negligible and the PA has strong nonlinearity, the MCS table in operation 2-3-1) may be used.
[0423] For example, if both the PN and PA nonlinearities cannot be ignored, the MCS table in operation 3-3-1) may be used.
[0424] Here, conventional reference signals and estimation methods may be considered to estimate PN variance, PA saturation point, PA model, and / or CSI-related information (e.g., SNR, SINR, CQI, CSI, etc.) In addition, AI may be used to estimate PN variance, PA saturation point, PA model, and / or CSI-related information for selecting an MCS.
[0425] Below, operation 4) UE capability definition will be explained.
[0426] The following UE capabilities 1)-4) may be defined:
[0427] 1) UE capability indicating whether the UE supports APSK, 2) UE capability indicating whether the UE supports hexagonal constellations, 3) UE capability indicating whether the UE supports using NN for modulation and demodulation, and 4) UE capability indicating whether the UE supports estimation of PN variance and / or PA saturation point.
[0428] UE capabilities may be reported from the UE to the BS via UE capability signaling, which may be, for example, RRC, MAC-CE, or UCI signaling.
[0429] According to the above-described embodiment, the terminal 20 can communicate effectively even when there is the influence of phase noise and / or nonlinear PA by including the constellation generated by the trained neural network in the MCS table and using it. Furthermore, by using a normalized constellation generated by the trained neural network, performance and complexity can be optimized. Furthermore, the normalized constellation can reduce the complexity of demodulation.
[0430] That is, in a wireless communication system, it is possible to reduce the degradation of performance caused by the PA (Power Amplifier) saturation point.
[0431] (Device Configuration) Next, a functional configuration example of the base station 10 and the terminal 20 that execute the processes and operations described above will be described. The base station 10 and the terminal 20 include functions for implementing the above-described embodiments. However, the base station 10 and the terminal 20 may each include only a part of the functions of the embodiments.
[0432] <Base Station 10> Figure 42 is a diagram showing an example of the functional configuration of the base station 10 in an embodiment of the present invention. As shown in Figure 42, the base station 10 has a transmitting unit 110, a receiving unit 120, a setting unit 130, and a control unit 140. The functional configuration shown in Figure 42 is merely an example. The names of the functional divisions and functional units may be any as long as they can execute the operations related to the embodiment of the present invention.
[0433] The transmitter 110 has a function of generating a signal to be transmitted to the terminal 20 side and transmitting the signal wirelessly. The transmitter 110 also transmits inter-network node messages to other network nodes. The receiver 120 has a function of receiving various signals transmitted from the terminal 20 and acquiring, for example, information of higher layers from the received signals. The transmitter 110 also has a function of transmitting NR-PSS, NR-SSS, NR-PBCH, DL / UL control signals, etc. to the terminal 20. The receiver 120 also receives inter-network node messages from other network nodes.
[0434] The setting unit 130 stores preset setting information and various setting information to be transmitted to the terminal 20. The setting information includes, for example, information related to a modulation method, an encoding method, a channel bandwidth, a waveform, and / or a subcarrier spacing.
[0435] The control unit 140 performs control to realize the functions described in the embodiments. Furthermore, as described in the embodiments, the control unit 140 performs control related to the modulation method, the coding method, the channel bandwidth, the waveform, and / or the subcarrier spacing. The function unit related to signal transmission in the control unit 140 may be included in the transmitting unit 110, and the function unit related to signal reception in the control unit 140 may be included in the receiving unit 120.
[0436] <Terminal 20> Figure 43 is a diagram showing an example of the functional configuration of terminal 20 in an embodiment of the present invention. As shown in Figure 43, terminal 20 has a transmitting unit 210, a receiving unit 220, a setting unit 230, and a control unit 240. The functional configuration shown in Figure 43 is merely an example. The names of the functional divisions and functional units may be any as long as they can execute the operations related to the embodiment of the present invention.
[0437] The transmitter 210 creates a transmission signal from transmission data and transmits the transmission signal wirelessly. The receiver 220 wirelessly receives various signals and acquires higher layer signals from the received physical layer signals. The receiver 220 also has a function of receiving NR-PSS, NR-SSS, NR-PBCH, DL / UL / SL control signals, and the like transmitted from the base station 10. For example, the transmitter 210 transmits a PSCCH (Physical Sidelink Control Channel), a PSSCH (Physical Sidelink Shared Channel), a PSDCH (Physical Sidelink Discovery Channel), a PSBCH (Physical Sidelink Broadcast Channel), and the like to another terminal 20 as D2D communication, and the receiver 220 receives the PSCCH, PSSCH, PSDCH, or PSBCH, and the like, from the other terminal 20.
[0438] The setting unit 230 stores various setting information received from the base station 10 by the receiving unit 220. The setting unit 230 also stores setting information that is set in advance. The setting information includes, for example, information related to a modulation method, a coding method, a channel bandwidth, a waveform, and / or a subcarrier spacing.
[0439] The control unit 240 performs control to realize the functions described in the embodiments. Furthermore, as described in the embodiments, the control unit 240 performs control related to the modulation method, coding method, channel bandwidth, waveform, and / or subcarrier spacing. The function unit related to signal transmission in the control unit 240 may be included in the transmitting unit 210, and the function unit related to signal reception in the control unit 240 may be included in the receiving unit 220.
[0440] The transmitting unit and receiving unit in the embodiment of the present invention may execute operations that cause the transmitter and receiver shown in Fig. 22, Fig. 29, or Fig. 38 to function. The terminal 20 and the base station 10 may have the transmitting unit and the receiving unit.
[0441] (Hardware Configuration) The block diagrams (FIGS. 42 and 43) used to explain the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method for realizing each functional block is not particularly limited. That is, each functional block may be realized using a single device that is physically or logically coupled, or may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wires, wirelessly, etc.) and these multiple devices. The functional block may be realized by combining software with the single device or the multiple devices.
[0442] Functions include, but are not limited to, judgment, determination, judgment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, consideration, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assignment. For example, a functional block (component) that performs transmission is called a transmitting unit or transmitter. As mentioned above, there are no particular limitations on how these functions are implemented.
[0443] For example, the base station 10, the terminal 20, etc. according to an embodiment of the present disclosure may function as a computer that performs processing of the wireless communication method of the present disclosure. Fig. 44 is a diagram illustrating an example of the hardware configuration of the base station 10 and the terminal 20 according to an embodiment of the present disclosure. The base station 10 and the terminal 20 described above may be physically configured as a computer device including a processor 1001, a storage device 1002, an auxiliary storage device 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, etc.
[0444] In the following description, the term "apparatus" can be interpreted as a circuit, a device, a unit, etc. The hardware configuration of the base station 10 and the terminal 20 may be configured to include one or more of the apparatuses shown in the drawings, or may be configured to exclude some of the apparatuses.
[0445] Each function in the base station 10 and the terminal 20 is realized by loading specified software (programs) onto hardware such as the processor 1001, the memory device 1002, etc., so that the processor 1001 performs calculations, controls communication by the communication device 1004, and controls at least one of reading and writing data in the memory device 1002 and the auxiliary memory device 1003.
[0446] The processor 1001 controls the entire computer by running, for example, an operating system. The processor 1001 may be configured as a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, etc. For example, the above-mentioned control unit 140, control unit 240, etc. may be realized by the processor 1001.
[0447] Furthermore, the processor 1001 reads programs (program codes), software modules, data, etc. from at least one of the auxiliary storage device 1003 and the communication device 1004 into the storage device 1002 and executes various processes in accordance with the programs. The programs used are those that cause a computer to execute at least some of the operations described in the above-described embodiments. For example, the control unit 140 of the base station 10 shown in FIG. 42 may be implemented by a control program stored in the storage device 1002 and running on the processor 1001. Furthermore, for example, the control unit 240 of the terminal 20 shown in FIG. 43 may be implemented by a control program stored in the storage device 1002 and running on the processor 1001. While the above-described various processes have been described as being executed by one processor 1001, they may also be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The programs may also be transmitted from a network via a telecommunications line.
[0448] The storage device 1002 is a computer-readable recording medium and may be configured, for example, by at least one of a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a random access memory (RAM), etc. The storage device 1002 may also be called a register, a cache, a main memory, etc. The storage device 1002 can store executable programs (program codes), software modules, etc. for implementing a communication method according to an embodiment of the present disclosure.
[0449] The secondary storage device 1003 is a computer-readable recording medium, and may be, for example, at least one of an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy (registered trademark) disk, a magnetic strip, etc. The above-mentioned storage medium may be, for example, a database, a server, or other appropriate medium including at least one of the storage device 1002 and the secondary storage device 1003.
[0450] The communication device 1004 is hardware (transmission / reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may be configured to include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc. to realize at least one of frequency division duplex (FDD) and time division duplex (TDD). For example, a transmission / reception antenna, an amplifier unit, a transmission / reception unit, a transmission path interface, etc. may be realized by the communication device 1004. The transmission / reception unit may be implemented as a transmission unit and a reception unit that are physically or logically separated.
[0451] The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that receives input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that outputs to the outside. Note that the input device 1005 and the output device 1006 may be integrated into one device (e.g., a touch panel).
[0452] Furthermore, each device such as the processor 1001 and the storage device 1002 is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses between each device.
[0453] Furthermore, the base station 10 and the terminal 20 may be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA), and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented using at least one of these pieces of hardware.
[0454] Fig. 45 shows a configuration example of a vehicle 2001. As shown in Fig. 45, the vehicle 2001 includes a drive unit 2002, a steering unit 2003, an accelerator pedal 2004, a brake pedal 2005, a shift lever 2006, front wheels 2007, rear wheels 2008, an axle 2009, an electronic control unit 2010, various sensors 2021 to 2029, an information service unit 2012, and a communication module 2013. Each aspect / embodiment described in the present disclosure may be applied to a communication device mounted on the vehicle 2001, and may be applied to the communication module 2013, for example.
[0455] The drive unit 2002 is configured, for example, by an engine, a motor, or a hybrid of an engine and a motor. The steering unit 2003 includes at least a steering wheel (also called a handle) and is configured to steer at least one of the front wheels and the rear wheels based on the operation of the steering wheel operated by the user.
[0456] The electronic control unit 2010 is composed of a microprocessor 2031, a memory (ROM, RAM) 2032, and a communication port (IO port) 2033. Signals are input to the electronic control unit 2010 from various sensors 2021 to 2029 provided in the vehicle 2001. The electronic control unit 2010 may also be called an ECU (Electronic Control Unit).
[0457] The signals from the various sensors 2021 to 2029 include a current signal from a current sensor 2021 that senses the current of the motor, a front or rear wheel rotation speed signal obtained by a rotation speed sensor 2022, a front or rear wheel air pressure signal obtained by an air pressure sensor 2023, a vehicle speed signal obtained by a vehicle speed sensor 2024, an acceleration signal obtained by an acceleration sensor 2025, an accelerator pedal depression amount signal obtained by an accelerator pedal sensor 2029, a brake pedal depression amount signal obtained by a brake pedal sensor 2026, a shift lever operation signal obtained by a shift lever sensor 2027, and a detection signal for detecting obstacles, vehicles, pedestrians, etc. obtained by an object detection sensor 2028.
[0458] The information service unit 2012 is composed of various devices, such as a car navigation system, an audio system, speakers, a television, and a radio, for providing (outputting) various types of information, such as driving information, traffic information, and entertainment information, and one or more ECUs for controlling these devices. The information service unit 2012 uses information acquired from external devices via the communication module 2013 or the like to provide various types of multimedia information and multimedia services to the occupants of the vehicle 2001. The information service unit 2012 may include input devices (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, a touch panel, etc.) that accept input from the outside, and may also include output devices (e.g., a display, a speaker, an LED lamp, a touch panel, etc.) that output information to the outside.
[0459] The driving assistance system unit 2030 is composed of various devices that provide functions for preventing accidents and reducing the driving burden on the driver, such as millimeter-wave radar, LiDAR (Light Detection and Ranging), cameras, positioning locators (e.g., GNSS, etc.), map information (e.g., high-definition (HD) maps, autonomous vehicle (AV) maps, etc.), gyro systems (e.g., IMU (Inertial Measurement Unit), INS (Inertial Navigation System), etc.), AI (Artificial Intelligence) chips, and AI processors, as well as one or more ECUs that control these devices. In addition, the driving assistance system unit 2030 transmits and receives various information via the communication module 2013 to realize the driving assistance function or the autonomous driving function.
[0460] The communication module 2013 can communicate with the microprocessor 2031 and components of the vehicle 2001 via the communication port. For example, the communication module 2013 transmits and receives data via the communication port 2033 to and from the drive unit 2002, steering unit 2003, accelerator pedal 2004, brake pedal 2005, shift lever 2006, front wheels 2007, rear wheels 2008, axle 2009, microprocessor 2031 and memory (ROM, RAM) 2032 in the electronic control unit 2010, and sensors 2021 to 29, which are provided in the vehicle 2001.
[0461] The communication module 2013 is a communication device that can be controlled by the microprocessor 2031 of the electronic control unit 2010 and can communicate with an external device. For example, it transmits and receives various information to and from the external device via wireless communication. The communication module 2013 may be located either inside or outside the electronic control unit 2010. The external device may be, for example, a base station, a mobile station, or the like.
[0462] The communication module 2013 may transmit at least one of signals from the above-mentioned various sensors 2021-2028 input to the electronic control unit 2010, information obtained based on the signals, and information based on input from the outside (user) obtained via the information service unit 2012 to an external device via wireless communication. The electronic control unit 2010, the various sensors 2021-2028, the information service unit 2012, etc. may be referred to as input units that accept input. For example, the PUSCH transmitted by the communication module 2013 may include information based on the above-mentioned input.
[0463] The communication module 2013 receives various information (traffic information, traffic signal information, vehicle-to-vehicle information, etc.) transmitted from external devices and displays it on an information service unit 2012 provided in the vehicle 2001. The information service unit 2012 may be called an output unit that outputs information (for example, outputs information to a device such as a display or speaker based on the PDSCH (or data / information decoded from the PDSCH) received by the communication module 2013). The communication module 2013 also stores the various information received from external devices in a memory 2032 that can be used by the microprocessor 2031. Based on the information stored in the memory 2032, the microprocessor 2031 may control the drive unit 2002, steering unit 2003, accelerator pedal 2004, brake pedal 2005, shift lever 2006, front wheels 2007, rear wheels 2008, axles 2009, sensors 2021 to 2029, etc. provided in the vehicle 2001.
[0464] (Summary of the embodiment) As described above, according to the embodiment of the present invention, there is provided a terminal having a control unit that pre-trains a neural network (NN) that maps a bit sequence to an index, outputs a one-hot encoder vector corresponding to the index, and outputs modulation symbols using the one-hot encoder vector and a PA (power amplifier) saturation point as input, and a transmission unit that transmits to a base station using modulation symbols determined based on a constellation generated by the trained NN.
[0465] With the above configuration, the terminal 20 can communicate effectively even when there is an influence of a PA (Power Amplifier) saturation point by using the constellation generated by the trained NN in the MCS table. In other words, it is possible to reduce the performance degradation caused by the PA saturation point in the wireless communication system.
[0466] The transmitter may normalize the constellation and convert it to amplitude phase shift keying (APSK). This configuration optimizes performance and complexity by using a normalized constellation generated by a trained neural network. Furthermore, the normalized constellation reduces the complexity of demodulation.
[0467] The transmitter may use a plurality of different modulation schemes based on a plurality of different constellations generated by the neural network that has learned using a plurality of different PA saturation points as input. With this configuration, the terminal 20 can communicate effectively even when affected by the PA saturation point by using the constellations generated by the learned neural network in an MCS table.
[0468] The transmission unit may transmit, to the base station, information indicating whether to use a modulation scheme based on the constellation. With this configuration, the terminal 20 can communicate preferably even when there is an influence of a PA saturation point by using the constellation generated by the trained NN by including it in an MCS table.
[0469] The terminal 20 may further include a receiving unit that receives, from the base station, information indicating whether to use a modulation scheme based on the constellation. With this configuration, the terminal 20 can communicate effectively even when there is an influence of a PA saturation point by using the constellation generated by the trained NN by including it in the MCS table.
[0470] Furthermore, according to an embodiment of the present invention, there is provided a communication method in which a terminal executes the steps of: mapping a bit sequence to an index, outputting a one-hot encoder vector corresponding to the index, and pre-training a neural network (NN) that outputs modulation symbols using the one-hot encoder vector and a PA (Power Amplifier) saturation point as input; and transmitting to a base station using modulation symbols determined based on a constellation generated by the trained NN.
[0471] With the above configuration, the terminal 20 can communicate effectively even when there is an influence of a PA (Power Amplifier) saturation point by using the constellation generated by the trained NN in the MCS table. In other words, it is possible to reduce the performance degradation caused by the PA saturation point in the wireless communication system.
[0472] (Supplementary Notes on the Embodiments) Although the embodiments of the present invention have been described above, the disclosed invention is not limited to such embodiments, and those skilled in the art will understand various modifications, alterations, alternatives, and substitutions. While specific numerical examples have been used to facilitate understanding of the invention, unless otherwise specified, these numerical values are merely examples, and any appropriate values may be used. The division of items in the above description is not essential to the present invention; matters described in two or more items may be used in combination as needed, and matters described in one item may apply to matters described in another item (as long as there is no contradiction). Boundaries between functional units or processing units in functional block diagrams do not necessarily correspond to boundaries between physical components. The operations of multiple functional units may be performed by a single physical component, or the operations of a single functional unit may be performed by multiple physical components. The order of processing steps described in the embodiments may be reversed as long as there is no contradiction. For convenience of processing description, the base station 10 and terminal 20 have been described using functional block diagrams, but such devices may be realized by hardware, software, or a combination thereof. The software operated by the processor of the base station 10 in accordance with an embodiment of the present invention and the software operated by the processor of the terminal 20 in accordance with an embodiment of the present invention may each be stored in random access memory (RAM), flash memory, read-only memory (ROM), EPROM, EEPROM, register, hard disk (HDD), removable disk, CD-ROM, database, server, or any other suitable storage medium.
[0473] Furthermore, the notification of information is not limited to the aspects / embodiments described in the present disclosure, and may be performed using other methods. For example, the notification of information may be performed by physical layer signaling (e.g., Downlink Control Information (DCI), Uplink Control Information (UCI)), higher layer signaling (e.g., Radio Resource Control (RRC) signaling, Medium Access Control (MAC) signaling), broadcast information (Master Information Block (MIB), System Information Block (SIB)), other signals, or a combination thereof. Furthermore, the RRC signaling may be referred to as an RRC message, and may be, for example, an RRC Connection Setup message, an RRC Connection Reconfiguration message, or the like.
[0474] Each aspect / embodiment described in the present disclosure may be implemented using any of the following standards: LTE (Long Term Evolution), LTE-Advanced (LTE-A), SUPER 3G, IMT-Advanced, 4G (4th generation mobile communication system), 5G (5th generation mobile communication system), 6th generation mobile communication system (6G), xth generation mobile communication system (xG) (xG (x is, for example, an integer or a decimal number)), FRA (Future Radio Access), NR (new Radio), New radio access (NX), Future generation radio access (FX), W-CDMA (registered trademark), GSM (registered trademark), CDMA2000, UMB (Ultra Mobile Broadband), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE 802.17 (WiMAX (registered trademark)), IEEE 802.19 (WiMAX (registered trademark)), IEEE 802.20 (WiMAX (registered trademark)), IEEE 802.21 (Wi-Fi (registered trademark)), IEEE 802.22 (WiMAX (registered trademark)), IEEE 802.23 (WiMAX (registered trademark)), IEEE 802.24 (WiMAX (registered trademark)), IEEE 802.25 (WiMAX (registered trademark)), IEEE 802.26 (WiMAX (registered trademark)), IEEE 802.27 (WiMAX (registered trademark)), IEEE 802.28 (WiMAX (registered trademark)), IEEE 802.29 (WiMAX (registered trademark)), IEEE 802.30 (WiMAX (registered trademark)), IEEE 802.31 (Wi-Fi (registered trademark)), IEEE 802.32 (WiMAX (registered trademark)), IEEE 802.33 (WiMAX (registered trademark)), IEEE 802.34 ( The present invention may be applied to at least one of systems using 802.20, UWB (Ultra-Wide Band), Bluetooth (registered trademark), or other suitable systems, and next-generation systems that are extended, modified, created, or defined based on these systems. The present invention may also be applied to a combination of multiple systems (e.g., a combination of LTE and / or LTE-A with 5G).
[0475] The order of the procedures, sequences, flowcharts, etc. of each aspect / embodiment described herein may be rearranged unless it is consistent. For example, the methods described in this disclosure present elements of various steps using an example order and are not limited to the particular order presented.
[0476] In this specification, a specific operation described as being performed by the base station 10 may be performed by its upper node in some cases. In a network consisting of one or more network nodes having the base station 10, it is clear that various operations performed for communication with the terminal 20 may be performed by at least one of the base station 10 and another network node other than the base station 10 (such as, but not limited to, an MME or an S-GW). Although the above example illustrates a case where there is one other network node other than the base station 10, the other network node may be a combination of multiple other network nodes (such as an MME and an S-GW).
[0477] The information, signals, etc. described in the present disclosure may be output from a higher layer (or a lower layer) to a lower layer (or a higher layer), or may be input / output via multiple network nodes.
[0478] Input and output information may be stored in a specific location (for example, memory) or may be managed using a management table. Input and output information may be overwritten, updated, or added to. Output information may be deleted. Input information may be transmitted to another device.
[0479] In the present disclosure, the determination may be made based on a value represented by one bit (0 or 1), a Boolean value (true or false), or a numerical comparison (e.g., comparison with a predetermined value).
[0480] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.
[0481] Software, instructions, information, etc. may also be transmitted or received over a transmission medium. For example, if software is transmitted from a website, server, or other remote source using wired technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave), then these wired and / or wireless technologies are included within the definition of transmission media.
[0482] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.
[0483] Note that terms described in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of a channel and a symbol may be a signal (signaling). Furthermore, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, a cell, a frequency carrier, etc.
[0484] As used in this disclosure, the terms "system" and "network" are used interchangeably.
[0485] Furthermore, the information, parameters, etc. described in the present disclosure may be expressed using absolute values, may be expressed using relative values from a predetermined value, or may be expressed using other corresponding information. For example, a radio resource may be indicated by an index.
[0486] The names used for the above-described parameters are not intended to be limiting in any way. Furthermore, the mathematical expressions using these parameters may differ from those explicitly disclosed in this disclosure. The various channels (e.g., PUCCH, PDCCH, etc.) and information elements may be identified by any suitable names, and therefore the various names assigned to these various channels and information elements are not intended to be limiting in any way.
[0487] In the present disclosure, terms such as "base station (BS)," "radio base station," "base station device," "fixed station," "NodeB," "eNodeB (eNB)," "gNodeB (gNB)," "access point," "transmission point," "reception point," "transmission / reception point," "cell," "sector," "cell group," "carrier," and "component carrier" may be used interchangeably. A base station may also be referred to by terms such as a macrocell, a small cell, a femtocell, and a picocell.
[0488] A base station can accommodate one or more (e.g., three) cells. When a base station accommodates multiple cells, the overall coverage area of the base station can be partitioned into multiple smaller areas, and each smaller area can also be provided with communication services by a base station subsystem (e.g., a small indoor base station (RRH: Remote Radio Head)). The terms "cell" or "sector" refer to part or all of the coverage area of a base station and / or base station subsystem that provides communication services within that coverage.
[0489] In the present disclosure, the base station transmitting information to a terminal may be interpreted as the base station instructing the terminal to control or operate based on the information.
[0490] In this disclosure, the terms "Mobile Station (MS)," "user terminal," "User Equipment (UE)," "terminal," and the like may be used interchangeably.
[0491] A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or some other suitable terminology.
[0492] At least one of the base station and the mobile station may be referred to as a transmitting device, a receiving device, a communication device, etc. At least one of the base station and the mobile station may be a device mounted on a mobile object, the mobile object itself, etc. The mobile object refers to a movable object, and may move at any speed. Naturally, this also includes cases where the mobile object is stationary. Examples of the mobile object include, but are not limited to, vehicles, transport vehicles, automobiles, motorcycles, bicycles, connected cars, excavators, bulldozers, wheel loaders, dump trucks, forklifts, trains, buses, handcars, rickshaws, ships and other watercraft, airplanes, rockets, satellites, drones (registered trademark), multicopters, quadcopters, balloons, and objects mounted thereon. The mobile object may also be a mobile object that moves autonomously based on an operational command. It may be a vehicle (e.g., a car, an airplane, etc.), an unmanned mobile object (e.g., a drone, an autonomous vehicle, etc.), or a robot (manned or unmanned). At least one of the base station and the mobile station may be a device that does not necessarily move during communication operations. For example, at least one of the base station and the mobile station may be an IoT (Internet of Things) device such as a sensor.
[0493] Furthermore, a base station in the present disclosure may be read as a user terminal. For example, the aspects / embodiments of the present disclosure may be applied to a configuration in which communication between a base station and a user terminal is replaced with communication between multiple terminals 20 (which may be called, for example, Device-to-Device (D2D) or Vehicle-to-Everything (V2X)). In this case, the terminal 20 may be configured to have the functions of the base station 10 described above. Furthermore, terms such as "uplink" and "downlink" may be read as terms corresponding to terminal-to-terminal communication (for example, "side"). For example, terms such as an uplink channel and a downlink channel may be read as a side channel.
[0494] Similarly, the user terminal in the present disclosure may be read as a base station, in which case the base station may be configured to have the functions of the user terminal described above.
[0495] As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, inquiring (e.g., searching in a table, database, or other data structure), ascertaining, and the like. "Determining" and "determining" may also include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in memory), and the like. Furthermore, "judgment" and "decision" can include regarding resolving, selecting, choosing, establishing, comparing, etc. as having been "judged" or "decided." In other words, "judgment" and "decision" can include regarding some action as having been "judged" or "decided." Furthermore, "judgment (decision)" can be interpreted as "assuming," "expecting," "considering," etc.
[0496] The terms "connected," "coupled," or any variation thereof, refer to any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are "connected" or "coupled" to each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, "connected" may be read as "access." As used in this disclosure, two elements may be considered to be "connected" or "coupled" to each other using one or more wires, cables, and / or printed electrical connections, as well as electromagnetic energy having wavelengths in the radio frequency range, microwave range, and optical (both visible and invisible) range, as some non-limiting and non-exhaustive examples.
[0497] The reference signal may be abbreviated as RS (Reference Signal) or may be called a pilot depending on the applicable standard.
[0498] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."
[0499] As used in this disclosure, any reference to an element using a designation such as "first," "second," etc. does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must in some way precede the second element.
[0500] The "means" in the configuration of each of the above devices may be replaced with "part," "circuit," "device," etc.
[0501] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Furthermore, when the term "or" is used in this disclosure, it is not intended to be an exclusive or.
[0502] A radio frame may be composed of one or more frames in the time domain. Each of the one or more frames in the time domain may be called a subframe. A subframe may further be composed of one or more slots in the time domain. A subframe may have a fixed time length (e.g., 1 ms) that is independent of numerology.
[0503] Numerology may be communication parameters that apply to the transmission and / or reception of a signal or channel, such as subcarrier spacing (SCS), bandwidth, symbol length, cyclic prefix length, transmission time interval (TTI), number of symbols per TTI, radio frame structure, specific filtering operations performed by the transceiver in the frequency domain, specific windowing operations performed by the transceiver in the time domain, etc.
[0504] A slot may be composed of one or more symbols (such as an Orthogonal Frequency Division Multiplexing (OFDM) symbol or a Single Carrier Frequency Division Multiple Access (SC-FDMA) symbol) in the time domain. A slot may be a time unit based on numerology.
[0505] A slot may include multiple minislots. Each minislot may consist of one or multiple symbols in the time domain. A minislot may also be called a subslot. A minislot may consist of fewer symbols than a slot. A PDSCH (or PUSCH) transmitted in a time unit larger than a minislot may be called PDSCH (or PUSCH) mapping type A. A PDSCH (or PUSCH) transmitted using a minislot may be called PDSCH (or PUSCH) mapping type B.
[0506] The radio frame, subframe, slot, minislot, and symbol all represent time units for transmitting signals, and may be referred to by other names corresponding to the radio frame, subframe, slot, minislot, and symbol.
[0507] For example, one subframe may be called a transmission time interval (TTI), multiple consecutive subframes may be called a TTI, or one slot or one minislot may be called a TTI. That is, at least one of the subframe and the TTI may be a subframe (1 ms) in existing LTE, a period shorter than 1 ms (for example, 1-13 symbols), or a period longer than 1 ms. Note that the unit representing the TTI may be called a slot, minislot, etc. instead of a subframe.
[0508] Here, TTI refers to, for example, the smallest time unit for scheduling in wireless communication. For example, in an LTE system, a base station performs scheduling to allocate wireless resources (such as frequency bandwidth and transmission power that can be used by each terminal 20) to each terminal 20 in TTI units. Note that the definition of TTI is not limited to this.
[0509] The TTI may be a transmission time unit for a channel-encoded data packet (transport block), a code block, a code word, etc., or may be a processing unit for scheduling, link adaptation, etc. When a TTI is given, the time interval (e.g., the number of symbols) to which a transport block, a code block, a code word, etc. is actually mapped may be shorter than the TTI.
[0510] When one slot or one minislot is called a TTI, one or more TTIs (i.e., one or more slots or one or more minislots) may be the minimum time unit for scheduling. Also, the number of slots (minislots) constituting the minimum time unit for scheduling may be controlled.
[0511] A TTI having a time length of 1 ms may be called a regular TTI (TTI in LTE Rel. 8-12), normal TTI, long TTI, regular subframe, normal subframe, long subframe, slot, etc. A TTI shorter than a regular TTI may be called a shortened TTI, short TTI, partial or fractional TTI, shortened subframe, short subframe, minislot, subslot, slot, etc.
[0512] In addition, a long TTI (e.g., a normal TTI, a subframe, etc.) may be interpreted as a TTI having a time length of more than 1 ms, and a short TTI (e.g., a shortened TTI, etc.) may be interpreted as a TTI having a TTI length shorter than the TTI length of a long TTI and greater than or equal to 1 ms.
[0513] A resource block (RB) is a resource allocation unit in the time domain and the frequency domain, and may include one or more consecutive subcarriers in the frequency domain. The number of subcarriers included in an RB may be the same regardless of the numerology, for example, 12. The number of subcarriers included in an RB may be determined based on the numerology.
[0514] The time domain of an RB may include one or more symbols and may have a length of one slot, one minislot, one subframe, or one TTI. One TTI, one subframe, etc. may each be composed of one or more resource blocks.
[0515] Note that one or more RBs may also be called a physical resource block (PRB), a sub-carrier group (SCG), a resource element group (REG), a PRB pair, an RB pair, etc.
[0516] Furthermore, a resource block may be composed of one or more resource elements (REs). For example, one RE may be a radio resource region of one subcarrier and one symbol.
[0517] A Bandwidth Part (BWP) (which may also be referred to as a fractional bandwidth) may represent a subset of contiguous common resource blocks (RBs) for a given numerology on a given carrier, where the common RBs may be identified by their index relative to a Common Reference Point of the carrier. PRBs may be defined in a BWP and numbered within the BWP.
[0518] The BWP may include a BWP for UL (UL BWP) and a BWP for DL (DL BWP). One or more BWPs may be configured for a UE within one carrier.
[0519] At least one of the configured BWPs may be active, and the UE may not expect to transmit or receive a given signal / channel outside the active BWP. Note that the terms "cell," "carrier," etc. in this disclosure may be read as "BWP."
[0520] The above-described structures of radio frames, subframes, slots, minislots, symbols, etc. are merely examples, and various changes may be made to the number of subframes included in a radio frame, the number of slots per subframe or radio frame, the number of minislots included in a slot, the number of symbols and RBs included in a slot or minislot, the number of subcarriers included in an RB, the number of symbols in a TTI, the symbol length, the cyclic prefix (CP) length, etc.
[0521] In this disclosure, where articles are added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.
[0522] In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "coupled" may also be interpreted in the same way as "different."
[0523] The aspects / embodiments described in this disclosure may be used alone, in combination, or switched depending on the implementation. Notification of predetermined information (e.g., notification that "X is true") is not limited to explicit notification, but may be implicit (e.g., not notifying the predetermined information).
[0524] Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not have any limiting meaning on the present disclosure.
[0525] REFERENCE SIGNS LIST 10 Base station 110 Transmitter 120 Receiver 130 Setting unit 140 Control unit 20 Terminal 210 Transmitter 220 Receiver 230 Setting unit 240 Control unit 30 Power transmission equipment 40 Core network 1001 Processor 1002 Storage device 1003 Auxiliary storage device 1004 Communication device 1005 Input device 1006 Output device 2001 Vehicle 2002 Drive unit 2003 Steering unit 2004 Accelerator pedal 2005 Brake pedal 2006 Shift lever 2007 Front wheel 2008 Rear wheel 2009 Axle 2010 Electronic control unit 2012 Information service unit 2013 Communication module 2021 Current sensor 2022 RPM sensor 2023 Air pressure sensor 2024 Vehicle speed sensor 2025 Acceleration sensor 2026 Brake pedal sensor 2027 Shift lever sensor 2028 Object detection sensor 2029 Accelerator pedal sensor 2030 Driving assistance system unit 2031 Microprocessor 2032 Memory (ROM, RAM) 2033 Communication port (IO port)
Claims
1. A terminal having a control unit that pre-trains a neural network (NN) that maps a bit sequence to an index, outputs a one-hot encoder vector corresponding to the index, and outputs a modulation symbol using the one-hot encoder vector and a PA (Power Amplifier) saturation point as input, and a transmission unit that transmits to a base station using a modulation symbol determined based on a constellation generated by the trained NN.
2. The terminal of claim 1, wherein the transmitting unit normalizes the constellation and converts it to APSK (Amplitude phase shift keying).
3. The terminal according to claim 1, wherein the transmitting unit uses a plurality of different modulation schemes based on a plurality of different constellations generated by the neural network that has learned using a plurality of different PA saturation points as input.
4. The terminal according to claim 1, wherein the transmitting unit transmits information indicating whether or not to use a modulation method based on the constellation to the base station.
5. The terminal according to claim 1, further comprising a receiving section for receiving information indicating whether or not to use a modulation method based on said constellation from said base station.
6. A communication method in which a terminal executes the steps of: mapping a bit sequence to an index, outputting a one-hot encoder vector corresponding to the index, and pre-training a neural network (NN) that uses the one-hot encoder vector and a PA (Power Amplifier) saturation point as input to output a modulation symbol; and transmitting to a base station using a modulation symbol determined based on a constellation generated by the trained NN.
Citation Information
Patent Citations
Reception quality evaluation device
JP2021136490A
Transmission device, reception device, transmission method, and reception method
JP2022033051A
Learning in communication systems
JP2022534603A
A personalized air interface
JP2023503111A
Modulation scheme discrimination device, control circuit, storage medium and modulation scheme discrimination method
WO2022130625A1
