Intelligent transceiving system for communication based on dynamic channel perception

Abstract

The invention belongs to the field of communication engineering, and particularly relates to an intelligent transceiving system for communication based on dynamic channel perception. The invention discloses a system named as an intelligent transceiving system for communication based on dynamic channel perception, and aims to solve the problems of link reliability reduction and response lag caused by separation of perception and transceiving decisions and conflict between perception overhead and resource efficiency of an existing communication system in a dynamic channel environment. The system comprises a channel dynamic sensing front end, a channel state evolution modeling engine, a central cooperative controller, an adaptive modulation coding unit, a power and beam joint regulation and control module and a feedback information compression encoder. Through multi-dimensional channel feature extraction, implicit channel deduction without pilot frequency dependence and prospective parameter adjustment based on a channel evolution trend, high robustness and high resource efficiency under high-speed movement and high-interference scenes are realized on the premise of not increasing pilot frequency overhead.

Description

Technical Field

[0001] This invention belongs to the field of communication engineering, and more specifically, relates to an intelligent transceiver system for communication based on dynamic channel awareness. Background Technology

[0002] Against the backdrop of the continuous evolution of modern wireless communication systems, the reliability of communication links, spectrum utilization efficiency, and anti-interference capabilities have become core indicators for measuring the overall performance of a system. With the rapid popularization of emerging application scenarios such as the Internet of Things (IoT), industrial automation, vehicle-to-everything (V2X) communication, and high-density mobile communication, the wireless channel environment exhibits unprecedented dynamism and complexity. Its time-varying, multipath, fading, and external electromagnetic interference characteristics pose severe challenges to traditional transceiver mechanisms. In this technological ecosystem, how to construct a communication system capable of real-time sensing of channel conditions and intelligently adjusting transceiver strategies accordingly has become a key research direction for improving communication robustness and resource scheduling efficiency.

[0003] For a long time, mainstream communication systems have generally adopted a transceiver architecture based on a pre-defined channel model or periodic channel estimation. Such schemes typically rely on the periodic insertion of pilot signals, combined with channel estimation algorithms at the receiver (such as least mean square error estimation or least squares estimation) to obtain channel state information, and then perform subsequent operations such as modulation and coding scheme selection, power control, or antenna beamforming. In static or quasi-static channel environments, this type of method, with its clear structure and mature implementation, effectively ensures the basic performance of the communication link. Especially in fourth-generation mobile communication systems and their early evolutions, this technical path based on fixed feedback periods and static channel assumptions has achieved a good balance between system overhead and performance, supporting large-scale commercial deployment.

[0004] However, with the continuous development of related technologies and the increasingly stringent performance requirements of application scenarios, some inherent characteristics of the aforementioned technical solutions at the principle level have gradually revealed their limitations in addressing new challenges. Specifically, the traditional periodic channel estimation mechanism is essentially based on the premise that "the channel remains stable within the estimation interval." This assumption is difficult to hold in typical dynamic scenarios such as high-speed movement, sudden interference, or dense multi-user access. When the channel coherence time is significantly shortened, even if the pilot density is increased to enhance the estimation frequency, it will inevitably crowd out valuable data transmission resources, leading to a decrease in effective throughput. On the other hand, if the original pilot overhead is maintained, the timeliness of channel state information will be severely delayed, resulting in modulation and coding scheme mismatch, increased bit error rate, and even link interruption. Furthermore, existing systems typically treat channel sensing and transmit / receive decision-making as two relatively independent functional modules. The former only provides discrete channel snapshots, while the latter performs static configuration based on these snapshots, lacking the ability to continuously track channel evolution trends and proactively respond. The reason for this is that traditional architectures do not incorporate channel dynamics itself as a core input variable into the transmit / receive control closed loop, but instead simplify it into an external disturbance that needs to be passively adapted to, thus fundamentally limiting the system's ability to actively manage rapidly changing environments.

[0005] Building on this, a deep-seated technical contradiction is becoming increasingly apparent: on the one hand, improving the adaptability of communication systems to dynamic channels urgently requires higher frequency and finer-grained channel state perception; on the other hand, the additional overhead introduced by the perception process itself (including time, power, and bandwidth) directly erodes the total amount of resources available for effective data transmission, thus restricting the overall system performance. This contradiction is particularly acute in low-latency, high-reliability communication scenarios—for example, in industrial control or autonomous driving applications, not only is millisecond-level end-to-end latency required, but also extremely low packet loss rates. In such cases, any retransmission or rate regression caused by channel perception lag or rigid transmission and reception strategies may trigger system-level failures. Correspondingly, existing technological systems struggle to achieve coordinated optimization among "perception accuracy," "response speed," and "resource efficiency," as their inherent architecture dictates that the other two performance indicators cannot be improved simultaneously without sacrificing any one of them.

[0006] Therefore, how to construct a communication system architecture that can deeply integrate dynamic channel awareness and intelligent transceiver decision-making, so as to achieve continuous and accurate capture of the time-varying characteristics of the channel without significantly increasing system overhead, and adjust the transceiver parameters in real time and adaptively to balance transmission reliability, spectrum efficiency and energy consumption control, has become a key challenge and a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0007] This invention provides an intelligent transceiver system for communication based on dynamic channel awareness, aiming to solve the technical problems of existing communication systems in dynamic channel environments, such as decreased link reliability, limited spectrum utilization, and response lag caused by the separation of channel awareness and transceiver decision-making, and the conflict between sensing overhead and resource efficiency. To achieve the above-mentioned objectives, this invention constructs a closed-loop architecture that deeply integrates channel dynamic modeling, real-time sensing feedback, and adaptive transceiver control. By introducing a multi-dimensional channel feature extraction mechanism, a pilot-independent implicit channel state inference module, and a forward-looking parameter adjustment strategy based on channel evolution trend prediction, it achieves continuous tracking and efficient utilization of the time-varying characteristics of the wireless channel. Without increasing additional pilot overhead, it significantly improves the communication robustness and resource scheduling efficiency of the system in high-speed mobile, sudden interference, and high-density access scenarios.

[0008] The intelligent transceiver system for communication based on dynamic channel awareness includes a channel dynamic sensing front-end, a channel state evolution modeling engine, a central collaborative controller, an adaptive modulation and coding unit, a power and beam joint control module, and a feedback information compression encoder. The channel dynamic sensing front-end is deployed after the receiver's RF link, with its input connected to the output of the analog-to-digital converter, used to extract multi-dimensional channel features from the received baseband signal in real time. The channel state evolution modeling engine is connected to the channel dynamic sensing front-end via a high-speed parallel bus, used to construct a time-series model of the channel state based on the extracted multi-dimensional channel features, and to deduce the channel state vector for the current moment and the next transmission time slot. The central coordinating controller is connected to the channel state evolution modeling engine, the adaptive modulation and coding unit, and the power and beam joint control module, respectively, used to receive the channel state vector and generate corresponding transmit / receive control commands. The adaptive modulation and coding unit dynamically switches the modulation order and forward error correction code rate according to the transmit / receive control commands. The power and beam joint control module synchronously adjusts the transmit power level and the beam pointing of the antenna array according to the transmit / receive control commands. The feedback information compression encoder is connected to the output of the central coordinating controller, used to compress and encode the transmit / receive control commands and transmit them back to the transmitter through the uplink control channel to complete closed-loop coordination.

[0009] Furthermore, the channel dynamic sensing front-end includes a time-frequency joint feature extractor, a multipath component clustering identifier, and a channel noise covariance estimator. The time-frequency joint feature extractor receives a baseband complex signal sequence and extracts the amplitude fading slope, phase rotation rate, and time delay spread index on each subcarrier using a combination of sliding window Fourier transform and time-domain differentiation. The multipath component clustering identifier, based on the extracted time delay spread index, uses a density peak clustering algorithm to group multipath components and identify the dominant path cluster and its corresponding angle of arrival distribution. The channel noise covariance estimator, based on the received signal residual sequence, calculates the local mean of the noise power spectral density using a moving average filter and combines the multipath component clustering results to correct channel estimation bias. The outputs of these three components together constitute a multidimensional channel feature vector, which contains five core elements: channel amplitude change rate, phase jitter intensity, number of multipath clusters, dominant path angle center, and noise covariance matrix.

[0010] In a preferred embodiment of the present invention, the channel state evolution modeling engine employs a two-layer recurrent neural network structure. The bottom layer consists of long short-term memory (LSTM) network units used for time-series modeling of multi-dimensional channel feature vectors, capturing long-term dependencies in the channel state. The top layer consists of gated recurrent units used to fuse the bottom layer output with the symbol energy distribution characteristics of the current transport block, generating a physically meaningful channel state vector. The channel state vector consists of four deterministic parameters: predicted channel gain, estimated coherence time, upper limit of Doppler shift, and channel condition number. The predicted channel gain characterizes the expected channel strength in the next transmission slot, the estimated coherence time reflects the maximum duration for which the channel remains stable, the upper limit of Doppler shift defines the highest rate of channel phase change, and the channel condition number quantifies the ill-conditioned nature of the channel matrix. This channel state vector is cached in a dual-port RAM and written to the input register of the central coordinating controller by a DMA controller at fixed timing.

[0011] The central coordinating controller integrates a rule-based inference engine and a parameter mapping table. The rule-based inference engine has pre-set decision logic based on a four-dimensional parameter combination of the channel state vector. Its judgment conditions include: triggering low-order modulation and high-redundancy coding modes when the predicted channel gain is below a first threshold and the estimated coherence time is below a second threshold; forcing the use of a constant envelope modulation scheme to suppress phase distortion when the upper limit of Doppler shift exceeds a third threshold; and activating an antenna selection mechanism to reduce spatial correlation when the channel condition number is greater than a fourth threshold. The parameter mapping table stores the index of the optimal modulation and coding scheme, power level number, and beamforming weight coefficient set corresponding to different channel state intervals. The central coordinating controller directly looks up the table based on the output of the rule-based inference engine to obtain transmit / receive control commands, avoiding the delay caused by online optimization calculations.

[0012] The adaptive modulation and coding unit includes a modulation format selection switch, a coding rate configuration register, and an interleaving depth controller. The modulation format selection switch receives a modulation scheme index from the central coordinating controller and switches the output of the QPSK, 16QAM, or π / 2-BPSK modulator through a multiplexer. The coding rate configuration register loads the corresponding LDPC code generation matrix according to the coding scheme index. The interleaving depth controller dynamically adjusts the number of rows of the bit interleaver based on the coherence time estimate to ensure that the interleaving period covers an integer multiple of the channel coherence time, thereby maximizing diversity gain.

[0013] The power and beamforming joint control module includes a digital predistortion compensator, a power amplifier bias controller, and a phased array beamforming network. The digital predistortion compensator receives the baseband modulation signal and applies corresponding nonlinear compensation coefficients according to the current power level number to suppress spectral regeneration at high power operating points. The power amplifier bias controller outputs an analog voltage through a digital-to-analog converter to adjust the gate bias current of the final-stage power amplifier, achieving a step-wise adjustment of the transmit power. The phased array beamforming network consists of a phase shifter array and a weighted summer. Its phase shifter control word is generated by converting the beamforming weight coefficient set provided by the central coordinating controller using the CORDIC algorithm, ensuring that the main lobe is precisely aligned with the angular center of the dominant path cluster.

[0014] The feedback information compression encoder employs a context-adaptive binary arithmetic coding mechanism, where the coding context is dynamically updated by the historical sequence of the current channel state vector. The feedback information includes the modulation and coding scheme index, power level number, and differential coding results of the beamforming weight coefficient set. The encoded bitstream is transmitted via a reserved uplink control channel in a single symbol time slot, ensuring that the end-to-end delay of the transmit and receive parameter adjustments does not exceed one transmission time interval.

[0015] Furthermore, the system of the present invention synchronously deploys a channel state buffer and a parameter preloading unit at the transmitting end. The channel state buffer stores the historical channel state vectors of the most recent N transmission time slots to support cross-time slot analysis of channel evolution trends. After receiving feedback information, the parameter preloading unit preloads the modulation and coding parameters, power configuration, and beam weights required for the next transmission time slot into the configuration register of the corresponding hardware module, achieving zero-wait handover. N is a positive integer, and its value is equal to the maximum allowed retransmission number of the system plus one, ensuring that parameter consistency is maintained even when HARQ retransmission occurs.

[0016] In another preferred embodiment of the present invention, an abnormal channel event detector is provided between the channel dynamic sensing front-end and the channel state evolution modeling engine. The abnormal channel event detector continuously monitors the abrupt changes in the multi-dimensional channel feature vector. When the rate of change of any feature component exceeds a preset abrupt change threshold, an emergency sensing mode is immediately triggered, shortening the feature extraction window to the minimum processable length and increasing the update frequency of the channel state vector to once per symbol period until the channel returns to stability. In this mode, the central coordination controller prioritizes link reliability, forcibly employing the lowest-order modulation and highest-redundancy coding, and suspends beam scanning operations to quickly stabilize the communication link.

[0017] The system described in this invention transforms channel dynamics from passive disturbances into active control variables, achieving end-to-end closed-loop coordination of sensing, modeling, decision-making, and execution. Channel sensing no longer relies on periodic pilot insertion but utilizes the statistical characteristics of the data symbols themselves for implicit inference, thereby eliminating sensing overhead and data transmission resource contention in traditional schemes. Channel state information includes not only instantaneous snapshots but also evolutionary trend predictions, enabling forward-looking adjustments to transmit and receive parameters. The generation of transmit and receive control commands is based on deterministic rules and pre-set mapping tables, avoiding the real-time computational burden of complex optimization algorithms and ensuring millisecond-level response capabilities. These technical features work together to enable this invention to simultaneously optimize transmission reliability, spectral efficiency, and energy consumption control in dynamic channel environments, effectively overcoming the inherent contradictions among "sensing accuracy," "response speed," and "resource efficiency" in existing technologies.

[0018] The hardware implementation of the intelligent transceiver system based on dynamic channel awareness adopts a system-on-a-chip architecture. The channel dynamic awareness front-end, channel state evolution modeling engine, and central coordinating controller are integrated into the same baseband processing chip, achieving low-latency data exchange through an on-chip high-speed interconnect bus. The adaptive modulation and coding unit and the power and beam joint control module are deployed in the radio frequency integrated circuit, receiving control commands through a dedicated configuration interface. The feedback information compression encoder is embedded in the media access control layer protocol stack, tightly coupled with the physical layer. The entire system is driven by a unified clock source, ensuring that the operation of each module is strictly synchronized with the system frame structure, meeting the timing constraints of low-latency, high-reliability communication.

[0019] In summary, this invention, by constructing an intelligent transceiver architecture driven by channel dynamics, achieves continuous channel sensing without additional pilot overhead, forward-looking parameter adjustment based on evolutionary trends, and hardware-friendly deterministic control logic. It fundamentally solves the technical bottlenecks of traditional communication systems in dynamic channel environments, such as perception lag, rigid decision-making, and resource waste, and provides an engineering-featured technical path for future high-mobility, high-density, and high-reliability wireless communication scenarios. Attached Figure Description

[0020] The invention will now be further described with reference to the accompanying drawings.

[0021] Figure 1 This is a schematic diagram of the overall structure of the intelligent transceiver system for communication based on dynamic channel awareness as described in this invention. Detailed Implementation

[0022] This invention provides an intelligent transceiver system for communication based on dynamic channel awareness. Its overall architecture consists of a dynamic channel awareness front-end, a channel state evolution modeling engine, a central collaborative controller, an adaptive modulation and coding unit, a power and beam joint control module, and a feedback information compression encoder. The modules interact and synchronize control through a high-speed interconnect bus or a dedicated configuration interface. The technical solution of this invention will be described in detail below, focusing on the specific composition of the system's functional modules, signal processing flow, parameter generation mechanism, and hardware deployment method.

[0023] The channel dynamic sensing front-end is deployed after the receiver's RF link, with its input connected to the output of an analog-to-digital converter. It is used to extract multi-dimensional channel features in real time from the received baseband complex signal sequence. This front-end comprises three sub-modules: a time-frequency joint feature extractor, a multipath component clustering identifier, and a channel noise covariance estimator. The time-frequency joint feature extractor receives a continuous stream of baseband complex signal samples and uses a combination of sliding window Fourier transform and time-domain differentiation to calculate the amplitude fading slope, phase rotation rate, and time delay spread for each orthogonal frequency division multiplexing subcarrier. Specifically, the sliding window length is set to several symbol periods. The signal within the window undergoes a fast Fourier transform to obtain the frequency domain response, and the phase rotation rate is obtained through phase difference operations between adjacent windows. The amplitude fading slope is obtained by performing a first-order difference on the amplitude envelope within the window and then normalizing it. The time delay spread is derived by inversely calculating the second moment of the power spectrum of the frequency domain response. These three types of indicators collectively characterize the local dynamic characteristics of the channel in the time-frequency two-dimensional plane.

[0024] The multipath component clusterer receives the delay spread index output by the joint time-frequency feature extractor and, combined with the spatial sampling information of the received signal, uses a density peak clustering algorithm to group the multipath components. This algorithm first calculates the Euclidean distance between each multipath component and other components, then determines two key parameters: local density and minimum high-density distance, ultimately identifying several dominant path clusters. Each dominant path cluster corresponds to an angle-of-arrival (AOA) distribution center, which is obtained by weighted averaging of the AOAs of all multipath components within the cluster, with the weight being their received power. The channel noise covariance estimator estimates the noise power spectral density based on the received signal residual sequence. The residual sequence is obtained by subtracting the signal reconstructed based on the preliminary channel estimation from the original received signal, and then calculating the local noise power mean using a moving average filter. This mean is further corrected based on the multipath component clustering results: if a subcarrier is within the coverage area of ​​a dominant path cluster, its noise estimate is lowered by a certain proportion to compensate for the channel estimation bias; otherwise, it remains unchanged. Ultimately, the multidimensional channel feature vector output by the channel dynamic sensing front-end contains five core elements: channel amplitude change rate, phase jitter intensity, number of multipath clusters, dominant path angle center, and noise covariance matrix.

[0025] In a preferred embodiment of the present invention, the channel state evolution modeling engine is connected to the channel dynamic sensing front-end via a high-speed parallel bus, receives the aforementioned multi-dimensional channel feature vector sequence, and constructs a time-series model of the channel state. The engine employs a two-layer recurrent neural network structure, with a long short-term memory (LSTM) network unit at the bottom and a gated recurrent unit at the top. The LSM network unit receives the multi-dimensional channel feature vector at a fixed time step, and through the synergistic effect of forget gates, input gates, and output gates, learns the dependencies of the channel state over a long time scale, outputting a hidden vector containing the long-term memory state. This hidden vector, along with the symbol energy distribution characteristics of the current transport block, is input to the gated recurrent unit at the top layer. The symbol energy distribution characteristics are provided by the physical layer parsing module at the receiver, characterizing the energy concentration of each symbol in the current transport block, and are used to enhance the model's sensitivity to sudden interference or power fluctuations. After fusing the above two types of information, the gated recurrent unit generates a channel state vector with clear physical meaning. This vector consists of four deterministic parameters: channel gain prediction, coherence time estimate, Doppler shift upper limit, and channel condition number. The channel gain prediction is obtained through a linear mapping of the network output layer, representing the expected channel strength in the next transmission slot. The coherence time estimate is derived by inversely calculating the time decay coefficient of the model's internal state. The upper limit of Doppler shift is calculated using the product of phase jitter intensity and carrier frequency. The channel condition number is determined by both the noise covariance matrix and the channel gain prediction, reflecting the ill-conditioning of the channel matrix. The generated channel state vector is written to a dual-port RAM buffer and synchronously transmitted to the input register of the central co-controller by the DMA controller at a fixed timing.

[0026] The central coordinating controller is connected to the channel state evolution modeling engine, the adaptive modulation and coding unit, and the power and beamforming joint control module, respectively, and is used to receive the channel state vector and generate corresponding transmit and receive control commands. The controller integrates a rule-based inference engine and a parameter mapping table. The rule-based inference engine has multiple pre-set decision logics based on the four-dimensional parameter combination of the channel state vector. When the predicted channel gain is lower than the first threshold and the estimated coherence time is lower than the second threshold, the system determines that the current channel is in a deep fading and rapidly time-varying state, triggering low-order modulation and high-redundancy coding modes. When the upper limit of Doppler frequency shift exceeds the third threshold, it indicates a significant increase in phase distortion risk, forcibly enabling a constant envelope modulation scheme to maintain a constant signal envelope. When the channel condition number is greater than the fourth threshold, it indicates severe correlation in the spatial channel, activating the antenna selection mechanism to reduce dimensionality and improve decoupling performance. The parameter mapping table stores the optimal modulation and coding scheme index, power level number, and beamforming weight coefficient set corresponding to different channel state intervals. This mapping table is pre-trained through offline simulation in typical channel scenarios, covering various operating conditions from static to high-speed mobile and from single-user to high-density access. The central collaborative controller directly looks up the table based on the output of the rule inference engine to obtain the control commands for sending and receiving, avoiding the delay caused by online optimization calculations and ensuring that the control commands are generated within a single transmission time slot.

[0027] The adaptive modulation and coding unit dynamically switches the modulation order and forward error correction code rate according to transmit / receive control commands. This unit includes a modulation format selection switch, a coding rate configuration register, and an interleaving depth controller. The modulation format selection switch receives the modulation scheme index from the central coordinating controller and switches the output between QPSK, 16QAM, or π / 2-BPSK modulators via a multiplexer. The coding rate configuration register loads the corresponding LDPC code generation matrix according to the coding scheme index; this matrix is ​​stored in on-chip ROM and supports multiple code rate configurations. The interleaving depth controller dynamically adjusts the number of rows in the bit interleaver based on the coherence time estimate, ensuring that the interleaving period covers an integer multiple of the channel coherence time. Specifically, the interleaver adopts a block interleaving structure; its number of rows is obtained by dividing the coherence time estimate by the symbol duration and rounding up, while the number of columns is fixed at the LDPC code block length. This design ensures that bits in the same codeword experience as independent a channel fading as possible, thereby maximizing time diversity gain.

[0028] The power and beamforming joint control module synchronously adjusts the transmit power level and the beam pointing of the antenna array according to transmit and receive control commands. This module includes a digital predistortion compensator, a power amplifier bias controller, and a phased array beamforming network. The digital predistortion compensator receives the baseband modulation signal and loads the corresponding nonlinear compensation coefficient from a lookup table based on the current power level number. This coefficient is obtained by offline measurement of the AM-AM and AM-PM characteristics of the power amplifier at different operating points. The compensated signal is sent to the power amplifier bias controller, which outputs an analog voltage through a digital-to-analog converter to adjust the gate bias current of the final-stage power amplifier, achieving a step-wise adjustment of the transmit power. The phased array beamforming network consists of a phase shifter array and a weighted summer. Its phase shifter control word is generated by converting the beamforming weight coefficient set provided by the central coordinating controller using the CORDIC algorithm. The CORDIC algorithm iteratively calculates the amplitude and phase representation of the complex weights and outputs the phase control word to each phase shifter, ensuring that the main lobe is precisely aligned with the angular center of the dominant path cluster. This process is completed before the start of each transmission time slot, ensuring strict alignment between the beam pointing and the channel spatial characteristics.

[0029] The feedback information compression encoder is connected to the output of the central coordination controller and is used to compress and encode transmit / receive control commands before transmitting them back to the transmitter via the uplink control channel. This encoder employs a context-adaptive binary arithmetic coding mechanism, where the coding context is dynamically updated by the historical sequence of the current channel state vector. The feedback information includes the modulation and coding scheme index, power level number, and differential coding results of the beamforming weight coefficient set. Differential coding uses the most recently successfully transmitted parameters as a reference, transmitting only the changed parameters, significantly reducing the number of feedback bits. The encoded bit stream is transmitted via the reserved uplink control channel in a single symbol time slot, ensuring that the end-to-end delay of transmit / receive parameter adjustments does not exceed one transmission time interval.

[0030] Furthermore, the system described in this invention synchronously deploys a channel state buffer and a parameter preloading unit at the transmitting end. The channel state buffer stores the historical channel state vectors of the most recent N transmission time slots, where N is a positive integer equal to the maximum allowed retransmission count plus one. This buffer supports cross-time slot channel evolution trend analysis, such as predicting the channel gain trend of multiple future time slots through sliding window linear regression. After receiving feedback information, the parameter preloading unit preloads the modulation and coding parameters, power configuration, and beam weights required for the next transmission time slot into the configuration register of the corresponding hardware module. This preloading operation is executed immediately after the data transmission of the current transmission time slot is completed, ensuring that all parameters are ready at the start of the next transmission time slot, achieving zero-wait handover. In the event of a HARQ retransmission, the system can retrieve the channel state vector of the corresponding original transmission time slot from the buffer, restore consistent transmit and receive parameters, and avoid retransmission failure due to parameter drift.

[0031] In another preferred embodiment of the present invention, an abnormal channel event detector is provided between the channel dynamic sensing front-end and the channel state evolution modeling engine. This detector continuously monitors the abrupt change amplitude of each component in the multidimensional channel feature vector, calculates its first-order difference absolute value, and compares it with a preset abrupt change threshold. When the rate of change of any feature component exceeds this threshold, an emergency sensing mode is immediately triggered. In this mode, the time-frequency joint feature extractor shortens the feature extraction window to the minimum processable length, typically a single symbol period; the multipath component clustering identifier suspends clustering operations and instead uses fixed-angle partitioning for coarse grouping; the channel state evolution modeling engine increases the update frequency of the channel state vector to once per symbol period. Upon receiving the emergency mode flag, the central coordination controller prioritizes link reliability, forcibly employs the lowest-order modulation (e.g., π / 2-BPSK) and the highest-redundancy coding (e.g., LDPC code with a code rate of 1 / 3), and suspends beam scanning operations, maintaining the current beam pointing unchanged. This mechanism ensures that the system can quickly stabilize the communication link and prevent connection interruption when encountering abnormal events such as sudden interference, rapid obstruction, or high-speed intrusion.

[0032] The hardware implementation of the system described in this invention adopts a system-on-a-chip (SoC) architecture. The channel dynamic sensing front-end, channel state evolution modeling engine, and central coordinating controller are integrated into the same baseband processing chip. Low-latency data exchange is achieved through an on-chip high-speed interconnect bus with a bus width of 512 bits and a clock frequency of 500MHz, ensuring that multi-dimensional channel feature vectors can be transmitted completely within a single clock cycle. The adaptive modulation and coding unit and the power and beam joint control module are deployed in the RF integrated circuit, receiving control commands through a dedicated configuration interface using the SPI protocol, supporting a configuration rate of 10Mbps. The feedback information compression encoder is embedded in the media access control layer protocol stack, tightly coupled with the physical layer, sharing the same clock domain to avoid cross-layer synchronization overhead. The entire system is driven by a unified clock source, which is strictly aligned with the wireless frame structure, ensuring that the operation of each module is synchronized at the system frame boundary, meeting the timing constraints of low-latency, high-reliability communication.

[0033] To verify the technical effectiveness of this invention, a specific embodiment and a comparative example are provided for comparative testing. In the embodiment, the system is deployed in the 3.5GHz band with a bandwidth of 100MHz, using a 64-antenna phased array, with a moving speed of 120km / h, and the channel model is the UMA scenario defined in 3GPP TR 38.901. The system operates according to the above scheme. The channel dynamic sensing front-end outputs a multi-dimensional channel feature vector every millisecond, and the channel state evolution modeling engine generates a channel state vector every transmission time slot (0.5ms). The central coordinating controller generates transmit and receive control commands based on these vectors. In the comparative example, a traditional channel estimation scheme based on periodic pilots is used. Pilot overhead accounts for 10% of the total resources, and transmit and receive parameters are adjusted every 5ms, with no forward-looking prediction capability.

[0034] The test results are shown in the table below:

[0035] Performance indicators Example Comparative Example Average spectral efficiency (bps / Hz) 4.82 3.15 Block Error Rate (BLER) 0.008 0.032 Parameter adjustment delay (ms) 0.5 5.0 Pilot overhead percentage (%) 0 10 Standard deviation of throughput fluctuation at high speed 0.31 0.78 Performance indicators Example Comparative Example Average spectral efficiency (bps / Hz) 4.82 3.15

[0036] Experimental data show that, under the same channel conditions, the embodiment significantly outperforms the comparative embodiment in terms of spectral efficiency, link reliability, and response speed, while completely eliminating pilot overhead. Especially in high-speed mobile scenarios, the embodiment exhibits significantly smaller throughput fluctuations, demonstrating its superior ability to track channel dynamics.

[0037] In summary, this invention constructs an intelligent transceiver system that requires no additional pilot overhead, possesses forward-looking adjustment capabilities, and is hardware-friendly by deeply integrating channel dynamics modeling, real-time sensing feedback, and adaptive transceiver control. In engineering practice, this system can effectively cope with complex channel environments such as high-speed movement, sudden interference, and high-density access, providing a highly robust and efficient solution for future wireless communication systems.

Claims

1. A communication intelligent transceiver system based on dynamic channel awareness, characterized in that, include: The channel dynamic sensing front end is deployed after the receiver's radio frequency link and is used to extract multi-dimensional channel features from the baseband signal in real time. A channel state evolution modeling engine, connected to the channel dynamic sensing front end, is used to construct a time series model of the channel state based on the multi-dimensional channel features, and deduce the channel state vector at the current moment and the next transmission time slot. The central coordinating controller is connected to the channel state evolution modeling engine, the adaptive modulation and coding unit, and the power and beam joint control module, respectively, and is used to generate transmit and receive control commands based on the channel state vector. The adaptive modulation and coding unit is used to dynamically switch the modulation order and forward error correction code rate according to the transmit and receive control command; The power and beam joint control module is used to synchronously adjust the transmit power level and the beam pointing of the antenna array according to the transmit and receive control command. A feedback information compression encoder, connected to the output of the central collaborative controller, is used to compress and encode the transmit and receive control commands and then transmit them back to the transmitter via the uplink control channel.

2. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The channel dynamic sensing front-end includes: A time-frequency joint feature extractor is used to extract the amplitude fading slope, phase rotation rate, and time delay spread index of each subcarrier through sliding window Fourier transform and time-domain differentiation. A multipath component clustering identifier is used to group multipath components based on the aforementioned delay spread index using a density peak clustering algorithm, and to identify the dominant path clusters and their angle of arrival distribution centers. The channel noise covariance estimator is used to calculate the local mean of the noise power spectral density based on the received signal residual sequence, and to correct the channel estimation bias by combining the multipath component clustering results. The multidimensional channel features include channel amplitude variation rate, phase jitter intensity, number of multipath clusters, dominant path angle center, and noise covariance matrix.

3. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The channel state evolution modeling engine adopts a two-layer recurrent neural network structure, with the bottom layer being a long short-term memory network unit, used to perform time series modeling of multi-dimensional channel feature vectors. The top layer is a gated loop unit, which is used to fuse the symbol energy distribution characteristics of the bottom layer output and the current transport block to generate a channel state vector consisting of channel gain prediction, coherence time estimate, upper limit of Doppler shift, and channel condition number.

4. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The central coordinating controller integrates a rule-based inference engine and a parameter mapping table. The rule-based inference engine triggers one of the following operations based on the four-dimensional parameter combination of the channel state vector: when the channel gain prediction value is lower than the first threshold and the coherence time estimate value is less than the second threshold, low-order modulation and high-redundancy coding are enabled; when the upper limit of Doppler frequency shift exceeds the third threshold, constant envelope modulation is enabled; when the channel condition number is greater than the fourth threshold, the antenna selection mechanism is activated. The parameter mapping table stores the modulation and coding scheme index, power level number, and beamforming weight coefficient set corresponding to different channel state intervals.

5. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The adaptive modulation and coding unit includes: The modulation format selection switch is used to switch the output of the QPSK, 16QAM, or π / 2-BPSK modulator according to the modulation scheme index. The encoding rate configuration register is used to load the corresponding LDPC code generation matrix; An interleaving depth controller is used to adjust the number of rows of the bit interleaver based on the coherence time estimate, so that the interleaving period covers an integer multiple of the channel coherence time.

6. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The power and beam joint control module includes: A digital predistortion compensator is used to apply nonlinear compensation coefficients according to power level numbering to suppress spectral regeneration; A power amplifier bias controller is used to adjust the gate bias current of the final stage power amplifier by outputting an analog voltage from a digital-to-analog converter. The phased array beamforming network consists of a phase shifter array and a weighted summer. Its phase shifter control word is generated by converting the beamforming weight coefficient set through the CORDIC algorithm so that the main lobe is aligned with the angular center of the dominant path cluster.

7. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The feedback information compression encoder adopts a context-adaptive binary arithmetic coding mechanism, and its coding context is dynamically updated by the historical sequence of the channel state vector. The feedback information includes the differential coding results of the modulation coding scheme index, power level number and beamforming weight coefficient set, and is transmitted uplink within a single symbol time slot.

8. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The transmitter is equipped with a channel state buffer and a parameter preloading unit. The channel state buffer stores the channel state vector history of the most recent N transmission time slots, where N is the maximum number of retransmissions allowed by the system plus one. After receiving feedback information, the parameter preloading unit preloads the modulation and coding parameters, power configuration and beam weight required for the next transmission time slot into the configuration register of the corresponding hardware module.

9. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, An abnormal channel event detector is installed between the channel dynamic sensing front-end and the channel state evolution modeling engine. The abnormal channel event detector monitors the abrupt change amplitude of the multi-dimensional channel feature vector. When the rate of change of any feature component exceeds the preset abrupt change threshold, an emergency sensing mode is triggered, the feature extraction window is shortened to the minimum processable length, and the channel state vector update frequency is increased to once per symbol period. In the emergency sensing mode, the central coordinating controller forces the use of the lowest order modulation and the highest redundancy coding, and suspends beam scanning operation.

10. The intelligent transceiver system for communication based on dynamic channel awareness according to claim 1, characterized in that, The channel dynamic sensing front-end, channel state evolution modeling engine, and central collaborative controller are integrated into the same baseband processing chip and exchange data through an on-chip high-speed interconnect bus. The adaptive modulation and coding unit and the power and beam joint control module are deployed in the radio frequency integrated circuit and receive control commands through a dedicated configuration interface. The entire system is driven by a unified clock source to ensure that the operation of each module is strictly synchronized with the system frame structure.

Images