A dual-mode adaptive communication method and system based on SDR

Abstract

This invention relates to the field of software-defined radio communication technology, and discloses a dual-mode adaptive communication method and system based on SDR. The method includes receiving raw wireless signals and initially determining the protocol family, then activating the corresponding signal parsing engine and channel evaluation engine. The signal parsing engine reconstructs complete communication message units, and the channel evaluation engine quantifies the channel state. Both are input into a dynamic decision network, which learns the implicit correlation rules between message semantic integrity and channel transmission reliability, and outputs a collaborative decision instruction containing a mode priority weight vector. The instruction is ultimately compiled into a hardware configuration parameter sequence, driving the reconfiguration of the RF front-end and baseband unit to achieve adaptive mode switching. This invention achieves cross-layer intelligent decision-making and flexible multi-mode scheduling, improving the reliability and efficiency of communication in complex wireless environments.

Description

Technical Field

[0001] This invention relates to the field of software-defined radio communication technology, specifically to a dual-mode adaptive communication method and system based on SDR. Background Technology

[0002] Current adaptive communication technologies typically rely on real-time monitoring of the physical layer state of the channel. The system determines whether to switch communication modes or adjust transmission parameters based on these preset, explicit physical index thresholds. The core logic of existing solutions remains at the level of ensuring reliable signal transmission over the physical channel, with decisions based on a single, directly measurable channel quality parameter. This single-dimensional decision-making mechanism, based on explicit physical layer indicators, fails to incorporate the ultimate goal of communication transmission—the complete and effective delivery of message units carrying specific service semantics—into the decision-making loop.

[0003] Traditional multi-mode handover mechanisms often employ fixed-threshold decision-making methods, performing hard handovers between multiple available communication modes, choosing one or the other sequentially. This approach treats mode selection within each communication cycle as a deterministic, either-or problem, lacking the flexible scheduling capability for dynamic coordination and priority combination among multiple modes. This rigid handover strategy struggles to cope with the challenges of rapidly changing service requirements and channel conditions in complex wireless environments, easily leading to low resource utilization efficiency or sudden changes in service experience. An adaptive communication method is needed that can deeply understand service semantic requirements and perform fine-grained multi-mode collaborative scheduling. Summary of the Invention

[0004] The purpose of this invention is to provide a dual-mode adaptive communication method and system based on SDR to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a dual-mode adaptive communication method based on SDR, the method comprising:

[0006] Receive raw wireless signal streams from the radio frequency front end and obtain a preliminary determination of the communication protocol family category based on the raw signal streams;

[0007] Based on the initially determined communication protocol family category, the corresponding signal analysis engine and channel evaluation engine are activated. The signal analysis engine performs in-depth analysis of the signal stream to reconstruct the complete communication message unit, and the channel evaluation engine quantifies and characterizes the noise floor, multipath delay, and interference spectrum during signal transmission.

[0008] The communication message unit and the quantized channel state are input together into a dynamic decision network, which is configured to learn the implicit association rules between message semantic integrity and channel transmission reliability.

[0009] Based on the learned implicit association rules, the dynamic decision network outputs a collaborative decision instruction containing a weight vector, which defines the priority order of the primary communication mode and alternative communication modes to be used in the next communication cycle.

[0010] The collaborative decision-making instructions are compiled into a sequence of configuration parameters that can be recognized by the software-defined radio hardware, driving the RF front-end and baseband processing unit to reconfigure according to the new parameter sequence, thereby completing the adaptive switching of communication modes.

[0011] Preferably, the receiving of the raw wireless signal stream from the radio frequency front-end, and the acquisition of the preliminarily determined communication protocol family category based on the raw signal stream, includes:

[0012] The original wireless signal stream is subjected to joint detection of signal energy distribution and protocol frame start symbol to locate potential valid communication data segments in the signal stream;

[0013] Blind parsing is performed on the potential valid communication data segments. During the blind parsing process, candidate modulation patterns, symbol rate features, and frame structure features are extracted simultaneously. Based on the extraction results, multiple rounds of fuzzy matching are performed in a preset protocol feature library to preliminarily determine the communication protocol family category to which the current signal stream belongs. Specifically, this includes:

[0014] Sliding window energy detection is performed on the potential valid communication data segments to identify signal pulse intervals with energy exceeding the silence threshold, and the time-domain envelope waveform of each signal pulse interval is extracted.

[0015] A multi-scale wavelet transform is performed on the time-domain envelope waveform to separate the periodic component representing the symbol rate and the aperiodic component representing the modulation pattern from the wavelet transform coefficients.

[0016] Calculate the constellation point distribution of the aperiodic component in the complex plane, and perform point-by-point matching of the constellation point distribution with the known modulation constellation templates stored in the protocol feature library to generate a modulation pattern matching degree list;

[0017] Simultaneously, autocorrelation analysis is performed on the periodic components to extract the main peak interval as a candidate symbol period. The candidate symbol period is then compared with the standard symbol periods of the protocol stored in the protocol feature library to generate a symbol period matching degree list.

[0018] By combining the modulation pattern matching list and the symbol period matching list, a comprehensive matching score is calculated for each candidate protocol, and protocols with comprehensive matching scores exceeding the judgment threshold are selected as the communication protocol family category for the preliminary determination.

[0019] Preferably, activating the corresponding signal parsing engine and channel evaluation engine includes:

[0020] Based on the preliminarily determined communication protocol family category, the corresponding protocol syntax rule library and frame synchronization word template are loaded from the engine resource pool into the working memory of the signal parsing engine;

[0021] The signal parsing engine performs precise frame synchronization positioning of the signal stream based on the loaded frame synchronization word template, and performs syntax parsing of the synchronized data bit stream based on the loaded protocol syntax rule library, decomposing the message header, payload and check fields.

[0022] The channel evaluation engine captures background noise samples before and after the signal pulse interval while parsing the signal, and estimates the power spectral density of the background noise samples to obtain a quantitative characterization of the noise floor.

[0023] The channel evaluation engine further estimates the quantitative characterization of the multipath delay by performing broadening analysis on the temporal correlation peak of the frame synchronization word template.

[0024] The channel evaluation engine also identifies unwanted high-power frequency points by monitoring the power of frequency bands outside the signal pulse interval, and generates a quantitative representation of the interference spectrum.

[0025] Preferably, the step of inputting the communication message unit and the quantized channel state into a dynamic decision network includes:

[0026] The communication message units are classified according to semantic importance, and different semantic weights are assigned to the message header field, control payload field and data payload field according to the field priority defined in the protocol syntax rule base.

[0027] Availability mapping is performed on the quantized channel state, and the quantized representation of the noise floor, the quantized representation of the multipath delay, and the quantized representation of the interference spectrum are respectively mapped to negative factor levels that affect the reliability of signal transmission.

[0028] A dynamic decision network is constructed with semantic weight vector and negative factor level vector as input. The dynamic decision network contains multi-layer cross-sensing units to calculate the probability estimate of successful transmission of messages with different semantic weights under a specific combination of negative factors.

[0029] Preferably, the dynamic decision network outputs a collaborative decision instruction containing a weight vector, including:

[0030] The dynamic decision network inputs the calculated probability estimate of successful transmission into a strategy optimization layer, which has a variety of preset communication mode switching strategies built in.

[0031] The strategy optimization layer evaluates the expected change in the estimated success transmission probability when maintaining the current communication mode and applying each preset communication mode switching strategy in the next communication cycle.

[0032] Select a target strategy that maximizes the expected change in the estimated probability of successful transmission, and generate a corresponding weight vector based on the target strategy. The weight vector includes a determination identifier and parameter configuration for the primary communication mode, as well as preparation instructions for alternative communication modes.

[0033] Preferably, compiling the collaborative decision-making instructions into a software-defined radio hardware-recognizable sequence of configuration parameters includes:

[0034] Parse the weight vector in the collaborative decision-making instruction and extract the standard hardware configuration template of the primary communication mode corresponding to the determination identifier;

[0035] Based on the real-time channel parameters fed back by the current signal analysis engine, the adjustable parameters such as center frequency, bandwidth, and transmit power in the standard hardware configuration template are fine-tuned to generate an accurate parameter set adapted to the current channel.

[0036] The precise parameter set is encapsulated into a continuous sequence of configuration parameters according to the data format and order specified by the software-defined radio hardware driver interface, and the end of the configuration parameter sequence contains a checksum.

[0037] Preferably, the driving RF front-end and the baseband processing unit are reconfigured according to a new parameter sequence, including:

[0038] The configuration parameter sequence is sent to the programmable register group of the RF front end via the control bus. The programmable register group updates the local oscillator frequency, filter bandwidth and power amplifier bias voltage according to the received data.

[0039] Synchronously, the parameter portion of the configuration parameter sequence related to baseband processing is sent to the programmable logic array of the baseband processing unit, and the programmable logic array reloads the corresponding digital filter coefficients, modem cores and codec cores according to the received parameters;

[0040] After both the programmable register group and the programmable logic array have been loaded, a synchronization enable signal is sent, causing the RF front-end and the baseband processing unit to start working simultaneously under the new configuration.

[0041] Preferably, the construction steps of the signal parsing engine include:

[0042] The communication protocol family categories initially determined are subjected to protocol feature extraction, and the corresponding syntax rule set and frame synchronization word mode are obtained from the protocol feature library;

[0043] Configure the state machine of the protocol parser according to the set of syntax rules, wherein the state machine defines the parsing order and verification rules of message fields;

[0044] The correlator parameters of the frame detector are configured according to the frame synchronization word mode, and the correlator parameters include the coefficients of the matched filter and the synchronization threshold;

[0045] The configured protocol parser and frame detector are loaded into the execution environment of the signal parsing engine to establish a processing pipeline from signal stream input to communication message unit output.

[0046] Preferably, the construction steps of the channel evaluation engine include:

[0047] Background noise is sampled from the original wireless signal stream to establish a statistical model of the noise floor;

[0048] Configure a multipath delay estimation module, use a generalized cross-correlation algorithm to calculate the broadening of the signal correlation peak, and initialize the delay expansion parameters;

[0049] Configure the interference spectrum monitoring module, set the frequency band scanning range and power threshold, to identify unwanted interference signals;

[0050] The noise floor statistical model, multipath delay estimation module, and interference spectrum monitoring module are integrated into the channel evaluation engine to form an output interface for quantitative representation of channel state.

[0051] Preferably, when the processor executes the computer program, it implements the steps of the SDR-based dual-mode adaptive communication method as described in any of the above-described embodiments.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] By learning the implicit correlation rules between message semantic integrity and channel transmission reliability through a dynamic decision network, the decision-making mechanism evolves from relying solely on explicit physical layer indicators to a joint optimization process that integrates both physical layer state and application layer semantic state. The communication message units reconstructed by the signal parsing engine provide business semantic-level input for decision-making, enabling the system to identify the differentiated requirements of different service data for transmission reliability. The quantitative representation provided by the channel evaluation engine characterizes the constraints of the physical environment. Through continuous learning of these two types of heterogeneous information, the dynamic decision network constructs a non-obvious correlation model between them. This ensures that the final handover decision not only pursues channel-level reliability but also simultaneously guarantees the effectiveness of business semantics, achieving intelligent decision-making through cross-layer optimization and avoiding the transmission risk of critical signaling or data semantic corruption even with good physical layer indicators alone.

[0054] The priority order of primary and alternative communication modes is defined by learning-generated weight vectors, changing the traditional single-mode occupancy method of hard handover. Cooperative decision-making instructions plan resource allocation and mode usage within a communication cycle as a flexibly configurable cooperative task. The weight vectors indicate the utility priority of different modes under the current channel and service combination, enabling the system to implement refined strategies such as weighted network reconfiguration of primary and backup modes and priority-based scheduling of service flows, supported by the rapid reconfiguration capabilities of software-defined radio. This endows the system with the ability to perform smooth and gradual mode migration and load sharing in the face of sudden interference or service changes, improving the overall robustness of communication and the flexibility of resource utilization, and overcoming the service interruption or sharp quality degradation problems that may occur during hard handover. Attached Figure Description

[0055] Figure 1 This is a schematic diagram illustrating the working principle of the SDR-based dual-mode adaptive communication method described in this invention.

[0056] Figure 2 The flowchart is constructed as input for the dynamic decision network;

[0057] Figure 3 Output a flowchart for policy optimization for dynamic decision networks;

[0058] Figure 4 A scatter plot showing the correlation between parameter weights and configuration time;

[0059] Figure 5 A dual-axis composite chart for SDR interference spectrum monitoring. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1This invention provides a dual-mode adaptive communication method based on SDR. The method includes: receiving a raw wireless signal stream from a radio frequency (RF) front-end; obtaining a preliminary determination of the communication protocol family category based on the raw signal stream; activating the corresponding signal parsing engine and channel evaluation engine based on the preliminary determination of the communication protocol family category; the signal parsing engine performing deep parsing of the signal stream to reconstruct complete communication message units; and the channel evaluation engine quantifying and characterizing the noise floor, multipath delay, and interference spectrum during signal transmission. The communication message units and the quantized channel state are input into a dynamic decision network, which is configured to learn implicit association rules between message semantic integrity and channel transmission reliability. Based on the learned implicit association rules, the dynamic decision network outputs a collaborative decision instruction containing a weight vector, which defines the priority order of the primary and alternative communication modes to be used in the next communication cycle. The collaborative decision instruction is compiled into a configuration parameter sequence recognizable by the software-defined radio hardware, driving the RF front-end and baseband processing unit to reconfigure according to the new parameter sequence, thereby completing the adaptive switching of communication modes.

[0062] Example 1: The system receives a raw wireless signal stream from an RF front-end. Based on the raw signal stream, a preliminary classification of the communication protocol family is determined. This includes joint detection of signal energy distribution and protocol frame start symbols in the raw wireless signal stream to locate potential valid communication data segments. Blind parsing is performed on these potential valid communication data segments, simultaneously extracting candidate modulation patterns, symbol rate features, and frame structure features. Based on the extraction results, multiple rounds of fuzzy matching are performed in a pre-defined protocol feature library to preliminarily determine the communication protocol family to which the current signal stream belongs. Sliding window energy detection is performed on the potential valid communication data segments to identify signal pulse intervals with energy exceeding a silence threshold. The time-domain envelope waveform of each signal pulse interval is extracted, and multi-scale wavelet transform is performed on the time-domain envelope waveform. The periodic component representing the symbol rate and the aperiodic component representing the modulation pattern are separated from the wavelet transform coefficients. The constellation point distribution of aperiodic components in the complex plane is calculated, and the constellation point distribution is matched point-by-point with known modulation constellation templates stored in the protocol feature library to generate a modulation pattern matching degree list. Simultaneously, autocorrelation analysis is performed on the periodic components to extract the main peak interval as candidate symbol periods. These candidate symbol periods are compared with the standard symbol periods of protocols stored in the protocol feature library to generate a symbol period matching degree list. The modulation pattern matching degree list and the symbol period matching degree list are fused, and a comprehensive matching score is calculated for each candidate protocol. Protocols with comprehensive matching scores exceeding a judgment threshold are selected as the preliminary communication protocol family category.

[0063] In practice, receiving the raw wireless signal stream from the RF front-end is a continuous process. The RF front-end converts the captured analog signal into a digital signal stream for subsequent processing. Determining the initial communication protocol family category based on the raw signal stream involves two main stages: signal preprocessing and feature matching. Joint detection of signal energy distribution and protocol frame start symbols is performed on the raw wireless signal stream to locate potential valid communication data segments. This process involves continuously calculating the energy of the digital signal stream and comparing it with a preset energy threshold. When a continuous signal segment with energy exceeding a silence threshold is detected, it is marked as a potential valid communication data segment.

[0064] In some embodiments, blind parsing is performed on potential valid communication data segments. During blind parsing, candidate modulation patterns, symbol rate features, and frame structure features are extracted simultaneously. Based on the extraction results, multiple rounds of fuzzy matching are performed in a preset protocol feature library to preliminarily determine the communication protocol family category to which the current signal stream belongs. Sliding window energy detection is performed on potential valid communication data segments to identify signal pulse intervals with energy exceeding a silence threshold. The time-domain envelope waveform of each signal pulse interval is extracted, and the width of the sliding window is set according to the lowest detectable symbol rate. Multi-scale wavelet transform is performed on the time-domain envelope waveform to separate the periodic component representing the symbol rate and the aperiodic component representing the modulation pattern from the wavelet transform coefficients. The wavelet transform scale covers multiple frequency bands from low to high.

[0065] In practice, the constellation point distribution of the aperiodic components in the complex plane is calculated, and this distribution is then matched point-by-point with known modulation constellation templates stored in the protocol feature library to generate a modulation pattern matching degree list. The known modulation constellation templates stored in the protocol feature library include ideal constellation diagrams for various modulation patterns such as binary phase shift keying, quadrature phase shift keying, and hexadecimal quadrature amplitude modulation. Simultaneously, autocorrelation analysis is performed on the periodic components, and the main peak interval is extracted as candidate symbol periods. These candidate symbol periods are then compared with the standard symbol periods of the protocols stored in the protocol feature library to generate a symbol period matching degree list. The standard symbol periods stored in the protocol feature library include, for example, the symbol periods of the Global System for Mobile Communications (GSMA) standard and the symbol periods of the Long Term Evolution (LTE) standard.

[0066] It is understandable that by fusing the modulation pattern matching list and the symbol period matching list, a comprehensive matching score is calculated for each candidate protocol, and protocols with comprehensive matching scores exceeding a judgment threshold are selected as the initial classification of communication protocol families. The formula for calculating the comprehensive matching score is:

[0067]

[0068] in: This represents the overall matching score. This represents the modulation pattern matching score. Indicates the symbol period matching score. and These are the pre-defined modulation pattern matching weights and symbol period matching weights. Modulation pattern matching score. The sign period matching score reflects the Euclidean distance between the measured constellation point distribution and the template constellation diagram. This reflects the relative error between the candidate symbol period and the standard symbol period. In some embodiments, when the binary phase-shift keying modulation matching score... Achieved a symbol periodicity score of 0.95 When the score reaches 0.98, its overall matching score will be higher than the combination of a quadrature phase shift keying modulation matching score of 0.90 and a symbol period matching score of 0.85, thus making it more likely to be identified as the corresponding protocol family.

[0069] Optionally, the standard symbol period of the protocol stored in the protocol feature library is a numerical range rather than a single fixed value, to accommodate clock tolerances in actual communication. Each entry in the symbol period matching list corresponds to a candidate protocol family and its matching confidence. The matching process is iterative. The first round of matching is based on the most significant features, such as symbol rate, for screening. Subsequent rounds combine more refined modulation pattern features for confirmation. Finally, one or more communication protocol family categories with the highest matching degree and their corresponding matching scores are output as preliminary judgment results.

[0070] Example 2: See Figure 2The corresponding signal parsing engine and channel evaluation engine are activated. Based on the initially determined communication protocol family category, the corresponding protocol syntax rule library and frame synchronization word template are loaded from the engine resource pool into the working memory of the signal parsing engine. The signal parsing engine performs precise frame synchronization positioning of the signal stream based on the loaded frame synchronization word template, and performs syntax parsing of the synchronized data bit stream based on the loaded protocol syntax rule library, decomposing the message header, payload, and checksum fields. While parsing the signal, the channel evaluation engine captures background noise samples before and after the signal pulse interval and estimates the power spectral density of the background noise samples to obtain a quantitative characterization of the noise floor. The channel evaluation engine further estimates the quantitative characterization of multipath delay by broadening the time-domain correlation peak of the frame synchronization word template. The channel evaluation engine also identifies unwanted high-power frequency points by monitoring the frequency band power outside the signal pulse interval and generates a quantitative characterization of the interference spectrum. The communication message units and the quantized channel state are input into a dynamic decision network. The semantic importance of the communication message units is classified, and different semantic weights are assigned to the message header, control payload, and data payload fields based on the field priorities defined in the protocol syntax rule base. Availability mapping is performed on the quantized channel state, mapping the quantized representations of the noise floor, multipath delay, and interference spectrum to negative factor levels affecting signal transmission reliability. A dynamic decision network is constructed, taking the semantic weight vector and negative factor level vector as inputs. This dynamic decision network contains multiple layers of cross-sensing units to calculate the probability estimate of successful transmission of messages with different semantic weights under specific combinations of negative factors.

[0071] In practice, the activation of the corresponding signal parsing engine and channel evaluation engine is based on the initially determined communication protocol family category. According to this category, the corresponding protocol syntax rule library and frame synchronization word template are loaded from the engine resource pool into the working memory of the signal parsing engine. The protocol syntax rule library is stored in Extensible Markup Language (EXPLAIN) file format, defining the message structure, field lengths, encoding methods, and checksum algorithms. The signal parsing engine performs precise frame synchronization positioning of the signal stream based on the loaded frame synchronization word template. Frame synchronization positioning is achieved by calculating the cross-correlation function between the input signal and the frame synchronization word template. When the cross-correlation value exceeds a preset threshold, it is determined to be the start of a frame. The synchronized data bit stream is then parsed according to the loaded protocol syntax rule library, extracting the message header, payload, and checksum fields. The parsing process follows the state transition diagram defined in the protocol syntax rule library.

[0072] In some embodiments, the channel evaluation engine operates in parallel with signal parsing, capturing background noise samples before and after the signal pulse interval, which is identified by the previous blind parsing step. It then estimates the power spectral density of the background noise samples to obtain a quantitative characterization of the noise floor. The power spectral density estimation employs the Welch method. The channel evaluation engine further estimates the quantitative characterization of multipath delay by performing broadening analysis on the time-domain correlation peaks of the frame synchronization word template. It analyzes the time width of the correlation peaks after exceeding a certain amplitude threshold, which is proportional to the multipath delay expansion. The channel evaluation engine also identifies unwanted high-power frequency points by monitoring the power of frequency bands outside the signal pulse interval, generating a quantitative characterization of the interference spectrum. The monitoring process converts the time-domain signal into a frequency-domain power spectrum using a Fast Fourier Transform.

[0073] In practical implementation, the communication message units and the quantized channel state are input into a dynamic decision network. This requires semantic importance classification of the communication message units, assigning different semantic weights to the message header, control payload, and data payload fields based on field priorities defined in the protocol syntax rule base. For example, in the parsing of a wireless LAN protocol stack, the frame control field in the message header is assigned the highest semantic weight of 0.8, while the data payload field is assigned a lower semantic weight of 0.2. Availability mapping is then performed on the quantized channel state, mapping the quantized representations of the noise floor, multipath delay, and interference spectrum to negative factor levels that affect signal transmission reliability. The negative factor level is a value between 0 and 1, where 0 represents no impact and 1 represents a severe impact. Essentially, a dynamic decision network is constructed with semantic weight vectors and negative factor level vectors as inputs. This dynamic decision network contains multiple layers of cross-sensing units used to calculate the probability estimate of successful transmission of messages with different semantic weights under specific combinations of negative factors. The multilayer cross-sensing unit performs a nonlinear transformation on the input vector. Its computation involves concatenating the semantic weight vector with the negative factor ranking vector, and then passing it through multiple fully connected layers and activation functions. A simplified intermediate layer computation can be represented as:

[0074]

[0075] in: Represents the hidden layer feature vector. This represents the Sigmoid activation function. Represents the weight matrix. Represents the semantic weight vector. This represents the channel negative factor vector, which is composed of the noise floor negative factor level, the multipath delay negative factor level, and the interference spectrum negative factor level. This represents the bias vector. The final output layer will take the hidden layer feature vectors. It is mapped to a scalar representing an estimate of the probability of successful transmission.

[0076] In some embodiments, when the semantic weight vector indicates a high-priority control message, and the channel negative factor vector indicates a noise floor level of 0.1 and a multipath delay level of 0.3, the estimated probability of successful transmission calculated by the dynamic decision network may be 0.95. Optionally, the interference spectrum quantization representation generated by the channel evaluation engine can be a list containing the center frequencies and power values ​​of multiple interference frequencies. The availability mapping process compares the power of each interference frequency with the current signal power; if the ratio exceeds a threshold, a high-level negative factor is generated. Optionally, the field priorities defined in the protocol syntax rule base are statically configured, but can also be dynamically adjusted by a higher-level strategy according to the communication scenario. The dynamically adjusted information is fed back to the semantic importance classification module in real time.

[0077] Example 3: See Figure 3 The dynamic decision network outputs a collaborative decision instruction containing a weight vector. The calculated successful transmission probability estimate is input into a policy optimization layer. This layer incorporates multiple preset communication mode switching strategies. The strategy optimization layer evaluates the expected change in the successful transmission probability estimate during the next communication cycle, considering both maintaining the current communication mode and applying each preset communication mode switching strategy. A target strategy that maximizes the expected change in the successful transmission probability estimate is selected, and a corresponding weight vector is generated based on this strategy. This weight vector contains a identifier for the primary communication mode and its parameter configuration, as well as preparation instructions for alternative communication modes. The collaborative decision instruction is compiled into a configuration parameter sequence recognizable by the software-defined radio hardware. The weight vector in the collaborative decision instruction is parsed, and the standard hardware configuration template for the primary communication mode corresponding to the identifier is extracted. Based on the real-time channel parameters fed back by the current signal analysis engine, the adjustable parameters of the center frequency, bandwidth, and transmit power in the standard hardware configuration template are fine-tuned to generate a precise parameter set adapted to the current channel. The precise parameter set is encapsulated into a continuous configuration parameter sequence according to the data format and order specified by the software-defined radio hardware driver interface, with a checksum at the end of the configuration parameter sequence.

[0078] In practical implementation, the dynamic decision network outputs a collaborative decision instruction containing a weight vector. The dynamic decision network then inputs the calculated successful transmission probability estimate into a policy optimization layer. This layer incorporates multiple preset communication mode switching strategies, including switching from orthogonal frequency division multiplexing (OFDM) to direct sequence spread spectrum (DSSS), from high-frequency bands to low-frequency bands, and from high-order modulation schemes to low-order modulation schemes, as well as various combinations thereof. The policy optimization layer evaluates the expected change in the successful transmission probability estimate in the next communication cycle, considering both maintaining the current communication mode and applying each preset communication mode switching strategy. This evaluation is based on a historical channel state transition probability and policy execution delay model. The strategy execution delay model constructs an execution time delay library for each preset communication mode switching strategy by analyzing the time interval data from the completion of the compilation of the collaborative decision instruction to the full readiness of the RF front-end and the baseband processing unit in historical operations. The strategy execution delay model uses the loading history of the programmable register group and the programmable logic array, combined with the trigger timing of the synchronization enable signal, to quantify the fixed delay and variable delay components in the strategy execution process, and matches these delay data with the current hardware state parameters. Thus, when the strategy optimization layer evaluates the expected change, a time correction factor is introduced for each switching strategy to ensure that the calculation of the expected change in the successful transmission probability estimate includes the actual time overhead of mode switching.

[0079] In some embodiments, a target strategy is selected that maximizes the expected change in the estimated probability of successful transmission, and a corresponding weight vector is generated based on the target strategy. The weight vector includes a determination identifier for the primary communication mode and its parameter configuration, as well as a preparation indication for alternative communication modes. The determination identifier is an index pointing to a specific communication mode entry in the protocol feature library. The parameter configuration includes the center frequency, bandwidth, modulation scheme, and channel coding rate. The preparation indication includes the index of the alternative communication modes and the switching trigger condition. It can be understood that compiling the cooperative decision instructions into a sequence of configuration parameters recognizable by the software-defined radio hardware requires parsing the weight vector in the cooperative decision instructions and extracting the standard hardware configuration template for the primary communication mode corresponding to the determination identifier. The standard hardware configuration template is stored in the form of an Extensible Markup Language file, defining the initial register values ​​of each programmable module in the RF front-end and baseband processing unit.

[0080] In practical implementation, based on the real-time channel parameters fed back by the current signal analysis engine, the adjustable parameters of center frequency, bandwidth, and transmit power in the standard hardware configuration template are fine-tuned to generate a precise parameter set adapted to the current channel. The fine-tuning calculation is based on the quantization representation provided by the channel evaluation engine. For example, if the interference spectrum indicated by the channel evaluation engine's quantization representation shows strong interference at 2.412 GHz, and the center frequency of the standard hardware configuration template is 2.412 GHz, the fine-tuning calculation will shift the center frequency to 2.437 GHz. The formula for calculating the expected change is:

[0081]

[0082] in: Indicates the application of the first The expected change in the probability of successful transmission after a preset communication mode switching strategy. This indicates that the strategy optimization layer predicts the application of the first term in the next communication cycle. Estimated success rate of transmission after implementing this strategy This represents the estimated probability of successful transmission calculated by the dynamic decision-making network. The policy optimization layer calculates the probability of successful transmission for all preset policies. And select to make Maximum value strategy As the output of the target policy. In one example, maintaining the current orthogonal frequency division multiplexing mode. The value is 0.72, switching to direct sequence spread spectrum mode. The value is 0.90, while switching to a lower-order quadrature amplitude modulation mode. If it is 0.85, then the direct sequence spread spectrum mode corresponds to... Its value is 0.18, which is the largest, so it was selected as the target strategy.

[0083] Optionally, the precise parameter set is encapsulated into a continuous sequence of configuration parameters according to the data format and order specified by the software-defined radio hardware driver interface. The end of the configuration parameter sequence includes a checksum. The data format is typically a binary byte stream, with the order being: RF front-end configuration segment, baseband processing unit configuration segment, and checksum. In some embodiments, if the standard hardware configuration template defines the transmit power as 20 mW / dB, but the path loss calculated based on real-time channel parameters increases by 5 dB, then the fine-tuning calculation will adjust the transmit power parameter in the precise parameter set to 25 mW / dB.

[0084] Example 4: The RF front-end and baseband processing unit are reconfigured according to a new parameter sequence. The configuration parameter sequence is sent to the programmable register group of the RF front-end via the control bus. The programmable register group updates the local oscillator frequency, filter bandwidth, and power amplifier bias voltage based on the received data. Simultaneously, the parameters related to baseband processing in the configuration parameter sequence are sent to the programmable logic array of the baseband processing unit. The programmable logic array reloads the corresponding digital filter coefficients, modem cores, and codec cores based on the received parameters. After both the programmable register group and the programmable logic array have completed loading, a synchronization enable signal is sent, causing the RF front-end and baseband processing unit to start working simultaneously under the new configuration.

[0085] In practice, reconfiguring the RF front-end and baseband processing unit according to the new parameter sequence is a hardware control process. The configuration parameter sequence is sent to the programmable register group of the RF front-end through the control bus. The control bus adopts a serial peripheral interface protocol. The configuration parameter sequence contains a series of address-data pairs. The programmable register group updates the local oscillator frequency, filter bandwidth, and power amplifier bias voltage according to the received data. The local oscillator frequency is updated by writing to the divider ratio register of the frequency synthesizer. The filter bandwidth is updated by changing the capacitor array value of the adjustable filter. The power amplifier bias voltage is updated by the output voltage value of the digital-to-analog converter. In some embodiments, the parameter portion of the configuration parameter sequence related to baseband processing is synchronously sent to the programmable logic array of the baseband processing unit. The programmable logic array of the baseband processing unit reloads the corresponding digital filter coefficients, modem cores, and codec cores according to the received parameters. The digital filter coefficients are stored in the block random access memory of the field programmable gate array in the form of a coefficient table. The modem cores and codec cores exist in the form of hardware description language modules and are loaded through partial reconfiguration technology. After the programmable register group and the programmable logic array have been loaded, a synchronization enable signal is sent so that the RF front-end and the baseband processing unit start working simultaneously under the new configuration. The synchronization enable signal is a global hardware interrupt signal that triggers the RF front-end and the baseband processing unit to switch from the standby state to the working state, as shown in Table 1.

[0086] Table 1: Configuration Parameter Sequence Table

[0087]

[0088] In practical implementation, the transmission of configuration parameter sequences follows a strict timing sequence. After sending each address-data pair, the control bus waits for an acknowledgment signal from the programmable register group or programmable logic array. The transmission time of the entire sequence is proportional to the number of parameters. Optionally, the timing of the synchronization enable signal transmission is controlled by a hardware state machine. The hardware state machine monitors the configuration completion flags of the programmable register group and the programmable logic array. When both flags are set, the hardware state machine automatically sends a synchronization enable signal. It can be understood that updating the local oscillator frequency of the programmable register group involves the locking time of the phase-locked loop. After updating the local oscillator frequency, a delay is needed before sending the synchronization enable signal; this delay time is determined by the loop bandwidth of the phase-locked loop. The settling time after the local oscillator frequency update... It can be estimated using the following formula:

[0089]

[0090] in: Indicates the frequency lock time. This indicates the division ratio of the frequency synthesizer. This represents the loop bandwidth of the phase-locked loop. The system wait time after the programmable register set updates the divider ratio. The lock status is checked again before the synchronization enable signal is allowed to be sent. In some embodiments, the parameter portion of the configuration parameter sequence related to baseband processing is sent in the form of data packets. The data packets include a header, parameter body, and checksum. The programmable logic array receives the data packets, parses them, and loads them into the corresponding configuration memory. Optionally, the configuration of the programmable register group and the programmable logic array can be performed step by step, but the synchronization enable signal ensures that the start times of their operation are strictly aligned to avoid a situation where the RF front-end has already transmitted while the baseband processing unit is not ready.

[0091] See Figure 4 This is a scatter plot showing the correlation between parameter weights and configuration time. Higher parameter weights result in shorter configuration times (e.g., a parameter with a weight of 0.35 takes approximately 12ms; a parameter with a weight of 0.05 takes approximately 25ms). The slope of the fitted line is negative, further quantifying the trend of "increasing weight, decreasing time." This type of plot is commonly used in system parameter optimization analysis (such as the configuration process of an SDR communication system) to help determine the relationship between parameter priority and configuration efficiency, providing data support for simplifying the configuration process and shortening time.

[0092] Example 5: The construction steps of the signal parsing engine include extracting protocol features from the initially determined communication protocol family categories, obtaining the corresponding syntax rule set and frame synchronization word pattern from the protocol feature library, configuring the state machine of the protocol parser according to the syntax rule set, and defining the parsing order and verification rules of message fields. The correlator parameters of the frame detector are configured according to the frame synchronization word pattern, including the coefficients of the matched filter and the synchronization threshold. The configured protocol parser and frame detector are loaded into the execution environment of the signal parsing engine, establishing a processing pipeline from signal stream input to communication message unit output. The construction steps of the channel evaluation engine include sampling background noise from the original wireless signal stream, establishing a statistical model of the noise floor, configuring a multipath delay estimation module, using a generalized cross-correlation algorithm to calculate the signal correlation peak broadening, and initializing the delay extension parameters. An interference spectrum monitoring module is configured, setting the frequency band scanning range and power threshold to identify unwanted interference signals. The noise floor statistical model, multipath delay estimation module, and interference spectrum monitoring module are integrated into the channel evaluation engine to form an output interface for quantified channel state representation.

[0093] In practical implementation, the construction steps of the signal parsing engine include extracting protocol features from the initially determined communication protocol family categories, obtaining the corresponding syntax rule set and frame synchronization word pattern from the protocol feature library. The protocol feature library is a knowledge base stored in a local or remote database. The syntax rule set defines the complete structure of the message, the bit length of each field, the encoding method, and the cyclic redundancy check algorithm in an Extensible Markup Language (EXPLAIN) file format. The frame synchronization word pattern is stored in the form of a binary sequence. The state machine of the protocol parser is configured according to the syntax rule set. The state machine defines the parsing order and verification rules of the message fields. For example, when parsing a wireless LAN data frame, the state machine sequentially defines the state transition path for parsing the preamble, parsing the frame header, parsing the media access control address, parsing the sequence control field, parsing the payload, and parsing the frame check sequence, as well as the field verification logic in each state. In some embodiments, the correlator parameters of the frame detector are configured according to the frame synchronization word pattern. The correlator parameters include the coefficients of the matched filter and the synchronization threshold. The coefficients of the matched filter are directly generated from the binary sequence of the frame synchronization word pattern through a specific mapping rule. The synchronization threshold is dynamically calculated based on the length of the frame synchronization word pattern and the expected signal-to-noise ratio. The configured protocol parser and frame detector are loaded into the execution environment of the signal parsing engine. The execution environment is a software container or lightweight virtual machine with real-time processing capabilities. A processing pipeline is established from the input of the signal stream to the output of the communication message unit. The processing pipeline connects multiple processing modules such as frame detection, bit synchronization, field parsing and verification with a first-in-first-out buffer.

[0094] In practical implementation, the construction steps of the channel evaluation engine include: sampling background noise from the original wireless signal stream; establishing a statistical model of the noise floor; collecting multiple fixed-length noise samples during known periods of no signal transmission; and using the statistical model to describe the probability distribution characteristics of the noise amplitude, such as establishing a Gaussian distribution model of the noise samples and recording their mean and variance. A multipath delay estimation module is configured, employing a generalized cross-correlation algorithm to calculate the broadening of the signal correlation peak, initializing the delay spread parameters, and performing cross-correlation operations between the received signal and a locally known reference signal. The multipath delay spread value is estimated by analyzing the width of the correlation peak exceeding a specific threshold. The delay spread parameters include the average additional delay and the root mean square delay spread. Configure the interference spectrum monitoring module, setting the frequency band scanning range and power threshold to identify unwanted interference signals. The frequency band scanning range is set according to the frequency bands commonly used by the current communication protocol family, for example, 2.4 GHz to 2.5 GHz for wireless LAN protocols. The power threshold is set to be a fixed decibel higher than the average noise power calculated by the noise basis statistical model, such as 10 decibels higher. Essentially, the noise basis statistical model, multipath delay estimation module, and interference spectrum monitoring module are integrated into the channel evaluation engine to form the output interface for channel state quantification. The integration process involves allocating independent computing threads or hardware acceleration units to each module and defining a unified data structure to aggregate the output results of the three modules. The output of the noise basis statistical model is the noise power spectral density curve, the output of the multipath delay estimation module is the root mean square delay spread value, and the output of the interference spectrum monitoring module is a list of interferences containing frequency-power pairs. The output interface for channel state quantification encapsulates these data into a composite data structure. Signal-to-noise ratio (SNR) estimate. The calculation can be used as an intermediate step within the channel evaluation engine, and its formula is:

[0095]

[0096] in: This represents the estimated signal-to-noise ratio, expressed in decibels. This represents the average signal power calculated based on the parsed effective signal segments. This represents the average noise power calculated by the noise basis statistical model. Optionally, the frame synchronization word mode may include multiple variants to adapt to different transmission rates. Therefore, when configuring the correlator parameters of the frame detector, multiple matched filter coefficient sets need to be loaded simultaneously and automatically selected based on signal characteristics at runtime. In some embodiments, the verification rules defined in the protocol syntax rule set include parity check, cyclic redundancy check, or checksum. When the state machine of the protocol parser parses the verification field, it calls the corresponding verification algorithm for verification. If the verification fails, an error handling state is triggered. Optionally, the interference spectrum monitoring module identifies undesirable high-power frequency points. In addition to recording their frequency and power, it can further analyze their duty cycle or modulation characteristics. This additional information is supplemented into the quantitative characterization of the interference spectrum. It is understood that the reference signal used in the generalized cross-correlation algorithm of the multipath delay estimation module is usually the standard frame synchronization word sequence corresponding to the current protocol or a known training sequence.

[0097] See Figure 5 This is a dual-axis composite chart for SDR interference spectrum monitoring. There are two instances where the interference power significantly exceeds the threshold, with an interference duty cycle of approximately 60%; and another instance where the power is significantly higher than the threshold, with an interference duty cycle reaching 80% (strong persistent interference). Since the interference power exceeds the threshold in both instances, these are unwanted strong interference signals that will affect communication quality. This type of chart is used in interference analysis within the SDR channel assessment engine to help determine the location, intensity, and persistence characteristics of interference, providing data support for communication mode switching and frequency band avoidance.

[0098] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0099] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dual-mode adaptive communication method based on SDR, characterized in that, The method includes: Receive raw wireless signal streams from the radio frequency front end and obtain a preliminary determination of the communication protocol family category based on the raw signal streams; Based on the initially determined communication protocol family category, the corresponding signal analysis engine and channel evaluation engine are activated. The signal analysis engine performs in-depth analysis of the signal stream to reconstruct the complete communication message unit, and the channel evaluation engine quantifies and characterizes the noise floor, multipath delay, and interference spectrum during signal transmission. The communication message unit and the quantized channel state are input together into a dynamic decision network, which is configured to learn the implicit association rules between message semantic integrity and channel transmission reliability. Based on the learned implicit association rules, the dynamic decision network outputs a collaborative decision instruction containing a weight vector, which defines the priority order of the primary communication mode and alternative communication modes to be used in the next communication cycle. The collaborative decision-making instructions are compiled into a sequence of configuration parameters that can be recognized by the software-defined radio hardware, driving the radio frequency front end and the baseband processing unit to reconfigure according to the new parameter sequence, thereby completing the adaptive switching of communication modes. The step of inputting the communication message unit and the quantized channel state into a dynamic decision network includes: The communication message units are classified according to semantic importance, and different semantic weights are assigned to the message header field, control payload field and data payload field according to the field priority defined in the protocol syntax rule base. Availability mapping is performed on the quantized channel state, and the quantized representation of the noise floor, the quantized representation of the multipath delay, and the quantized representation of the interference spectrum are respectively mapped to negative factor levels that affect the reliability of signal transmission. A dynamic decision network is constructed with semantic weight vector and negative factor level vector as input. The dynamic decision network contains a multi-layer cross-sensing unit to calculate the probability estimate of successful transmission of messages with different semantic weights under a specific combination of negative factors. The multi-layer cross-sensing unit performs a non-linear transformation on the input vector. The computation process concatenates the semantic weight vector with the negative factor level vector, and then passes it through multiple fully connected layers and activation functions. The intermediate layer computation is represented as follows: ； in: Represents the hidden layer feature vector. This represents the Sigmoid activation function. Represents the weight matrix. Represents the semantic weight vector. This represents the channel negative factor vector, which is composed of the noise floor negative factor level, the multipath delay negative factor level, and the interference spectrum negative factor level. This represents the bias vector, and the final output layer will take the hidden layer feature vectors. It is mapped to a scalar representing an estimate of the probability of successful transmission; The construction steps of the signal parsing engine include: The communication protocol family categories initially determined are subjected to protocol feature extraction, and the corresponding syntax rule set and frame synchronization word mode are obtained from the protocol feature library; Configure the state machine of the protocol parser according to the set of syntax rules, wherein the state machine defines the parsing order and verification rules of message fields; The correlator parameters of the frame detector are configured according to the frame synchronization word mode, and the correlator parameters include the coefficients of the matched filter and the synchronization threshold; The configured protocol parser and frame detector are loaded into the execution environment of the signal parsing engine to establish a processing pipeline from signal stream input to communication message unit output; The construction steps of the channel evaluation engine include: Background noise is sampled from the original wireless signal stream to establish a statistical model of the noise floor; Configure a multipath delay estimation module, use a generalized cross-correlation algorithm to calculate the broadening of the signal correlation peak, and initialize the delay expansion parameters; Configure the interference spectrum monitoring module, set the frequency band scanning range and power threshold, to identify unwanted interference signals; The noise floor statistical model, multipath delay estimation module, and interference spectrum monitoring module are integrated into the channel evaluation engine to form an output interface for quantitative representation of channel state.

2. The SDR-based dual-mode adaptive communication method according to claim 1, characterized in that, The process of receiving the raw wireless signal stream from the radio frequency front-end and obtaining the preliminarily determined communication protocol family category based on the raw signal stream includes: The original wireless signal stream is subjected to joint detection of signal energy distribution and protocol frame start symbol to locate potential valid communication data segments in the signal stream; Blind parsing is performed on the potential valid communication data segments. During the blind parsing process, candidate modulation patterns, symbol rate features, and frame structure features are extracted simultaneously. Based on the extraction results, multiple rounds of fuzzy matching are performed in a preset protocol feature library to preliminarily determine the communication protocol family category to which the current signal stream belongs. Specifically, this includes: Sliding window energy detection is performed on the potential valid communication data segments to identify signal pulse intervals with energy exceeding the silence threshold, and the time-domain envelope waveform of each signal pulse interval is extracted. A multi-scale wavelet transform is performed on the time-domain envelope waveform to separate the periodic component representing the symbol rate and the aperiodic component representing the modulation pattern from the wavelet transform coefficients. Calculate the constellation point distribution of the aperiodic component in the complex plane, and perform point-by-point matching of the constellation point distribution with the known modulation constellation templates stored in the protocol feature library to generate a modulation pattern matching degree list; Simultaneously, autocorrelation analysis is performed on the periodic components to extract the main peak interval as a candidate symbol period. The candidate symbol period is then compared with the standard symbol periods of the protocol stored in the protocol feature library to generate a symbol period matching degree list. By combining the modulation pattern matching list and the symbol period matching list, a comprehensive matching score is calculated for each candidate protocol, and protocols with comprehensive matching scores exceeding the judgment threshold are selected as the communication protocol family category for the preliminary determination.

3. The SDR-based dual-mode adaptive communication method according to claim 2, characterized in that, The activation of the corresponding signal parsing engine and channel evaluation engine includes: Based on the preliminarily determined communication protocol family category, the corresponding protocol syntax rule library and frame synchronization word template are loaded from the engine resource pool into the working memory of the signal parsing engine; The signal parsing engine performs precise frame synchronization positioning of the signal stream based on the loaded frame synchronization word template, and performs syntax parsing of the synchronized data bit stream based on the loaded protocol syntax rule library, decomposing the message header, payload and check fields. The channel evaluation engine captures background noise samples before and after the signal pulse interval while parsing the signal, and estimates the power spectral density of the background noise samples to obtain a quantitative characterization of the noise floor. The channel evaluation engine further estimates the quantitative characterization of the multipath delay by performing broadening analysis on the temporal correlation peak of the frame synchronization word template. The channel evaluation engine also identifies unwanted high-power frequency points by monitoring the power of frequency bands outside the signal pulse interval, and generates a quantitative representation of the interference spectrum.

4. The SDR-based dual-mode adaptive communication method according to claim 3, characterized in that, The dynamic decision network outputs a collaborative decision instruction containing a weight vector, including: The dynamic decision network inputs the calculated probability estimate of successful transmission into a strategy optimization layer, which has a variety of preset communication mode switching strategies built in. The strategy optimization layer evaluates the expected change in the estimated success transmission probability when maintaining the current communication mode and applying each preset communication mode switching strategy in the next communication cycle. Select a target strategy that maximizes the expected change in the estimated probability of successful transmission, and generate a corresponding weight vector based on the target strategy. The weight vector includes a determination identifier and parameter configuration for the primary communication mode, as well as preparation instructions for alternative communication modes.

5. The SDR-based dual-mode adaptive communication method according to claim 4, characterized in that, The step of compiling the collaborative decision-making instructions into a sequence of configuration parameters recognizable by the software-defined radio hardware includes: Parse the weight vector in the collaborative decision-making instruction and extract the standard hardware configuration template of the primary communication mode corresponding to the determination identifier; Based on the real-time channel parameters fed back by the current signal analysis engine, the adjustable parameters in the standard hardware configuration template are fine-tuned to generate an accurate parameter set adapted to the current channel. The adjustable parameters are center frequency, bandwidth, and transmit power; The precise parameter set is encapsulated into a continuous sequence of configuration parameters according to the data format and order specified by the software-defined radio hardware driver interface, and the end of the configuration parameter sequence contains a checksum.

6. The SDR-based dual-mode adaptive communication method according to claim 5, characterized in that, The driving RF front-end and baseband processing unit are reconfigured according to a new parameter sequence, including: The configuration parameter sequence is sent to the programmable register group of the RF front end via the control bus. The programmable register group updates the local oscillator frequency, filter bandwidth and power amplifier bias voltage according to the received data. Synchronously, the parameter portion of the configuration parameter sequence related to baseband processing is sent to the programmable logic array of the baseband processing unit, and the programmable logic array reloads the corresponding digital filter coefficients, modem cores and codec cores according to the received parameters; After both the programmable register group and the programmable logic array have been loaded, a synchronization enable signal is sent, causing the RF front-end and the baseband processing unit to start working simultaneously under the new configuration.

7. A dual-mode adaptive communication system based on SDR, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the SDR-based dual-mode adaptive communication method described in any one of claims 1 to 6.

Images