Intelligent anti-interference method for DCSK communication system based on software radio platform

Abstract

The present invention relates to an intelligent anti-interference method for a DCSK communication system based on a software radio platform, and belongs to the field of communication anti-interference. A DCSK intelligent anti-interference system is built based on GNU Radio and a matching USRP device, comprising a DCSK transmitting and receiving module, an interference module, and an intelligent anti-interference decision module. A reinforcement learning algorithm is introduced into the decision module to establish a state set consisting of different powers, communication frequency bands, and chaotic signal waveforms, as well as an action set consisting of communication system adjustment schemes. The DCSK system can make communication quality evaluations when facing different types of interference, thereby making intelligent decisions and selections for system parameter configurations. By verifying the anti-interference performance in an actual electromagnetic environment, the method of the present invention has good anti-interference effects, can effectively resist interference without significantly increasing the transmitter transmission power, and achieves safe and reliable information transmission.

Description

Technical Field

[0001] The present invention relates to the field of communication anti-interference, and in particular to an intelligent anti-interference method for a differential chaotic keying (DCSK) communication system based on a software radio platform. The method has good anti-interference effect and can provide a basis for research on anti-interference methods for DCSK communication systems in actual electromagnetic environments. Background Art

[0002] With the increasing level of modern information technology and the increasingly complex electromagnetic environment, wireless communication systems are facing an increasingly severe threat of interference, significantly impacting communication security and reliability. To improve the performance of communication systems in complex electromagnetic environments, higher requirements must be placed on them. Chaotic communication is a communication method that uses chaotic signals as carriers. Due to the unpredictability, initial value sensitivity, and noise-like wide-spectrum characteristics of chaotic signals, it offers significant advantages in communication confidentiality and multi-user anti-interference. Chaotic communication is mainly categorized into chaotic spread spectrum communication, chaotic analog communication, and chaotic keying communication. Chaotic keying communication, with its relatively simple transceiver implementation, has long been a key research topic. Differential chaotic keying (DCSK), in particular, is a typical example of chaotic keying communication, requiring no additional transmit / receive synchronization and therefore simple to implement. However, in environments such as electronic warfare, human electromagnetic interference can seriously affect the communication effect of the DCSK system, resulting in an increase in the bit error rate or even an inability to transmit information. Therefore, it is of great practical significance to study how the DCSK system can achieve high-reliability and high-security communication in the face of interference. Common anti-interference solutions for DCSK include adjusting the transmission power, communication frequency, chaotic signal waveform, etc.

[0003] Software-defined radio (SDR) is a new concept and system for wireless communication. It leverages highly programmable DSP devices to achieve relative independence between system hardware structure and functionality, providing a relatively versatile hardware platform. Software-defined radio (SDR) allows for programmable control of communication parameters such as operating frequency and modulation, enabling diverse communication functions and enhancing system flexibility. GNU Radio, an open-source SDR development framework, enables customizable SDR signal processing. GNU Radio and the Universal Software-Defined Radio Peripheral (USRP) can form a software-defined radio system.

[0004] In recent years, intelligent anti-interference technology has become a key research area for wireless communication anti-interference, and reinforcement learning (RL) algorithms have been increasingly applied. Reinforcement learning is a type of machine learning that uses learning to maximize rewards or achieve specific goals. Integrating intelligent agents into communication systems allows them to make informed decisions based on the real-time electromagnetic environment and interference conditions. Previous research has applied artificial intelligence algorithms to switching communication waveforms to mitigate different types of interference. Other research has also developed an intelligent anti-interference decision-making model for satellite communications based on reinforcement learning, which can make decisions about mitigating a specific interference type over a period of time.

[0005] Currently, intelligent anti-interference methods have been studied for specific types of communication systems, but these methods suffer from low scalability, reconfigurability, and flexibility. Traditional anti-interference techniques for chaotic communication systems are no longer able to meet current performance requirements, and no literature has yet been published on the application of intelligent anti-interference in chaotic communication systems. Summary of the Invention

[0006] The present invention provides an intelligent anti-interference method for a DCSK communication system based on a software radio platform. By building a software radio platform and researching intelligent anti-interference methods, the method aims to enable the DCSK communication system to make intelligent decisions and configurations for parameters such as communication power, communication frequency, and chaotic waveforms when facing complex electromagnetic interference. This method effectively resists single-tone interference, multi-tone interference, and partial-band interference at the lowest possible transmit power, while ensuring the system's communication performance and maintaining a low bit error rate. The intelligent anti-interference decision module uses a reinforcement learning method based on sampled data from an environmental sensing component to determine the interference type of the interfering signal based on information such as the power and frequency of the interfering signal. The intelligent anti-interference decision module then makes decisions based on user requirements for the system's bit error rate and transmit power, combined with preset constraints, to obtain a DCSK communication system parameter configuration suitable for the current communication environment, thereby improving the communication system's anti-interference performance.

[0007] The technical solution adopted by the present invention comprises the following steps:

[0008] Step 1: Design a DCSK communication intelligent anti-interference system based on the GNU Radio software radio platform, including a DCSK system transmitter module, a receiver module, an interference module, and an intelligent anti-interference decision module;

[0009] Step 2: Build the DCSK system transmitter and receiver modules based on the GNU Radio software radio platform, and test and record the DCSK communication performance indicators in the absence of interference.

[0010] Step 3: Build various types of jamming modules based on the GNU Radio software radio platform, and test and record the performance indicators of the DCSK communication subsystem under jamming conditions.

[0011] Step 4: Design the DCSK communication intelligent anti-interference decision module, introduce the reinforcement learning algorithm into the intelligent anti-interference decision module, and formulate the intelligent decision-making standard;

[0012] Step 5: Set the reward value function for reinforcement learning. According to the relative importance of bit error rate and transmit power, the reward value function is calculated by weighting the two according to their relative importance, with weights ω1 and ω2 respectively.

[0013] Step 6: Normalize the parameter indicators in reinforcement learning. Since the reward value function is calculated using the weighted summation method, it is necessary to unify the dimensions of the parameter indicators and unify the two parameter indicators to the [0, 1] interval.

[0014] Step 7: Set the states and actions for reinforcement learning. Define a state set and initialize each state. Define an action set so that the reinforcement learning agent can traverse the entire state set using the actions in the action set.

[0015] In step 8, value iteration is performed based on the reward function of the current state of the DCSK communication system. The cumulative value of the reward function for all states during the decision-making process is used as the state value function until the state value function converges. A state transition diagram is obtained and visualized. The communication performance indicators of the optimal state are tested to see whether they meet the standards for intelligent decision-making.

[0016] In step 3 of the present invention, three types of interference signals are used, namely single-tone interference, multi-tone interference and partial-band interference.

[0017] In step 4 of the present invention, the bit error rate is used as a standard for intelligent decision-making to intuitively represent the communication quality. The bit error rate R1 of the DCSK communication system in the absence of interference, the bit error rate R2 of the DCSK communication system in the presence of interference, and the bit error rate R3 of the DCSK communication system in the presence of interference after being resisted by the intelligent anti-interference method are tested respectively. R3 is compared to see whether it is in the same order of magnitude as R1 and much smaller than R2, which is used as the standard for intelligent decision-making.

[0018] The reward value function in step 5 of the present invention is:

[0019] R=ω1f ber +ω2f power

[0020] Among them, f ber is the bit error rate evaluation index, fpower is the evaluation index of transmission power, according to Shannon formula:

[0021]

[0022] Among them, C is the channel capacity, B is the bandwidth, S is the signal power, and N is the interference power. The channel capacity is related to the bandwidth and the signal-to-interference ratio (S / N). Therefore, when the bandwidth and interference power are determined, the channel capacity is proportional to the transmission power. A weighted calculation is performed based on the relative importance of the bit error rate and transmission power to obtain the reward value function of the reinforcement learning system, with weights ω1 and ω2 respectively.

[0023] The present invention takes the bit error rate as the most important indicator in wireless communication, sets the weight ω1 of the bit error rate to 0.83, and sets the weight ω2 of the transmission power indicator to 0.17.

[0024] In step 6 of the present invention, a weighted sum is required, so the two parameter indicators are normalized, where the normalized result of the bit error rate is:

[0025]

[0026] Where B max Indicates the maximum bit error rate, which is set to 0.4 here, B real represents the bit error rate measured experimentally, B min Indicates the minimum bit error rate, B min Set to 10 -6 , transmit power index f power The normalized result is:

[0027]

[0028] Where, P max is the maximum signal-to-interference ratio, the value range of the signal-to-interference ratio is 0dB~10dB, so P max Indicates the maximum signal-to-interference ratio, which is 10dB, P s Indicates the current signal-to-interference ratio, P min Indicates the minimum signal-to-interference ratio, which is 0dB.

[0029] The state set described in step 7 of the present invention is composed of a combination of chaotic signal waveform selection, communication frequency selection and transmission power control of the DCSK communication system. m situations are set for the frequency used for communication. The chaotic signal waveform is considered from two perspectives. First, the type of chaotic signal. Different types of chaotic signals have different generation modes and generate completely different waveforms. For n types of chaotic signals, since chaotic signals are sensitive to initial values, using different initial values ​​for chaotic signals of the same type will also lead to inconsistency in the chaotic signal waveform. Therefore, j different chaotic signal initial values ​​are used. For the transmission power, k power selections can be set. With interference power as a reference, different signal-to-interference ratios are taken. The above three parameters are combined to form a plane with a total of m×n×j×k states.

[0030] The setting of the action set in step 7 of the present invention refers to the state mapping diagram of reinforcement learning. For the state set, a plane with a width of m and a length of (n×j×k) has been constructed. In order to simplify the setting of actions in reinforcement learning and the Markov transition model in the decision-making process, four actions of up, down, left, and right are set. By moving up, down, left, and right in the state mapping diagram, the purpose of traversing all states can be achieved. At the same time, the state transition diagram can be represented by a visual method. When in a certain state, the set action is taken to adjust to the next state according to the current state, that is, the Markov characteristic is satisfied.

[0031] The optimal state in step 8 described in the present invention adopts the idea of ​​dynamic programming. In order to make the state value function converge, a discount factor of 0.95 needs to be added. When the state value function converges, the optimal state can be obtained. The communication performance indicators under the optimal state are tested to compare whether they meet the standards of intelligent decision-making in step 4. If they meet the standards, the final target state is obtained.

[0032] The intelligent decision-making system method of the present invention belongs to model-based learning. The system itself is a Markov decision process with four known quadruples, which is a multi-step reinforcement learning task. The decision-making system internally simulates similar or identical environmental conditions and builds a corresponding model. The present invention adopts a reinforcement learning method based on value iteration. Since the value iteration method does not update the policy during the training process, the value function of each state does not need to be calculated during the iteration process, thereby achieving the purpose of reducing algorithm complexity and improving convergence speed.

[0033] The advantages of the present invention are: (1) When DCSK communication is subject to artificial electromagnetic interference, the system can automatically adjust the three parameters of communication frequency, power, and chaotic waveform, and use bit error rate and power as evaluation criteria to conduct a comprehensive analysis of the communication system performance before and after the adjustment. (2) The software radio-based method has good flexibility and strong later reconfigurability and scalability. (3) The introduction of a reinforcement learning intelligent algorithm can quickly provide the optimal adjustment strategy to resist different types of interference, and the effect of the intelligent anti-interference method can be intuitively observed. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a framework diagram of the DCSK communication intelligent anti-interference system in the present invention;

[0035] Figure 2 It is a flow chart of the DCSK communication intelligent anti-interference method of the present invention;

[0036] Figure 3 This is the flow chart of the transmitter module of the DCSK communication system built on the GNU Radio software radio platform;

[0037] Figure 4 This is the flow chart of the DCSK communication system receiver module built using the GNU Radio software radio platform;

[0038] Figure 5 This is the flow chart of the single-tone jammer module in the jammer module built using the GNU Radio software radio platform.

[0039] Figure 6 This is the flow chart of the multi-tone interference module in the interference module built on the GNU Radio software radio platform;

[0040] Figure 7 This is a flow chart of some interference modules in the interference module built using the GNU Radio software radio platform;

[0041] Figure 8 It is a state and parameter mapping diagram of the method of the present invention;

[0042] Figure 9 It is a graph of the iterative process of the state value function of the intelligent anti-interference decision module when the DCSK communication system is interfered with by a single tone;

[0043] Figure 10 It is a state transition diagram of the decision-making process of the intelligent anti-interference decision module when the DCSK communication system is interfered with by a single tone;

[0044] Figure 11 It is a graph of the iterative process of the state value function of the intelligent anti-interference decision module when the DCSK communication system is subject to multi-tone interference;

[0045] Figure 12 It is a state transition diagram of the decision-making process of the intelligent anti-interference decision module when the DCSK communication system is subject to multi-tone interference;

[0046] Figure 13 It is a curve diagram of the iterative process of the state value function of the intelligent anti-interference decision module when the DCSK communication system is subject to 30% partial interference;

[0047] Figure 14 It is a state transition diagram of the decision-making process of the intelligent anti-interference decision module when the DCSK communication system is subject to 30% partial interference. DETAILED DESCRIPTION

[0048] The technical solution of the present invention is described in detail below in conjunction with specific embodiments. Figure 2 FIG. 1 is a flow chart of an intelligent anti-interference method for a DCSK communication system according to the present invention. The present invention provides an intelligent anti-interference method for a DCSK communication system based on a software radio platform, comprising the following steps:

[0049] Step 1: Design a software radio based on GNU Radio. Figure 1 The DCSK communication intelligent anti-interference system shown includes a DCSK system transmitter module, a receiver module, an interference module, and an intelligent anti-interference decision module;

[0050] Step 2: Build the following software based on GNU Radio platform: Figure 3 The DCSK system transmitter module shown in the figure, where ChaosGenerator is a chaotic signal generator, and the data information generated by the signal source VectorSource is modulated by the DCSK chaotic modulator ChaosModulator and sent to the channel through the signal transmission module UHD: USRPSink. QTGUIFrequencySink can display the spectrum of the transmitted signal; build as follows Figure 4 The receiver module shown in the figure first receives the communication signal through the signal receiving module UHD: USRPsource, then demodulates the data information through the DCSK chaos demodulation module ChaosDemodulator and saves it to the sink file through FileSink. The DCSK communication performance indicators in the absence of interference are tested and recorded.

[0051] Step 3: Build various types of jammers based on the GNU Radio software radio platform, including Figure 5The single-tone interference module shown, where SignalSource is the signal source, is sent to the channel through the signal transmission module UHD:USRPSink. The spectrum of the interference signal can be seen through WXGUIFFTSink. Figure 6 The multi-tone interference module shown is composed of multiple single-tone interference modules combined by an adder Add; Figure 7 The figure shows the partial-band interference module, where NoiseSource is a Gaussian noise interference source. The bandpass filter BandPassFilter generates a partial-band interference signal. The actual spectrum of the interference signal is observed to see if it meets the interference requirements. The performance indicators of DCSK communication under interference are tested and recorded.

[0052] Step 4: Design an intelligent anti-interference decision-making module for DCSK communication, introduce a reinforcement learning algorithm into the module, and establish a standard for intelligent decision-making. The standard for intelligent decision-making is to maintain the same bit error rate as in normal communication without significantly increasing the transmit power.

[0053] As for the standard of intelligent decision-making, the bit error rate is used as the standard to intuitively express the communication quality. The bit error rate R1 of the DCSK communication system in the absence of interference, the bit error rate R2 of the DCSK communication system in the presence of interference, and the bit error rate R3 of the DCSK communication system after interference resistance by intelligent anti-interference methods are tested respectively. The comparison is made to see whether R3 is in the same order of magnitude as R1 and much smaller than R2, which is used as the standard for intelligent decision-making.

[0054] Step 5: Set the reinforcement learning reward function. When setting the reinforcement learning reward function, the bit error rate and transmission power should be considered. According to Shannon's formula:

[0055]

[0056] Where C is the channel capacity, B is the bandwidth, S is the signal power, and N is the interference power. The channel capacity is related to the bandwidth and the signal-to-interference ratio. Therefore, when the bandwidth and interference power are determined, the channel capacity is proportional to the transmit power. A weighted calculation is performed based on the relative importance of the bit error rate and transmit power to obtain the reward value function of the reinforcement learning system, with weights ω1 and ω2 respectively.

[0057] Generally, the bit error rate is regarded as the most important indicator in wireless communication, so the weight of the bit error rate ω1 is set to 0.83, and the weight of the transmit power indicator ω2 is set to 0.17;

[0058] R=ω1f ber +ω2f power

[0059] Among them, fber is the bit error rate indicator, f power is the transmit power indicator;

[0060] Step 6: Normalize the parameter indicators in reinforcement learning. As mentioned in step 5, the reward value function uses a weighted summation method, which requires unifying the dimensions of the function parameter indicators and normalizing the two parameter indicators to the [0,1] range. Among them, the bit error rate indicator f ber The normalized result is:

[0061]

[0062] Where B max Indicates the maximum bit error rate, which is set to 0.4 here, B real represents the bit error rate measured experimentally, B min Indicates the minimum bit error rate, B min Set to 10 -6 , transmit power index f power The normalized result is:

[0063]

[0064] Where, P max is the maximum signal-to-interference ratio, the value range of the signal-to-interference ratio is 0dB~10dB, so P max Indicates the maximum signal-to-interference ratio, which is 10dB, P s Indicates the current signal-to-interference ratio, P min Indicates the minimum signal-to-interference ratio, which is 0dB.

[0065] Step 7: Set states and actions in the reinforcement learning algorithm. First, define a state set, initialize each state, randomly assign an action to each state, and calculate the reward function for each state. A larger reward function indicates better communication performance. The cumulative value of the reward functions for all states passed during the decision-making process is used as the state function. The DCSK communication system can adjust the communication bit error rate through various combinations of chaotic waveform selection, communication frequency selection, and power control selection. The intelligent anti-interference decision module can find the communication state suitable for the current environment by changing the above parameters. Therefore, changes in these three parameters are defined as the actions of the decision method, and the communication states obtained by various combinations of adjustment actions are defined as the states of the decision method.

[0066] The state map of reinforcement learning is as follows Figure 8As shown, the present invention sets five conditions for the communication frequency, ranging from 550MHz to 590MHz, divided in steps of 10MHz, generating five options. The chaotic signal waveform is considered from two perspectives: first, the type of chaotic signal. Different types of chaotic signals have different generation modes and produce completely different waveforms. Therefore, the present invention provides two examples: logistic chaotic signal and Tent chaotic signal. Second, the initial value of the chaotic signal. Since chaotic signals are sensitive to initial values, using different initial values ​​for the same type of chaotic signal can also lead to inconsistent chaotic signal waveforms. The present invention uses initial values ​​of 0.3 and 0.7 for the two chaotic signals, respectively, for experimentation. Three power levels are available for transmission power, with interference power as a reference: 0dB, 5dB, and 10dB. Combining these three parameters creates a 5×(2×2×3) plane, with a total of 60 states.

[0067] The action setting for reinforcement learning refers to the state map of reinforcement learning. For the state set, a 5×(2×2×3) plane has been constructed. To simplify the action setting in reinforcement learning and the Markov transition model in the decision-making process of this method as much as possible, the present invention sets four actions: up, down, left, and right. By moving up, down, left, and right in the state map, the purpose of traversing all states can be achieved. At the same time, the state transition diagram can be represented by a visual method. When in a certain state, taking the set action can adjust the state to the next moment according to the current state, that is, satisfying the Markov characteristic.

[0068] In step 8, value iteration is performed based on the reward value function of the current state of the DCSK communication system to obtain a state transition diagram, which is then visualized. This method employs dynamic programming to solve for the optimal state. To ensure convergence of the state value function, a discount factor of 0.95 is added. Once the state value function converges, the optimal state is achieved. Communication performance indicators in the optimal state are tested to see if they meet the intelligent decision-making criteria in step 4. If so, this state is selected as the target state.

[0069] The effects of the present invention are further illustrated below through experimental examples.

[0070] Experimental example: Experimental verification of intelligent anti-interference method of DCSK communication system based on software radio platform.

[0071] This experiment is based on the GNU Radio software radio platform and the supporting USRP device. Figures 3 to 7 The DCSK communication system transmitter module, receiver module and interference module shown in FIG are used to establish an intelligent anti-interference decision module. Figure 8The state mapping diagram shown in the figure, the experimental conditions are shown in Table 1.

[0072] Table 1

[0073]

[0074] (1) Verification of anti-interference performance of DCSK communication system

[0075] The experiment was conducted under a signal-to-interference ratio (SIR) of -5dB. A logistic chaotic waveform with an initial value of 0.3 was used as the test waveform for DCSK modulation and information transmission. A corresponding interference signal was also added to the communication system. The bit error rates of the communication system before and after interference were recorded, as shown in Table 2. It can be seen that the bit error rate of information transmission increased significantly after interference, but decreased significantly after the waveform adjustment, demonstrating a certain degree of interference resistance.

[0076] Table 2

[0077]

[0078] (2) Anti-single-tone interference experiment

[0079] When single-tone interference is added during DCSK communication, the curve diagram of the iteration process of the state value function of the intelligent anti-interference decision module is as follows: Figure 9 As shown, it can be seen that the function value has converged after multiple iterations; the state transition diagram is as follows Figure 10 As shown in the state transition diagram, states 9, 24, and 51 are selected as communication states suitable for the current interference environment. Bit error rates were tested before and after the adjustment. The results show that the bit error rates in these three states are much lower than those before the adjustment, indicating that the decision-making method is correct.

[0080] (3) Anti-multi-tone interference experiment

[0081] When multi-tone interference is added during DCSK communication, the curve diagram of the iteration process of the state value function of the intelligent anti-interference decision module is as follows: Figure 11 As shown, it can be seen that the function value has converged after multiple iterations; the state transition diagram is as follows Figure 12 As shown in the state transition diagram, states 6, 36, and 54 are suitable for communication in the current interference environment. Testing the bit error rate before and after the adjustment revealed that the bit error rate in these states was much lower than before the adjustment, demonstrating that the decision-making method was correct.

[0082] (4) Anti-partial band interference experiment

[0083] When 30% partial interference is added during DCSK communication, the curve diagram of the iteration process of the state value function of the intelligent anti-interference decision module is as follows: Figure 13 As shown, it can be seen that the function value has converged after multiple iterations; the state transition diagram is as follows Figure 14 As shown in the state transition diagram, states 7, 37, and 40 are suitable for communication in the current interference environment. Testing the bit error rate before and after the adjustment revealed that the bit error rate in these states was much lower than before the adjustment, demonstrating that the decision-making method was correct.

[0084] The experimental results show that the intelligent anti-interference method for the DCSK communication system based on software radio proposed in this invention has a good anti-interference effect. Reinforcement learning can be used on the software radio platform to intelligently select and combine the chaotic signal waveform, transmission power and communication frequency of the DCSK communication system, ensuring that information transmission has high security and reliability in the face of different interferences.

Claims

1. An intelligent anti-interference method for a DCSK communication system based on a software radio platform, characterized in that: The following steps are involved: Step 1: Design a DCSK communication intelligent anti-interference system based on the GNU Radio software radio platform, including a DCSK system transmitter module, a receiver module, an interference module, and an intelligent anti-interference decision module; Step 2: Build the DCSK system transmitter and receiver modules based on the GNU Radio software radio platform, and test and record the DCSK communication performance indicators in the absence of interference. Step 3: Build various types of jamming modules based on the GNU Radio software radio platform, and test and record the performance indicators of the DCSK communication subsystem under jamming conditions. Step 4: Design a DCSK communication intelligent anti-interference decision-making module and introduce a reinforcement learning algorithm into the module to establish intelligent decision-making standards. The bit error rate is used as a standard to intuitively represent communication quality. Test the bit error rate R1 of the DCSK communication system without interference, the bit error rate R2 of the DCSK communication system with interference, and the bit error rate R3 of the DCSK communication system with interference after intelligent anti-interference methods are used to mitigate interference. Compare R3 to see if it is on the same order of magnitude as R1 and much smaller than R2, which serves as the standard for intelligent decision-making. Step 5: Set the reward value function of reinforcement learning. According to the relative importance of bit error rate and transmission power, the reward value function is obtained by weighting the two according to their relative importance. The weights are and ; Step 6: Normalize the parameter indicators in reinforcement learning. The normalized result of the bit error rate is: ； Where, Indicates the maximum bit error rate, represents the bit error rate measured experimentally, Indicates the minimum bit error rate and transmit power index The normalized result is: ； Where, is the maximum signal-to-interference ratio, represents the maximum signal-to-interference ratio, Indicates the current signal-to-interference ratio, represents the minimum signal-to-interference ratio; Step 7: Set the state and action of reinforcement learning, define a state set, and initialize each state; Define an action set so that the reinforcement learning agent can traverse the entire state set by taking actions in the action set; In step 8, value iteration is performed based on the reward value function of the current state of the DCSK communication system. The cumulative value of the reward value function of all states passed during the decision-making process is used as the state value function until the state value function converges. The state transition diagram is obtained and represented by a visualization method. The communication performance indicators of the optimal state are tested to observe whether they meet the standards of intelligent decision-making.

2. The intelligent anti-interference method for a DCSK communication system based on a software radio platform according to claim 1, wherein: In step 3, three types of interference signals are used, namely single-tone interference, multi-tone interference and partial-band interference.

3. The intelligent anti-interference method for a DCSK communication system based on a software radio platform according to claim 1, wherein: The reward function in step 5 is: ； in, is the bit error rate evaluation index, is the evaluation index of transmission power, is the weight of the bit error rate, is the weight of the transmit power.

4. The intelligent anti-interference method for a DCSK communication system based on a software defined radio platform according to claim 3, wherein: Taking the bit error rate as the most important indicator in wireless communication, the weight of the bit error rate is Set to 0.83, the weight of the transmit power indicator Set to 0.

17.

5. The intelligent anti-interference method for DCSK communication system based on software radio platform according to claim 1 is characterized in that: The state set described in step 7 is composed of a combination of chaotic signal waveform selection, communication frequency selection and transmission power control of the DCSK communication system. For n kinds of chaotic signals, j different chaotic signal initial values ​​are used; for the transmission power, k power selections are set. The above three parameters are combined to form a total of states can form a plane.

6. The intelligent anti-interference method for a DCSK communication system based on a software radio platform according to claim 1, characterized in that: The setting of the action set in step 7 refers to the state mapping diagram of reinforcement learning. For the state set, a plane has been formed with a width of , long for Four actions, up, down, left, and right, are set. By moving up, down, left, and right in the state mapping diagram, the purpose of traversing all states can be achieved. At the same time, the state transition diagram can be represented by a visual method. When in a certain state, the set action is taken to adjust to the next state according to the current state, which satisfies the Markov characteristic.

7. The intelligent anti-interference method for DCSK communication system based on software radio platform according to claim 1 is characterized in that: The optimal state in step 8 adopts the idea of ​​dynamic programming and adds a discount factor of 0.

95. When the state value function converges, the optimal state can be obtained. The communication performance indicators under the optimal state are tested to compare whether they meet the standards of intelligent decision-making in step 4. If they meet the standards, the final target state is obtained.

Images