Anti-ssdf attack cooperative spectrum sensing method based on online learning sequential probability ratio test

Abstract

The application discloses a kind of based on online learning sequential probability ratio test anti-SSDF attack cooperative spectrum sensing method, first by secondary user energy detection is carried out, and the sensing result is reported to fusion center;Then, fusion center is based on Boltzmann distribution random selection secondary user, using preliminary fusion+dynamic addition step SPRT fusion strategy, calculate weighted likelihood ratio to complete decision;Finally, in combination with the number of participating secondary users and historical mean deviation design reward and punishment mechanism, through sliding window maintenance historical information, realize the reliability of secondary user and the adaptive optimization of participating scale.The application can dynamically adapt channel fading and probability type SSDF attack, while ensuring the sensing accuracy, reduce the number of participating secondary users and system energy consumption, significantly improve the sensing efficiency and attack robustness in low signal-to-noise ratio, high malicious proportion scene.

Description

Technical Field

[0001] This invention relates to the cross-technical field of cognitive radio networks and Internet security services, and more specifically to a cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio testing. Background Technology

[0002] With the rapid development of wireless communication technology, the demand for spectrum resources is constantly increasing. However, the current mainstream spectrum management model is still a static spectrum allocation system, resulting in a large number of idle spectrum gaps. Cognitive radio networks allow secondary users to intelligently sense and opportunistically access idle licensed frequency bands without interfering with the normal communication of primary users, making full use of spectrum gaps and potentially solving the dual problems of spectrum scarcity and low utilization.

[0003] One of the key tasks in realizing cognitive radio networks is spectrum sensing. Since the spectrum sensing capabilities of a single secondary user are limited, cooperative spectrum sensing, which integrates sensing information from multiple secondary users, can more accurately detect idle spectrum. While cooperative spectrum sensing improves sensing performance, its open nature makes it vulnerable to spectrum sensing data forgery (SSDF) attacks. Malicious users report false sensing data, leading the fusion center to make incorrect decisions, reducing spectrum utilization and interfering with the normal communication of primary users.

[0004] Existing anti-SSDF attack schemes still have shortcomings, such as the use of a fixed number of secondary users in the fusion, which cannot adapt to changes in channel quality and the proportion of malicious users; the reputation mechanism is not integrated with the variable sample size, making it difficult to balance energy efficiency and anti-attack capability; and the instantaneous misjudgment of normal users caused by channel fading is not considered, which can easily misclassify deeply fading secondary users as malicious. Summary of the Invention

[0005] The present invention provides a collaborative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test, which can dynamically select secondary users and has both high efficiency and strong anti-attack capability, and can solve at least one of the above-mentioned technical problems.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A collaborative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio testing includes the following steps: S1. Each secondary user performs local energy detection and reports the detection results to the fusion center; S2. The fusion center calculates the selection probability of each secondary user based on the Boltzmann distribution and learning parameters, and randomly selects secondary users based on the selection probability to form a preliminary fusion set; S3. Perform preliminary sequential probability ratio test (SPRT) fusion on the preliminary fusion set. The SPRT fusion calculates the weighted likelihood ratio as the decision statistic based on the variable sample size decision mechanism. S4. Regarding the judgment statistic, if it falls within the interval... If it falls outside the specified range, it is considered to have met the conditions for termination of the judgment. Within a certain range, secondary users are dynamically added and the weighted likelihood ratio is updated until the termination condition is met or all secondary users are traversed. and This is the decision threshold; S5. Output the final fusion decision result based on the final weighted likelihood ratio and complete the parameter update for the next time slot: Based on the final fusion decision result, perform learning parameter reward and punishment update by combining the deviation between the current number of participating secondary users and the historical average, and maintain the historical average number of participating secondary users through a sliding window.

[0007] Furthermore, in S3, the sequential probability ratio test SPRT fusion adopts a variable sample size decision mechanism, and the decision statistic is the weighted log-likelihood ratio:

[0008] in, for The number of secondary users in the initial fusion of time slots, Let be the credibility of the i-th secondary user in time slot t. for Statistical data of the initial fusion of time slots For the local perception result of the i-th secondary user in time slot t, This indicates that the primary user signal is present. This indicates that the primary user signal does not exist. This represents the probability of event A occurring given that event B has occurred.

[0009] Furthermore, Preliminary fusion of secondary users in time slots The dynamic adjustment is based on the previous time slot, and its expression is:

[0010] in, For t The final number of secondary users fused in slot 1, where c is the reduction step size. {1,2,3} To initially integrate the lower limit of the number of secondary users, N represents the total number of secondary users in the network.

[0011] Furthermore, the credibility of secondary users The expression is:

[0012] in, The probability of selection by the current secondary user. This represents the average probability of the initial fusion set.

[0013] Furthermore, in S4, the weighted likelihood ratio expression after adding secondary users is:

[0014] in, To supplement the updated data of secondary users, a weighted likelihood ratio decision statistic that incorporates data from n+1 secondary users is used. This is a weighted likelihood ratio decision statistic that incorporates data from n secondary users.

[0015] Furthermore, in S4, when the fusion of all secondary users still fails to meet the decision termination condition, the single-node fusion decision result of the secondary user with the highest selection probability is output to ensure normal operation under extreme attacks.

[0016] Furthermore, in step S5, the expression for the updated learning parameters is:

[0017] in, Let be the learning parameters of the i-th secondary user participating in the fusion in time slot t. For the t-slot fusion decision result, For the i-th secondary user participating in the fusion, the local perception result is... Let this be the probability of selection by the secondary user. As the reward weight, 0.01≤ ≤0.05, For penalty weights, 0.01 ≤ ≤0.03, This is a penalty factor for malicious attacks.

[0018] Furthermore, the malicious attack penalty factor ε is determined by the number of secondary users participating in the current time slot. Compared with the historical average of sliding windows The difference determines when Greater than hour, ,when Not greater than When ε=0.

[0019] The beneficial effects of this invention are reflected in: 1. This invention dynamically selects highly reliable secondary users to participate in decision-making by using a variable number of participating secondary users and step-by-step SPRT fusion. This can minimize the number of participating secondary users while ensuring decision-making accuracy, thereby reducing system energy consumption and perception latency.

[0020] 2. Based on the penalty mechanism of participation number deviation, this invention can accurately identify probabilistic SSDF attacks and has strong anti-malicious interference capabilities; by suppressing the impact of channel fading through a sliding window, it reduces misjudgments by normal secondary users and improves the robustness of non-stationary channels. Attached Figure Description

[0021] The accompanying drawings, which are provided to further illustrate this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application.

[0022] Figure 1 This is a schematic diagram of the overall process of the method according to an embodiment of the present invention.

[0023] Figure 2 This is a detailed flowchart of the method according to an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of the system model according to an embodiment of the present invention.

[0025] Figure 4 This is a graph comparing the probability of spectral collisions under different algorithms.

[0026] Figure 5 This is a curve comparing the average secondary user participation of the method in this embodiment of the invention with that of the traditional method. Detailed Implementation

[0027] 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 a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. 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.

[0028] It should be noted that the meaning of "and / or" throughout the text includes three parallel solutions. Taking "A and / or B" as an example, it includes solution A, solution B, or a solution that simultaneously satisfies A and B. Furthermore, "multiple" refers to two or more. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0029] See Figures 1-2 This invention provides a cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio testing, comprising the following steps: S1. Each secondary user performs local energy detection and reports the detection results to the fusion center; S2. The fusion center calculates the selection probability of each secondary user based on the Boltzmann distribution and learning parameters, and randomly selects secondary users based on the selection probability to form a preliminary fusion set; S3. Perform preliminary sequential probability ratio test (SPRT) fusion on the preliminary fusion set. The SPRT fusion calculates the weighted likelihood ratio as the decision statistic based on the variable sample size decision mechanism. S4. Regarding the judgment statistics, if they fall within the interval... If it falls outside the specified range, it is considered to have met the conditions for termination of the judgment. Within a certain range, secondary users are dynamically added and the weighted likelihood ratio is updated until the termination condition is met or all secondary users are traversed. and This is the decision threshold; S5. Output the final fusion decision result based on the final weighted likelihood ratio and complete the parameter update for the next time slot: Based on the final fusion decision result, perform learning parameter reward and punishment update by combining the deviation between the current number of participating secondary users and the historical average, and maintain the historical average number of participating secondary users through a sliding window.

[0030] In S1-S2, establish as follows Figure 3 The system model shown consists of a primary user (PU), N secondary users (SUs), and a fusion center (FC). The fusion center is fully trusted and is responsible for secondary user selection, data fusion, decision output, and learning parameter updates. Secondary users only perform local sensing and result reporting and do not participate in global decision-making. Some secondary users are malicious, while the rest are normal users. In each sensing cycle, SUs first perform local sensing of the PU's authorized channel status, and then send the local sensing results to the FC via the control channel. The FC fuses the sensing results and obtains the final decision. If the final decision PU is idle, the SUs attempt to access the authorized channel; if the final decision PU is busy, the SUs do not access it. It should be noted that the control channel is configured to transmit sensing results without errors.

[0031] Local sensing by SUs employs an energy detection method that does not rely on prior information from PU signals. The i-th secondary user SU... i Local detection probability and local false alarm probability It is expressed as follows:

[0032]

[0033] in, express, For SU i The results of local perception, A value of 0 or 1 indicates that the local sensing PU status is... or , This indicates that the PU signal is absent and the spectrum resources are idle. This indicates that the PU signal is present and the spectrum is occupied. This is an energy detection statistic. This is the energy detection decision threshold.

[0034] Malicious users employ probabilistic SSDF attacks, meaning they use [a specific method / mechanism] when reporting the perception results to the FC. Probability-based reporting of results that contradict local perception, in order to Report the actual data based on probability.

[0035] After receiving the secondary user perception results, the FC randomly selects results based on the selection probability for SPRT fusion, SU i The selection probability is calculated using the Boltzmann distribution as follows:

[0036] in, To select the lower probability limit, Let be the learning parameters for the i-th secondary user in time slot t. Let be the selection probability of the i-th secondary user in time slot t; Obtained during system standby, and then based on the secondary user SU i The estimated time-series difference error is updated.

[0037] In S3-S4, during the t-th sensing time slot, FC first determines the number of secondary users based on the selection probability and the initial fusion of secondary users. Random selection The data from each secondary user were initially fused, and statistics were calculated. SPRT fusion adopted a variable sample size decision mechanism, and the decision statistic was the weighted log-likelihood ratio.

[0038] in, Let t be the credibility of the i-th secondary user in time slot t. for Statistical data of the initial fusion of time slots For the local perception result of the i-th secondary user in time slot t, This indicates that the primary user signal is present. This indicates that the primary user signal is not present. This represents the probability of event A occurring given that event B has occurred.

[0039] Preliminary fusion of secondary users in time slots The dynamic adjustment is based on the previous time slot, and its expression is:

[0040] in, For t The final number of secondary users fused in slot 1, where c is the reduction step size. {1,2,3} To initially integrate the lower limit of the number of secondary users, N represents the total number of secondary users in the network.

[0041] Secondary user credibility The expression is:

[0042] in, The probability of selection by the current secondary user. This represents the average probability of the initial fusion set.

[0043] SPRT involves two opposing assumptions: the null hypothesis and alternative hypotheses In this method, represent the idle and busy states of the PU, respectively. The SPRT decision termination condition is: after each sampling, if the statistic... Less than the decision threshold If the sampling stops and the null hypothesis is accepted, then sampling is stopped. Output decision results If the statistic Greater than the decision threshold If so, then reject the null hypothesis and accept the alternative hypothesis. Stop sampling and output the decision result. If the statistic Between and If sampling continues between these points, a decision will be made. The decision threshold is... and The calculation method is as follows:

[0044]

[0045] in, and These represent the probabilities of Type I and Type II errors, respectively, i.e., the acceptable global false alarm probability and the global missed detection probability.

[0046] Statistics for preliminary integration The termination condition is checked. If the termination condition is met, the decision result D(t) = 0 or 1 is output, corresponding to PU. or If further sampling is needed, data from one of the remaining sub-users that were not selected will be randomly selected based on the selection probability. Update the statistics:

[0047] in, This is the updated statistic. (Regarding the updated statistic...) Perform a termination condition check. Repeat the above steps.

[0048] If a decision still cannot be made after integrating all secondary user data, that is... Then D(t) takes the result of the secondary user with the highest selection probability to ensure normal operation under extreme attacks.

[0049] In the S5, at the first Each perception cycle, the set of secondary users participating in SPRT fusion decision-making. For the common At the end of the perception phase, the learning parameters of the secondary users participating in the fusion are updated as follows:

[0050] in, Let be the learning parameters of the i-th secondary user participating in the fusion in time slot t. For the t-slot fusion decision result, For the i-th secondary user participating in the fusion, the local perception result is... Let this be the probability of selection by the secondary user. For reward weighting, 0.01 ≤ ≤0.05, For penalty weights, 0.01 ≤ ≤0.03, This is a penalty factor for malicious attacks; The malicious attack penalty factor ε is determined by the number of secondary users participating in the current time slot. Compared with the historical average of sliding windows The difference determines when Greater than hour, ,when Not greater than At that time, ε=0; Using a fixed-length sliding window to retain the information from the k most recent time slots, the following calculations are performed:

[0051] in, This represents the number of participating secondary users in the i-th time slot.

[0052] Figure 4 The chart shows the comparison curves of spectral conflict probabilities under different algorithms. OLSPRT, OL, and WSPRT represent: Online Learning with Stochastic Pruning; Online Learning; and Weighted Sequential Probability Ratio Test, respectively. Figure 4 As can be seen, the method of this invention can improve the accuracy of perception and reduce the probability of spectrum conflict.

[0053] Figure 5 The diagram shows a comparison curve of the mean values ​​of participating secondary users using the method of this invention and the traditional method. Here, SNR represents the signal-to-noise ratio; a higher SNR indicates better signal quality. Figure 4 and Figure 5 As can be seen, compared with the fixed number of secondary users algorithm, the method of the present invention can find the optimal number of participating secondary users according to the current environment, and increase the amount of fused data when the local perception performance deteriorates, so as to improve the accuracy of the fusion result.

[0054] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the above-described cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio testing.

[0055] This invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the above-described cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test.

[0056] This invention also provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the steps of the above-described cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio testing.

[0057] It is understood that the systems, devices, and storage media provided in the embodiments of the present invention correspond to the methods provided in the embodiments of the present invention, and the explanations, examples, and beneficial effects of the relevant content can be referred to the corresponding parts of the above-mentioned cooperative spectrum sensing method against SSDF attacks based on online learning sequential probability ratio test.

[0058] It should be noted that those skilled in the art will understand that all or part of the steps implemented in the embodiments of the present invention can be implemented entirely or partially by software, hardware, firmware, or any combination thereof. When implemented in hardware, it can be implemented entirely or partially by purchasing standard parts or modifications. When implemented in software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid state disks (SSDs)).

[0059] In summary, this invention addresses the imbalance between attack resistance and energy efficiency, as well as poor channel robustness, in existing anti-SSDF attack methods by providing a cooperative spectrum sensing method based on online learning sequential probability ratio testing. Specifically, this method first involves secondary users performing energy detection and reporting the results to a fusion center. Then, the fusion center randomly selects secondary users based on a Boltzmann distribution and employs a step-by-step SPRT fusion strategy of initial fusion followed by dynamic addition, calculating the weighted likelihood ratio to make a decision. Finally, a reward and punishment mechanism is designed based on the deviation between the number of participating secondary users and the historical mean. Historical information is maintained through a sliding window, achieving adaptive optimization of secondary user credibility and participation scale. This invention can dynamically adapt to channel fading and probabilistic SSDF attacks, reducing the number of participating secondary users and system energy consumption while maintaining sensing accuracy, significantly improving sensing efficiency and attack robustness in scenarios with low signal-to-noise ratio and high malicious activity.

[0060] It should be understood that the examples and embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Those skilled in the art can make various modifications or changes based on them. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the protection scope of the invention.

Claims

1. A cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test, characterized in that, Includes the following steps: S1. Each secondary user performs local energy detection and reports the detection results to the fusion center; S2. The fusion center calculates the selection probability of each secondary user based on the Boltzmann distribution and learning parameters, and randomly selects secondary users based on the selection probability to form a preliminary fusion set; S3. Perform preliminary sequential probability ratio test (SPRT) fusion on the preliminary fusion set. The SPRT fusion calculates the weighted likelihood ratio as the decision statistic based on the variable sample size decision mechanism. S4. Regarding the judgment statistic, if it falls within the interval... If it falls outside the specified range, it is considered to have met the conditions for termination of the judgment. Within a certain range, secondary users are dynamically added and the weighted likelihood ratio is updated until the termination condition is met or all secondary users are traversed. and This is the decision threshold; S5. Output the final fusion decision result based on the final weighted likelihood ratio and complete the parameter update for the next time slot: Based on the final fusion decision result, perform learning parameter reward and punishment update by combining the deviation between the current number of participating secondary users and the historical average, and maintain the historical average number of participating secondary users through a sliding window.

2. The cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test as described in claim 1, characterized in that, In S3, the sequential probability ratio test SPRT fusion adopts a variable sample size decision mechanism, and the decision statistic is the weighted log-likelihood ratio: in, for The number of secondary users in the initial fusion of time slots, Let be the credibility of the i-th secondary user in time slot t. for Statistical data of the initial fusion of time slots For the local perception result of the i-th secondary user in time slot t, This indicates that the primary user signal is present. This indicates that the primary user signal does not exist. This represents the probability of event A occurring given that event B has occurred.

3. The cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test as described in claim 2, characterized in that, Preliminary fusion of secondary users in time slots The dynamic adjustment is based on the previous time slot, and its expression is: in, For t The final number of secondary users fused in slot 1, where c is the reduction step size. {1,2,3} To initially integrate the lower limit of the number of secondary users, N represents the total number of secondary users in the network.

4. The online learning-based sequential probability ratio test-based anti-SSDF attack cooperative spectrum sensing method as described in claim 2, characterized in that, Secondary user credibility The expression is: in, The probability of selection by the current secondary user. This represents the average probability of the initial fusion set.

5. The online learning-based sequential probability ratio test-based anti-SSDF attack cooperative spectrum sensing method as described in claim 2, characterized in that, In step S4, the weighted likelihood ratio expression after adding secondary users is: in, To supplement the update of secondary users, a weighted likelihood ratio decision statistic that incorporates data from n+1 secondary users is used. This is a weighted likelihood ratio decision statistic that incorporates data from n secondary users.

6. The cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test as described in claim 1, characterized in that, In step S4, if the fusion of all secondary users still fails to meet the decision termination condition, the single-node fusion decision result of the secondary user with the highest selection probability is output to ensure normal operation under extreme attacks.

7. The online learning-based sequential probability ratio test-based anti-SSDF attack cooperative spectrum sensing method as described in claim 1, characterized in that, In S5, the expression for the updated learning parameters is: in, Let be the learning parameters of the i-th secondary user participating in the fusion in time slot t. For the t-slot fusion decision result, For the i-th secondary user participating in the fusion, the local perception result is... Let this be the probability of selection by this secondary user. As the reward weight, 0.01≤ ≤0.05, For penalty weights, 0.01 ≤ ≤0.03, This is a penalty factor for malicious attacks.

8. The cooperative spectrum sensing method for resisting SSDF attacks based on online learning sequential probability ratio test as described in claim 7, characterized in that, The malicious attack penalty factor ε is determined by the number of secondary users participating in the current time slot. Compared with the historical average of sliding windows The difference determines when Greater than hour, ,when Not greater than When ε=0.

Images