Cooperative energy detection spectrum sensing method based on Lightgbm algorithm

[0059] Example

[0060] The simulation experiment and the performance comparison of the SVM algorithm verify the availability and feasibility of the present invention based on the Lightgbm algorithm to solve the problem of cooperative energy detection spectrum sensing. The present invention is set in a geographical location-based cooperative spectrum sensing model, there are 16 unauthorized users, they are evenly distributed in a 4*4 grid, in addition to the attached figure 2 There are also two authorized users in the system model. The two-dimensional geographic locations are (500m, 1500m) and (-1500m, 0m). The simulation parameters are set as follows: the sensing time period is 100, the bandwidth is 5MHz, the noise power spectral density is -174dBm, the transmission power of each authorized user is 200mW, the path loss coefficient is 4, the multipath fading and shadow fading coefficients are both 1, The probability of each authorized user going online is 0.5. The excellent performance of the linear kernel function in this problem has been proved in the previous work, so the kernel function of SVM is selected as the linear kernel function. The training vector is 200 and the test vector is 1200. Attached image 3 , 4 , 5 is a box plot of the detection rate, misclassification risk and misclassification rate of the SVM algorithm with the best performance compared to the previous SVM algorithm when the standard fixed false alarm rate is 0.1. The figure is hollow. The cabinet is SVM algorithm, and the black cabinet is Lightgbm algorithm. Attached image 3 The box plot is available. When the standard fixed false alarm rate is 0.1 and the authorized user transmit power is 50mW or 100mW, the signal-to-noise ratio of the unauthorized receiver is too low, and the average detection rate is already below 50%, generally at least 70% The above detection rate is of practical significance. When the authorized user's transmit power is 200mW, the detection rate requirement is met. It can be seen that the detection rate of Lightgbm algorithm is better than SVM at 200mW, 300mW, and 400mW. Correspondingly, Figure 4 , 5 The Lightgbm algorithm is also better than the SVM algorithm. Image 6 , 7 , 8, and 9 show that the ROC curve of the present invention is 100mW, 200mW, 300mW, and 400mW for authorized users compared to the SVM algorithm with the best performance before. From Image 6 It can be seen that, like the previous results, the signal-to-noise ratio is too low at 100mW, and the performance of the two algorithms cannot meet the requirements. Figure 7 , 8 It is the performance of the two algorithms when the authorized user transmit power is 200mW/300mW. When the false alarm rate is 0.1, the detection rate of the present invention is increased by 6% to 7%. Picture 9 It shows that when the authorized user's transmit power is large, that is, when the signal-to-noise ratio at the receiving end is high, the performance of the two algorithms is closer. All this proves that the present invention performs better than the SVM algorithm when the signal-to-noise ratio is low and meets the actual requirements, and has better practicability and feasibility. The present invention can be used to solve the spectrum sensing under the cognitive wireless network. problem.