图6和图7分别表示在“随机错误”攻击下虚警率和误检率随着MIDs数量的

Figures 6 and 7 respectively show the change of false alarm rate and false detection rate with the number of MIDs under the "random error" attack. Because under the "random error" attack, MIDs do not undergo a single data reversal, but reverse each perception report. Therefore, we use two indicators, false alarm rate and false detection rate, to detect the performance of our proposed algorithm under "random error" attacks. Through comparison, it is found that our method has a lower false alarm rate and false detection rate than the K-rank criterion and the weighted fusion scheme based on credibility, and can more accurately detect the existence of the primary user. The detection method is effective when MIDs account for less than 45% of the total, but as the number of MIDs continues to increase, the false alarm rate and false detection rate have increased significantly. Experimental results show that the opposite sensory data will cause greater harm to the fusion process. This is because when there are more than half of the attackers in the Internet of Things devices, "random errors" attackers affect the judgment of FC.

Fig. 6 and Fig. 7 show the changes of false alarm rate and false detection rate with the number of mids under the "random error" attack, respectively. Because under the "random error" attack, mids does not flip a single data, but reverses each perception report. Therefore, we use false alarm rate and false detection rate to detect the performance of our proposed algorithm under "random error" attack. Through comparison, it is found that our method has smaller false alarm rate and false detection rate than k-rank criterion and weighted fusion scheme based on reputation, and can more accurately detect the existence of primary users. When the number of mids is less than 45%, the detection method is effective, but with the continuous increase of the number of mids, the false alarm rate and false detection rate increase significantly. The experimental results show that the opposite sensing data will cause greater harm to the fusion process. This is because when there are more than half of attackers in the Internet of things devices, "random error" attackers affect the judgment of FC.<br>

Fig. 6 and fig. 7 respectively show the change of false alarm rate and false detection rate with the number of MIDs under "random error" attack. Because MIDs does not have a single data flip under the "random error" attack, but reverses each perception report. Therefore, we use false alarm rate and false detection rate to detect the performance of our proposed algorithm under "random error" attack. By comparison, it is found that the false alarm rate and false detection rate of our method are smaller than those of K-rank criterion and weighted fusion scheme based on credibility, and it is more accurate to detect the existence of primary users. The detection method is effective when MIDs accounts for less than 45% of the total amount, but with the continuous increase of MIDs number, the false alarm rate and false detection rate increase obviously. The experimental results show that the opposite sensing data will do more harm to the fusion process, which is because when there are more than half of attackers in IoT devices, "randomly wrong" attackers will affect FC's judgment.