On Spectrum Sensing And Jamming Detection In Cognitive Radio

As the rapid development in wireless communication technology, corresponding de-vices and applications also proliferate. Meanwhile, the wireless spectrum resources are get-ting more and more rare and are unable to meet the growing demands of users. On the other hand, under current spectrum allocation mechanism, plenty of spectrum resources licensed to specific users are largely under-utilized. A great amount of licensed spectrum resources are wasted due to the absence or inactivity of the primary users. To solve this, in1999, Dr. Mitola proposed cognitive radio which enables secondary users access the idle licensed spectrum by finding spectrum holes so that the spectrum utilization can be improved and the situation of spectrum shortage can be mitigated.Spectrum sensing is a fundamental and important function in cognitive radio based systems. The efficiency of cognitive radio relies on the correct spectrum sensing results. Hence, research on improving the reliability of spectrum sensing is a popular topic in this area. Jamming is a common problem in wireless communications. In a cognitive radio sys-tem, both the primary user and the secondary user will be affected if there exists a jammer. A complete cognitive radio system is required to be able to detect jamming to ensure it can work normally in all kinds of wireless environments. For the above issues, this dissertation discusses our work from the following three aspects:1. When a cognitive radio performs cooperative spectrum sensing, applying a stat-ic data fusion rule and ignoring the actual network parameters often lead to low system reliability and large system cost. From this point of view, this dissertation proposes a data fusion rule selection algorithm for cooperative spectrum sensing. This algorithm first selects some typical data fusions from the most commonly used ones and the selected rules form the rule database. The rules in the rule database are then tested via simulations. An algorithm which selects the best-performing data fusion rule is proposed according to the current network conditions. The unknown parameters in the algorithm are estimated via Genetic Algorithm. The simulation results show that, compared with using a single and static data fusion rule, the proposed algorithm is able to provide higher spectrum sensing accuracy. Besides, bound with a trust value calculation method of low complexity, the proposed algo-rithm can output similar sensing accuracy as schemes of high complexity while its response time is much shorter.2. In reputation based cooperative spectrum sensing, the data fusion center will de-grade the trust value of a secondary user ignoring the real reason if it reports an erroneous local sensing result. This mechanism cannot let the calculated trust val-ue reflect the malice of the node. From this perspective, this dissertation proposes a new trust based cooperative spectrum sensing method. The proposed method first calculates the the sensing capabilities of the secondary users based on their channel status. Based on the capability values, the trust value of a node is then derived via maximum likelihood estimation. The data fusion center considers both the capability and the trust values of a secondary user to output the final sensing re-sult. According to the simulation results, the trust value calculated by the proposed scheme can better capture the malice of a node. The spectrum sensing accuracy can be significantly improved by the proposed scheme under spectrum sensing data falsification. Compared with existing schemes, the proposed scheme can provide higher sensing reliability under a series of different scenarios and the accuracy can be improved by as much as38%.3. Among all jamming classes, reactive jamming is considered as the most difficult one to detect. The cognitive radio system’s performance will be degraded if there is a reactive jammer attacking the primary base station. For this issue, this disser-tation proposes a joint reactive jammer detection and localization algorithm. This algorithm makes use of the cooperation among primary users, and builds a profile for each primary user based on the affection of the reactive jammer. The jamming detection is then performed by analyzing the similarity between the profile and the current data. With the same input, the proposed scheme can also perform the reac-tive jammer localization if the detection result is positive. Through thorough sim-ulations, the proposed scheme is able to provide fast and accurate detection. The true positive and true negative rates can simultaneously achieve over90%in the best situation. For more than70%of the cases the proposed scheme can position the jammer within one "hop" from its real location.To conclude, this dissertation proposes new ideas from the point of view of increasing the reliability of spectrum sensing and detecting and localizing a reactive jammer. It contributes to the current research from three different aspects while the former two focus on spectrum sensing and the latter one is about jamming detection.