Research On Energy Detection Algorithm Based On Noise Power Uncertainty And User Status Changes

With the rapid development of wireless communication technologies and services, thedemand for wireless spectrum resources is increasing. However, the lack of radiospectrum is one of the bottlenecks for the further development of wirelesscommunication. Cognitive Radio (CR) by opportunistically accessing the licensed idlespectrum can improve the utilization of the existing radio spectrum, and it issignificantly important to alleviate the spectrum shortage problem. Energy detection hasbecome the most common method of idle spectrum detecting because of its lowcomputational and implementation complexities. But its performance may sufferseriously from the Noise Power Uncertainty (NPU), user’s status changes randomlyduring the sensing period and the low Signal-to-Noise (SNR) environment. In this thesis,the key challenges of spectrum sensing mentioned above are studied, and the maincontributions are as follows.For the performance of energy detection is affected by NPU, a low computationalalgorithm is proposed to estimate the NPU interval, and the SNR WALL deteriorationphenomenon with estimated noise power is analyzed theoretically. As a result, the SNRWALL deterioration theorems are obtained. In addition, a new energy detectionalgorithm based on modified threshold is proposed to eliminate SNR WALLdeterioration. Numerical simulation results show that the proposed algorithm canestimate accurately the NPU interval, and verify the correctness of the SNR WALLdeterioration theorems. Furthermore, both analytical and simulation results show thatthe proposed energy detection under NPU outperforms the scheme of Robust StatisticsApproach (RSA). The SNR WALL deterioration can be reduced effectively, henceimproving the robustness of detection.The classical DC-MAC is integrated with energy detection at the PHY layer in thisthesis, and the sensing result is not reliability when multiple SU detects the samespectrum simultaneously. To achieve the desired detection probability in the low SNRenvironment, the Minimum Sampling Time (MST) is deduced for energy detection.Based on MST, a new Optimized DC-MAC (ODC-MAC) protocol is proposed to solvethe problem that amends the aforementioned problems of DC-MAC with energydetection. ODC-MAC jointly considers the spectrum sensing and access, and can improve the data transmission reliability for SU by cross-layer cooperation. It’s shownthat the simulation results match the theoretical analyses very well. The proposedODC-MAC can improve the data transmission reliability and enhance the throughputcompared to the traditional DC-MAC.For the spectrum sensing performance is significantly degraded by the PU statuschanges as arriving or leaving randomly and in a low SNR environment. A newweight-p energy detection is presented to improve the detection performance in thesesituations. To reduce the implementation complexities and save the energy consumption,the optimum weight operation of weight-p energy detection is modeled as MSToptimization problem, and the optimum and sub-optimal weights are obtained duringour analyses. Numerical simulation results show that the proposed weight-p energydetection offers better detection performances with a reduced probability of false alarmand compress sensing time compared to conventional energy detection for the samedetection performance.The performance of spectrum sensing can be significantly degraded by user (PU orSU) randomly arriving. To counter the aforementioned problem, a new feedbackadditional energy detection is proposed. The instantaneous energy of the samplingpoints in the later part of the sensing period is added to the former ones, as a result, theenergy statistics of the sensing period has been improved without the need forprolonging the sensing time. The simulation results show that the algorithm proposed bythis thesis not only significantly outperforms the existing schemes but also reduces theprobability of data collisions. Therefore, the throughput of SU can be enhanced.