| With the rapid development of communication technology, the spectrum isbecoming a scarce resource, and with the growing demand for communicationsservices, the effective use of spectrum resources has become an urgent problem.Cognitive radio is proposed to solve the rational allocation and utilization ofspectrum resources. Non-Contiguous Orthogonal Frequency DivisionMultiplexing (NC-OFDM) as a special kind of Orthogonal Frequency DivisionMultiplexing (OFDM), not only has the advantages of high spectrum utilizationand strong ability of anti-multipath interference, and because of using consecutivesub-carriers for data transmission, making it a common transmission mode ofcognitive radio. Channel estimation is one of the key technologies of theNC-OFDM system, but currently there is a lack of in-depth study.In recent years, compressed sensing has become the hotspot in the field ofsignal processing. The theory suggests that with the sampling rate below theNyquist sampling rate and at the receiving side the sparse signal can bereconstructed. In this paper, channel estimation of NC-OFDM system based oncompressed sensing will be investigated and studied.The principle and characteristics of the NC-OFDM system is first introduced,and the system model is described in detail, then a system simulation platform isbuilt with MATLAB. Channel estimation techniques are the focus point, includingthe design of pilot pattern, the estimation algorithm at the receiver andinterpolation algorithm. A systematic study of the compressed sensing theory byintroducing the theoretical framework of compressed sensing, contrast with thetraditional information acquisition system based on the Nyquist sampling theorem,the advantages of compressed sensing algorithm is obvious. Some keytechnologies of compressed sensing are focused on, including the signal sparserepresentation, the design of observation matrix, and signal reconstructionalgorithm. On the basis of fully understanding the theory, SAMP (SparsityAdaptive Matching Pursuit) is proposed in the NC-OFDM system channelestimation module, then the algorithm is simulated in the NC-OFDM systemsimulation platform, the analysis and comparison for estimated performance of thechannel estimation based on SAMP algorithm and traditional algorithm withsimulation results is taken. The superiority of channel estimation based oncompressed sensing is pointed out, that channel estimation algorithm based oncompressed sensing can get better performance with fewer pilots and improve the effectiveness of the system spectrum. At the same time, compared with the otherchannel estimation algorithms based on compressed sensing, the SAMP algorithmshows a better estimation performance. Because the algorithm does not require apriori knowledge of the channel sparsity, it’s more suitable for practicalapplications. |
