Dynamic Channelization Techniques For Broadband Signal Reconnaissance

Channelization technique is one of the key techniques in broadband signal reconnaissance, which is used to extract sub-band signals from the received bandwidth. In order to satisfy the requirements of multi-system, multi-standard communications, the IF bandwidth of software radio system usually contains a number of sub-band signals with different bandwidth and sub-band signals with non-uniform distribution. The parameters such as numbers and bandwidth of sub-band signals may change during receiving process. This dissertation studies the dynamic channelization methods with unknown sub-band signal parameters. The work finished in this thesis is a part of the overall task targeting at a key project undertaken by the laboratory the author works with. The main work and contributions are summarized as follows:The spectrum detection of the broadband signals is completed. The power spectrum is obtained by shaping the segmented signals with overlapping window function. Based on which, the affection of the varying energy of noise in different frequency is eliminated by using curve fitting methods. A variable step difference searching method is proposed. The frequency distribution of each sub-band signal is estimated exactly combining with SNR threshold.Channelization is based on the perfect reconstruction filter banks. By using prototype filters having a Square-Root cosine-roll-off (SRCR) transition band to design perfect reconstruction modulated filter banks, a new method is proposed, in which the prototype filter banks are designed using a second-order cone programming (SOCP) algorithm. With the prototype filter designed in this paper, the flatness condition required for perfect reconstruction filter banks can be better satisfied. Thus, perfect reconstruction filter banks with lower restructuring errors are designed.The structure, which completes the dynamic channelization of the broadband signals with non-uniform distributed sub-bands, is studied to meet the variation of the sub-band signals during receiving process. The channel number of filter banks is analyzed and determined by the spectrum detection. Based on which, the broadband signals are segmented uniformly by analysis filter banks. Then the reconstruction filter banks are designed according to the distribution of the sub-band signals, and the adjacent channel is reconstructed to recover sub-band signals. When the state of sub-band signals is changed, the channel number of the analysis filter banks and reconstruction filter banks is analyzed, and the sub-band signals are re-channelized according to the new state. Thus, the dynamic channelization is realized.The sample rate needs to be changed to further analyze the output signals of channelization. Based on the characteristic of the output signals, the arbitrary sample rate converting factor is decomposed, and is realized by hierarchical structure. It is shown from simulation that all stages of this scheme are stable with low complexity, which ensure its application in sample rate conversion of output signals of channelization. Meanwhile, a new scheme is proposed combing with the structure of channelization, which reduce the complexity dramatically with approximate processing.