Research On Channelization Techniques For Software Defined Radio

Channelization, as a key technology and component of wideband receiver and Software Defined Radio (SDR), is used to extract and separate a subband signal or a group of parallel subband signals contained within the received bandwidth for follow-up baseband processing. The primary functions of channelization usually comprise Digital Down Conversion (DDC), filtering, Sample Rate Conversion (SRC), despreading for spectrum-spread systems and so on. How to efficiently implement channelization has become a hot topic of research in the communication field. The main work and contributes offered in this dissertation is focusing on key technologies of channelization including single-channel channelization, multi-channel channelization and high-efficient architectures of SRC. The contents are organized as follows:In the aspect of single-channel receiving, wideband DDC and the related efficient structure based on polyphase filtering are discussed in depth. The kernel reason for its efficiency is analyzed and the existing problems to be solved are also pointed out. Based on above discussion an architecture based on Goertzel filter is proposed and the corresponding mathematical derivation is also provided. Due to ability of relieving the constraints on tuning frequency of mixer, the proposed structure can eliminate the blind zones of currently existing structures. Performance comparison via simulations has shown that the proposed structure is of relatively low computational complexity and finite cost in hardware.In the aspect of multi-channel receiving, channelization technique based on modulated filter bank for subband signals with non-uniform distribution and bandwidth is studied. First, the basic principles and the implementation architecture of non-uniform channelization processing based on modulated filter bank of perfect reconstruction are introduced. And then a modulated filter bank of nearly perfect reconstruction is designed, and the corresponding topics of blind channelization and dynamic channelization in a multi-standard wideband satellite link are discussed. Thanks to the use of nearly perfect reconstructive filter bank, the errors of the reconstructed signals are reduced remarkably and can be controlled within a very small range as long as a high-efficient prototype filter has been designed. Compared with the conventional method based on the parallel DDC of single-channel, the performance of new method is improved both in computational complexity and in hardware cost.Targeting at simplification of implementation and reduction of hardware cost, a preliminary study on the design and implementation of non-uniform filter bank is carried out for channelization-based receiving of subband signals with different bandwidths and arbitrary locations in frequency domain. The finally developed architecture is optimized by the combination of channelization with sampling rate conversion. And the corresponding blind channelization and dynamic channelizationthen in the multi-standard wideband satellite link are discussed. The advantages of the proposed method are the simplicity in implementation and low cost in hardware. New structure is rather flexible if the minimal guard band and maximal bandwidth of the subband signals are given beforehand. SRC in SDR is also discussed, and two improved structures of CIC filter are presented. The first one is a CIC decimation filter based on polyphase structure and partially sharpening technique, which is suitable for the situation where the requirements in passband attenuation is not very high while the requirements in alias suppression and operating speed are relatively high. The other one is an improved CIC-Rotated Sinc decimator which possesses improved frequency response, low operating rate and low cost in hardware as well.