Font Size: a A A

Research And Implementation Of The Waveform Component Technology Based On GPU

Posted on:2018-06-04Degree:MasterType:Thesis
Country:ChinaCandidate:N GuoFull Text:PDF
GTID:2348330542469879Subject:Electronic Science and Technology
Abstract/Summary:PDF Full Text Request
Software radio technology has been developed rapidly since it was proposed,and many countries,enterprises,universities and other research institutions have carried out in-depth study,and achieved a series of results.Although the CPU,DSP and FPGA-based waveform component development process is perfect,the problems also exist in the actual development and application.The CPU is powerful and versatile,but its capacity of data processing is inadequate,resulting in longer delay in the communication process,and affecting efficiency.The calculation of DSP and FPGA has been improved,but the development process of them is very complex,need higher requirements for developers,and the corresponding waveform components is targeted,but lack of versatility.GPU already has a strong computing power after years of development,and NVIDIA's CUDA platform makes GPU development becomes simple and convenient.But there is not much research work on the software radio technology in the GPU,and there is no more mature GPU-based waveform component development technology.In this case,this paper explores and studies the GPU-based waveform component development technology.Firstly,this paper studies the development method of CPU-based waveform components,especially the section of component model design and component generation,which is an indispensable step in the process of component development.On the basis,this paper analyzes the problems and shortcomings of CPU-based components,and further proposes to replace the CPU with GPU as the core processor for computing data,and improve the computing efficiency.Secondly,this paper studies the existing parallel computing model,analyzes the parallel computing model of multi-core CPU and GPU parallel computing model,and compares them from hardware structure,parallel computing mechanism and software development platform.GPU with its special structure,its computing power and storage bandwidth are significantly higher than the CPU's,but the difficulty of development is not bigger than the CPU's.This makes GPU-based waveform components in the implementation of a certain degree of feasibility.Then,according to the difference between GPU and CPU in the structure and software development process,this paper redesigns the GPU-based component development process,redefines the attributes and operations of the component interface,reconstructs the GPU component model.And this paper takes the QPSK communication process as an example,divides the modulation and demodulation process into four GPU components according to the GPU component development process,and realizes the QPSK waveform.Finally,QPSK waveform and USRP equipment together were built into a complete test platform,and were carried out the communication function test and communication performance test.The test results show that GPU-based waveform components can achieve the same function and have higher communication efficiency compared with CPU components under similar development conditions.Though this paper just explores the application of the GPU on the software radio,it also confirms that the scheme has certain feasibility.With the maturity of the technology,it will further reduce the waveform development periods,improve the communication efficiency,and promote the development of software radio technology.
Keywords/Search Tags:Software Defined Radio(SDR), waveform components, GPU, QPSK
PDF Full Text Request
Related items