【正文】 、天文、生物醫(yī)學(xué)工程等眾多領(lǐng)域。功率譜的估計大致可分為經(jīng)典功率譜估計和現(xiàn)代功率譜估計，針對經(jīng)典譜估計的分辨率低和方差性能不好等問題提出了現(xiàn)代譜估計，AR模型譜估計就是現(xiàn)代譜估計常用的方法之一。例如在現(xiàn)代軍事領(lǐng)域中，人們通過分析雷達信號，根據(jù)回波信號的功率譜密度、譜峰的寬度、高度與位置，可以確定運動目標的位置、輻射強度和運動速度。現(xiàn)代譜分析受算法的復(fù)雜度高和集成電路處理速度的影響，因此在引信系統(tǒng)中的應(yīng)用會受到很大的限制。研究內(nèi)容① 熟悉譜估計的發(fā)展歷程； ② 熟練掌握經(jīng)典譜估計方法：直接法和間接法、它們之間的關(guān)系、估計質(zhì)量以及估計性能比較； ③ 熟練掌握現(xiàn)代譜估計方法：信號建模、AR模型參數(shù)求解的LevinsonDurbin算法和BURG算法，階數(shù)的確定方法和原則，穩(wěn)定性以及對信號建模的討論； ④ 熟悉熟練使用MATLAB仿真。擬采取的研究方法、技術(shù)路線、實驗方案及可行性分析熟悉經(jīng)典譜估計和現(xiàn)代譜估計相關(guān)的理論知識，并掌握MATLAB的使用方法，達到熟練使用并融會貫通。最后總結(jié)歸納實驗結(jié)果，提出自己的觀點。2010年1月11日2010年3月5日：進行畢業(yè)實習(xí)，并填寫畢業(yè)實習(xí)報告。2010年3月15日2010年3月28日：學(xué)習(xí)并翻譯一篇與畢業(yè)設(shè)計相關(guān)的英文材料。2010年4月12日2010年4月18日：GUI設(shè)計。預(yù)期成果：了解譜估計的發(fā)展進程，掌握各種算法，并熟練掌握MATLAB的基本操作。特色或創(chuàng)新之處由于經(jīng)典譜估計中將數(shù)據(jù)工作區(qū)外的未知數(shù)據(jù)假設(shè)為零，相當(dāng)于數(shù)據(jù)加窗，導(dǎo)致其分辨率降低。而現(xiàn)代譜估計則不再簡單地將觀察區(qū)外的未知假設(shè)為而零，而是先將信號的觀測數(shù)據(jù)估計模型參數(shù)，按照求模型輸出功率的方法估計信號功率譜，回避了數(shù)據(jù)觀測區(qū)以外的數(shù)據(jù)假設(shè)問題，因而其譜的分辨率得到提高。已具備的條件和尚需解決的問題已具備的條件：可通過書籍和網(wǎng)絡(luò)了解功率譜估計相關(guān)的理論知識，并已學(xué)習(xí)過MATLAB程序設(shè)計。指導(dǎo)教師意見該生查閱了大量的相關(guān)資料，設(shè)計方案可行，同意開題。 for example, operations such as division and squareroot are often necessary. These arithmetic processes exhibit munication bottleneck and their hardware implementation can be inefficient when used in conjunction with multipliers. A programmable, bitserial, multiplier/divider, which overes the bottleneck problems by using a data interleaving scheme, is introduced in this paper. This interleaved processor is used to show how the parametric Modified Covariance spectral estimator can be efficiently routed on a field programmable gate array for realtime applications. 1. INTRODUCTIONDue to its ease of hardware and software implementation the shortterm fast Fourier transform(STFFT)is widely used for spectral estimation and is known as the conventional method. However, the technique has drawbacks in terms of spectral resolution and accuracy caused by the finite length of the input data sequence used. Windowing of input data causes spectral broadening and Gibb’s phenomenon of spectral leakage can mask the weaker frequency ponents of the true power spectral density(PSD)[1]. These unwanted effects can be reduced by using longer data sequence lengths, so that the transformed signal bees a better representation of the infinite data sequence, but in real life this usually is not feasible as the characteristics of the input data may change with time. Over short periods of time the data signals can often be assumed to exhibit widesense stationarity, where the signal characteristics are assumed approximately constant but the spectral resolution is therefore limited. In attempts to improve the PSD estimation, windowing functions, Bartlett or Hanning for example, can be used to reduce sidelobe levels but these lower spectral resolution by broadening the main lobe of the PSD[2].Model based, parametric spectral estimation techniques can alternatively be used, where the unrealistic assumption that data is zero outside the window of interest is dropped[1]. Either knowledge of the underlying process or reasonable assumptions about the nature of the unobserved data are used to improve frequency resolution over the conventional approaches. The putational burden of such processors is however much higher than the STFFT and arithmetic functions such as division and squareroot often bee necessary. In the division and squareroot nonrestoring algorithms there is an inherent dependency that the result bits must be puted in a most significant bit(MSB)first manner, with the putation of a bit directly dependent upon the result of the previous one[3]. This interdependency makes it difficult to efficiently realize such arithmetic functions in hardware, and implementations are usually much slower than other basic functions such as multiplication, addition and subtraction. Communication bottlenecks can therefore easily occur in systolic arrays where different types of processors are interconnected.The difficulties with hardware implementation of parametric spectral estimators have led to a preference of software implementation on homogeneous DSP networks[4]. However, high levels of processing capacity have not been fully reflected in system throughput since the increased munication incurred as a result of parallelism is constrained by munication bus performance. This restricts the range of problems that can be puted in realtime and the software approach may sometimes be inadequate for realtime spectral estimation.In this paper, hardware implementation of a parametric spectral estimator is addressed. A bitserial processor capable of division and inner product step putation is developed by bining separate processors for these functions. The design uses a high level of pipelining so that division can be puted at a high rate and multiplication is performed on a MSB first data stream, eliminating the bottleneck problem. The high level of pipelining allows many independent putations to be performed simultaneously or interleaved. The use of the interleaving scheme is demonstrated by implementing the design of a Modified Covariance type of parametric spectral estimator, to produce a field programmable gate array(FPGA)based system for the spectral analysis of Doppler signals from ultrasonic blood flow detectors.2. MODIFIED COVARIANCE SPECTRAL ESTIMATIONThe model order p=4 Modified Covariance(MC)spectral estimator, proven to be optimally cost efficient for the blood flow application where mean velocity and flow disturbance are of interest[5], involves solving the following linear system of covariance matrix equations: (1) where each element is obtained from: (2) for a window of length N data samples. The filter parameter estimates are obtained by solution of the linear system(1), using the Cholesky, forward elimination and back substitution algorithms. The signal white noise variance estimate, is calculated as: (3) and the power spectral de