In the 1950s, the development of sampled data system research and discrete system theory laid the mathematical foundation for digital signal processing. In 1965, J.W. Cooley and T.W. Tuji first proposed a fast algorithm for discrete Fourier transform (FFT), which greatly reduced the number of operations of discrete Fourier transform (DFT). This breakthrough has led to a major turning point in the concept and implementation of digital signal processing. At the same time, digital filtering theory, which uses computer approximation and simulation of analog filters, has also been developed. Fast Fourier transform and digital filtering theory form two pillars of digital signal processing. The emergence of large-scale digital integrated circuits provides favorable conditions for the realization of digital signal processing. In the mid-1970s, digital signal processing has become an independent discipline.
Digital signal processing has a wide range of applications. In terms of the characteristics of the signals it processes, it can be divided into speech signal processing and image signal processing. It has important applications in the field of communication engineering. For example, the application of digital filters instead of analog filters in communication equipment can make the equipment smaller and improve reliability. Fast Fourier transform and polyphase filter can realize multi-channel filter. Modulation and demodulation can be realized by sampling rate transform filtering. The application of adaptive filtering can realize channel equalization, echo cancellation, antenna array beamforming, etc. The application of nonlinear filtering can filter out the noise interference of the image. Therefore, digital signal processing technology plays an extremely important role in the development of communication technology.
In a broad sense, digital signal processing is a technical discipline that studies signal analysis, transformation, filtering, detection, modulation, demodulation and fast algorithms by digital methods. But many people think that digital signal processing is mainly about digital filtering technology, fast algorithm of discrete transformation and spectral analysis method. With the development of digital circuit and system technology as well as computer technology, digital signal processing technology has also been developed accordingly, and its application fields are very wide.
The applications of digital control and motion control mainly include disk drive control, engine control, laser printer control, inkjet control, motor control, power system control, robot control, high-precision servo system control, CNC machine tools, etc.
Applications for low-power, handheld devices and wireless terminals mainly include mobile phones, PDAs, GPS, data radio, etc.
digital filter
There are many practical types of digital filters, which can be roughly divided into finite impulse response type and infinite impulse response type, and can be realized in hardware and software. In hardware implementation, it is composed of adder, multiplier and other units, which is completely different from the analog filter composed of resistor, inductor and capacitor. The digital signal processing system is easily made of digital integrated circuits, showing the advantages of small size, high stability, programmability, etc. Digital filters can also be implemented in software. The software implementation method is to use a general digital computer to program the digital filter calculation according to the filter design algorithm.
Fourier transform
1965 J W. Cooley and T W. The graph base first proposes a fast algorithm for discrete Fourier transform, called Fast Fourier Transform for short, which is represented by FFT. With the fast algorithm, the number of discrete Fourier transform operations is greatly reduced, making it possible to implement digital signal processing. Fast Fourier transform can also be used for a series of related fast operations, such as correlation, convolution, power spectrum, etc. Fast Fourier transform can be made into special equipment or realized by software. Similar to FFT, other forms of transformation, such as Walsh transformation, number theoretic transformation, can also have their fast algorithms.
Spectral analysis
An analysis method describing signal characteristics in frequency domain can be used not only for deterministic signals, but also for random signals. The so-called deterministic signal can be expressed by a given time function, and its value at any time is certain; The random signal does not have such characteristics, and its value at a certain time is random. Therefore, random signal processing can only be analyzed and processed by statistical methods according to the theory of random process, such as using statistics such as mean value, mean square value, variance, correlation function, power spectral density function to describe the characteristics of random process or random signal.
In fact, most of the frequently encountered random processes are stationary and ergodic, so the average of its sample function set can be determined according to the time average of a sample function. Although stationary random signal itself is still uncertain, its correlation function is certain. When the mean value is zero, the Fourier transform or Z-transform of its correlation function can just be expressed as the power spectral density function of the random signal, generally referred to as the power spectrum. This feature is very important, so that fast transformation algorithm can be used for calculation and processing.
The data observed in practice is limited. This requires some estimation methods to estimate the power spectrum of the whole signal according to the limited measured data. For different requirements, such as reducing the deviation of spectral analysis, reducing the sensitivity to noise, improving spectral resolution, etc. Many different spectral estimation methods have been proposed. Among the linear estimation methods, there are periodogram method, correlation method and covariance method; Among the nonlinear estimation methods, there are maximum likelihood method, maximum entropy method, autoregressive moving average signal model method, etc. Spectral analysis and estimation are still under research and development.
Digital signal processing has a wide range of applications. In terms of the sources of the acquired signals, there are communication signal processing, radar signal processing, remote sensing signal processing, control signal processing, biomedical signal processing, geophysical signal processing, vibration signal processing, etc. According to the characteristics of the processed signal, it can be divided into speech signal processing, image signal processing, one-dimensional signal processing and multidimensional signal processing.
processing system
In any application, the required original signal must first be obtained through the process of information acquisition or data acquisition. If the original signal is a continuous signal, it must also be converted into a discrete signal through the sampling process, and then the binary digital signal that can be accepted by the digital computer or processor can be obtained through analog-to-digital conversion. If the collected data is already discrete data, binary numbers can be obtained by analog digital conversion only. The function of the digital signal processor is to properly process the digital signal sampled and converted from the original signal according to certain requirements, such as filtering requirements, to obtain the required digital output signal. After digital to analog conversion, the digital output signal is first converted into discrete signal, and then the discrete signal is connected into analog output signal through holding circuit. This processing system is suitable for various digital signal processing applications, but the special processor or software used is different.
Speech signal processing
Speech signal processing is one of the important branches of signal processing. It mainly includes speech recognition, language understanding, speech synthesis, speech enhancement, speech data compression, etc. Each application has its own special problems. Speech recognition is to extract the characteristic parameters of the speech signal to be recognized in real time, match with the known speech samples, and determine the phoneme attribute of the speech signal to be recognized. As for speech recognition methods, there are statistical pattern speech recognition, structure and sentence pattern speech recognition. These methods can be used to obtain formant frequency, tone, voice, noise and other important parameters. Speech understanding is the theoretical and technical basis for human computer dialogue with natural language. The main purpose of speech synthesis is to enable computers to speak. Therefore, first of all, it is necessary to study the change rule of voice feature parameters over time when pronouncing, and then use appropriate methods to simulate the process of pronunciation and synthesize into language. Other language processing problems also have their own characteristics. Speech signal processing is the basis for the development of intelligent computers and robots, and is the basis for the manufacture of vocoders. Speech signal processing is a rapidly developing signal processing technology.
Image signal processing
The application of image signal processing has penetrated into various scientific and technological fields. For example, image processing technology can be used to study the movement track of particles, the structure of biological cells, the state of landforms, the analysis of meteorological clouds, and the composition of cosmic stars. In the practical application of image processing, remote sensing image processing technology, tomography technology, computer vision technology and scene analysis technology have achieved great results. According to the application characteristics of image signal processing, the processing technology can be divided into image enhancement, restoration, segmentation, recognition, coding and reconstruction. These processing technologies have their own characteristics and are developing rapidly.
Vibration signal processing
The analysis and processing technology of mechanical vibration signal has been applied to the research and production of automobiles, aircraft, ships, mechanical equipment, housing construction, dam design, etc. The basic principle of vibration signal processing is to add an exciting force on the test body as the input signal. Monitor the output signal at the measuring point. The ratio of output signal to input signal is called the transfer function (or transfer function) of the system composed of the test body. According to the transfer function obtained, the so-called modal parameter identification is carried out to calculate the main parameters of the system, such as modal stiffness, modal damping, etc. In this way, the mathematical model of the system is established. Then the dynamic optimization design of the structure can be made. These works can be carried out by digital processors. This analysis and processing method is generally called modal analysis. In essence, it is a special method of signal processing in vibration engineering.
Geophysical processing
In order to explore the oil, natural gas and other mineral deposits stored deep underground, seismic exploration methods are usually used to detect the stratigraphic structure and lithology. The basic principle of this method is to apply artificial shock at a selected location. For example, if a vibration wave is generated by explosion and propagates underground, a reflected wave will be generated when encountering the stratum interface. A row of sensors will be placed at a certain distance from the vibration source to receive the reflected wave reaching the ground. The depth and structure of the formation can be judged from the delay time and intensity of the reflected wave. The seismic record received by the sensor is relatively complex and needs processing before geological interpretation. There are many processing methods, such as deconvolution, homomorphic filtering, etc. This is a problem that is still under study.
Biomedical processing
In biomedicine, signal processing is mainly used to assist the research of basic biomedical theory and to diagnose, check and monitor. For example, it is used for basic theoretical research in cytology, brain neurology, cardiology, genetics, etc. The human brain nervous system consists of about 10 billion nerve cells, which is a very complex and huge information processing system. In this processing system, information transmission and processing are carried out in parallel, and have special functions. Even if one part of the system is blocked, other parts can still work, which is impossible for computers. Therefore, the research on the information processing model of human brain has become an important topic of basic theoretical research. In addition, the research of neural cell model and chromosome function can be carried out with the help of the principle and technology of signal processing.
Signal processing is used for more successful examples of diagnostic tests, such as automatic analysis system of EEG or ECG, tomography technology, etc. Tomography is an important invention in the field of diagnostics. The basic principle of X-ray tomography is that X-ray passes through the observed object and forms a two-dimensional projection of the object. After receiving, the receiver can calculate two-dimensional projections in a series of different directions through restoration or reconstruction, and obtain the fault information of the entity through operation processing, so as to obtain the fault image on the large screen. The application of signal processing in biomedicine is developing rapidly.
Digital signal processing has many uses in other aspects, such as radar signal processing, geoscience signal processing, etc. Although they have their own special requirements, they use basically the same basic technology. In these aspects, digital signal processing technology plays a major role.
Digital signal processing system is a system of digital signal processing research and application. In any practical application, it is necessary to collect the original signal first. If the original signal is a continuous time signal, it must be sampled into a discrete signal, and then converted into a binary digital signal through analog digital conversion. If the collected original signal is already a discrete signal, it only needs to be converted into binary digital signal through analog digital conversion. The function of the digital signal processing system is to process the binary digital signal obtained from the sampling transformation of the original signal according to certain requirements, such as the requirements of transformation or filtering, to obtain the required output digital signal. The digital signal is converted into discrete signal through digital analog conversion, and then the discrete signal is restored to continuous time signal through holding circuit for output. Digital signal processing system can be composed of digital computer and software, general digital signal processor (DSP chip) or special signal processor.
The content of digital signal processing research is very extensive. Although different fields of application have different emphasis, research on fast and efficient algorithms, research on real-time hardware implementation of high processing rate, and research on new applications are the main topics to promote the development of digital signal processing technology.
With the development of digital signal processing technology and the improvement of the complexity and speed of DSP chips, many very complex computing or processing processes can be real-time processed in one chip, Even high-definition digital TV equipment has been widely used, and it has become a basic technology in modern information technology, artificial intelligence applications and other fields.
Both digital signal processing and analog signal processing are subsets of signal processing, and both belong to signal processing. The so-called "signal processing" refers to the process of processing the signal recorded on a certain media in order to extract useful information. It is a general term for processing processes such as signal extraction, transformation, analysis and synthesis. However, the objects processed by digital signal processing and analog signal processing are different, so the specific processes of processing are also different, but the purpose is to extract useful information.
In order to achieve high-performance digital signal processing, programmable DSP chips are generally used. The internal structure of the DSP chip is Harvard structure with separate program and data. It has a special hardware multiplier, widely uses pipeline operation, and provides special DSP instructions, which can be used to quickly implement various digital signal processing algorithms.
According to the requirements of digital signal processing, DSP chips generally have the following main characteristics:
(1) One multiplication and one addition can be completed in one instruction cycle.
(2) The program is separated from the data space and can access instructions and data at the same time.
(3) The on-chip has fast RAM, which can be accessed in two blocks simultaneously through independent data bus.
(4) Low overhead or no overhead loop and jump hardware support.
(5) Fast interrupt handling and hardware I/O support.
(6) There are multiple hardware address generators operating in a single cycle.
(7) Multiple operations can be performed in parallel.
(8) Pipeline operation is supported, so that operations such as fetching, decoding and execution can be overlapped.
Compared with general microprocessor, other general functions of DSP chip are relatively weak.
Figure 1 DSP chip
