Byungin Moon Yonsei University
Outline What is a DSP system? Why is important DSP? Advantages of DSP systems over analog systems Example DSP applications Characteristics of DSP systems Sample rates Clock rates Numeric representations Classes of DSP applications Low-cost embedded systems High-performance applications Personal computer-based multimedia 1
What is a DSP System Definition Electronic system making use of digital signal processing Digital signal processing Application of mathematical operations to digitally represent signals Contrasted with analog signal processing Signals are represented digitally as sequences of samples Obtained from physical signals through the use of transducers and analog-to-digital converters (ADCs) After digital processing, digital signals may be converted to physical signals via digital-to-analog converters (DACs) 2
Why is DSP Important? DSP is a key technology for many types of electronic products Modems and digital cellular phones Even primarily analog consumer electronics devices like audio amplifiers DSP-intensive tasks are the performance bottleneck in many computer applications today Computational demands of DSP-intensive tasks are increasing very rapidly 2000 market for DSP processors US$6.2 billion (2x 1998) 3
Advantages of over Analog Systems Accomplish tasks inexpensively that would be difficult or even impossible using analog electronics Speech synthesis, speech recognition, and high-speed modems involving error-correction coding Insensitivity to environment Whether in the snow or in the desert, a DSP system delivers the same response Insensitivity to component tolerances Two analog systems of exactly the same design will have slightly different responses due to slight variations in their components 4
Advantages of over Analog Systems Predictable and repeatable behavior DSP systems have exact, known responses that do not vary Reprogrammability A DSP system based on programmable processors can be reprogrammed Size The size of analog components varies with their values DSP can take advantages of the rapidly increasing density of IC process DSP increasingly becomes the solution of choice for signal processing 5
Example DSP Applications Speech coding and decoding Speech encryption and decryption Digital cellular telephones, personal communication systems, secure communications Speech recognition and synthesis Advanced user interfaces, robotics Speaker identification Security Hi-fi audio encoding and decoding Consumer audio and video Modem algorithms xdsl modems, Cable modems, Digital Cellular phones 6
Example DSP Applications Servo control Hard disk drives Audio equalization Vision Professional audio, advanced vehicular audio Multimedia computers, security, navigation Image compression and decompression Digital video, video on demand Beamforming Radar/sonar Echo cancellation Speaker phones, modems, telephone switches 7
Sample Rates Sample rate Rate at which samples are consumed, processed, or produced Sample rate determines the required speed of the implementation technology CD player produces samples at a rate of 44.1 khz on two channels Multirate DSP systems System using more than one sample rate The ratio between the highest and lowest rates can become quite large (sometimes exceeding 100,000) Converters from CD to DAT, filter banks 8
Sample Rates The range of sample rates in signal processing systems is huge The more complex, the lower sample rate At the highest rate, audio implementation is usual instead of digital implementation Hard real-time constraints Real-time DSP systems must meet performance goals in every instance (not on average) CD-to-DAT converter Must accept a new sample from the CD source every 22.6 μs (44.1 khz) Must produce a new output sample for the DAT every 20.8 μs (48 KHz) 9
Source: BuyerGuide Sample Rates vs. Algorithm Complexity 10
Clock Rates Definition The rate at which the system performs its most basic unit of work The ratio of system clock rate to sample rate One of the most important characteristics used to determine how the system will be implemented The relationship between the clock rate and the sample rate partially determines the amount of hardware needed to implement an algorithm Integral ration is preferred As the ratio increases, so does the amount of complexity of hardware 11
Numeric Representations Arithmetic operations are at the heart of DSP algorithms and systems Numeric representations and type of arithmetic used have an profound influence of the DSP system Fixed-point arithmetic Integer or fractional (-1.0 x < 1.0) Overflow Wrap around or saturate sign bit radix point bit weights -2 0 2-1 2-2 2-3 2-4 2-5 2-6 2-7 12
Numeric Representations Floating-point arithmetic value = mantissa 2 exponent Greater dynamic range Reduced possibility of overflow and the necessity of scaling More complicated to implement in hardware than fixedpoint arithmetic Slower and more expensive Mantissa Exponent sign bit sign bit radix point bit weights 2-1 2-2 2-3 2-4 2-5 2-6 2-7 -2 3 2 2 2 1 2 0 13
Classes of DSP Applications DSP, and DSP processors in particular, are used in an extremely diverse range of applications No one processor can meet the needs of all or even most applications Must weigh the relative importance of performance, cost, integration, ease of development, power consumption, etc. Low-cost embedded systems The largest applications Inexpensive, high-volume embedded systems Cost, integration and power consumption (for mobile) are critical Cellular phones, disk drives 14
Classes of DSP Applications High-performance Applications Applications processing large volumes of data with complex algorithms for specialized needs Production volumes are lower Algorithms are more demanding Product designs are larger and more complex Sonar, radar and seismic exploration Performance, ease of use, and support for multiprocessor configurations are critical features System assembling by using standard development boards Software development by using existing software libraries 15
Classes of DSP Applications Personal computer-based multimedia Voice mail, data and facsimile modems, music and speech synthesis, and image processing Low cost and high integration (like embedded) High performance demand (unlike embedded) Needed to perform multiple functions simultaneously Multitasking nature: efficient switch between functions Memory capacity Increasing trend to incorporate DSP-like functions into general-purpose processors Able to handle certain tasks without the need of a separate DSP processor in some cases The authors of the textbook and I believe the use of separate DSP processors will continue to be dominant 16