Page 128 - DSP Integrated Circuits
P. 128

Problems                                                             113

                  transversal FIR filter of the type shown in Figure 3.22. The length of the
                  filter isN= 12 and the sample rate is 8 kHz. Determine the arithmetic work
                  load for the adaptive filters.
             3.21 Draw a block diagram for the RLS lattice filter for N = 4.

             3.22 Show that






             3.23 Show that the DFT for N = 8 can be computed by essentially using two FFTs
                  with AT =4.
             3.24 Identify the complex constants in the FFT in Figure 3.23.

             3.25 Derive the signal-flow graph for the FFT, with N = 4, using the decimation-
                  in-frequency algorithm. What is the relationship between the decimation-in-
                  time and decimation-in-frequency algorithms?

             3.26 Show that an IDFT can be computed by using a DFT algorithm, by first
                  interchanging the real and imaginary parts of the sequence, performing the
                  DFT, and finally interchanging the real and imaginary parts of the
                  intermediate result.

             3.27 Show that the DFT of two real (or imaginary) sequences can be computed
                  simultaneously by only one FFT.
             3.28 Determine the available time for computing the FFT in the FFT coprocessor
                  if the I/O operations are overlapped with the FFT computation.
             3.29 In a transform coding system for images, the DCT-frequency components are
                  computed and their values are represented with different accuracy—number
                  of bits—to reduce the amount of data. However, the zero-frequency content of
                  most images is very large. Hence, if this component contributes an error to
                  the nonzero frequency components, then the image will be inefficiently coded
                  or even distorted. Determine which of the different DCTs can be used in such
                  an application.
             3.30 Determine the required channel capacity for ordinary TV signals. Determine
                  also the bit rates for the luminance (Y) and chrominance signals (U and V).


                                                       PAL        NTSC
                         Lines/frame                   625         525
                         Active lines                  576         480
                         Pixels/active line
                         Luminance, Y                  720         720
                         Chrominance: U and V          360         360
                         Frame rate (2:1 interlace)    50          59.94
                         Quantization                8-bit PCM   8-bit PCM
                                                     220 levels  220 levels
                         Sample frequency Y          13.5 MHz    13.5 MHz
                         UandV                       6.75 MHz    6.75 MHz
   123   124   125   126   127   128   129   130   131   132   133