Project #1: File I/O & Digital Signal Processing

Background:

We now apply everything we know about file I/O to process a digital signal. For our purposes a digital signal is a sequence of values representing the amplitude of a wave with respect to time. These values are time sequential at equal time intervals apart. The amount of time between each sample determines the sampling rate. If a series of amplitude values are 1000'th of second apart then the sampling rate is said to be 1000 samples per second or 1000Hz.

A few fundamental characteristics of a periodic signal (wave) are:

• Frequency: how often a wave repeats or cycles. Frequency is typically expressed in units called Hertz (Hz.) which is cycles per second.
  
  
• Period: the reciprocal of frequency. If a wave has a frequency of 10 Hz (cycles per second) then its period is 1/10th of a second or 100 milliseconds etc.
  
  
• Amplitude: the highest value of the wave in any given cycle. This term is overloaded in that we also speak of the value of the wave at any given time as the amplitude of the wave at that given time. Typically when someone asks "What is the amplitude of the wave"? - they refer to the highest value at the crest (peak). Conversely, the lowest value at the trough (nadir) of the wave could also be used as the amplitude of the wave.

In our above discussion an important assumption is being made. The wave is periodic.

The Assignment: Write two C programs

• The first program will be named txt2bin.c txt2bin.c_STARTER

  The txt2bin.c program will read a text file of numbers such as sine-1.txt which represents the digital signal. These values are a simulation of a 12 bit A/D conversion that is centered around zero. Thus the values will range from -2047 to + 2048. Each value will be written out in 16 bits (short int). Your txt2bin.c progam will read this text file one number at a time until end of file. Each time a number is read from the text file, that number will be written out (using fwrite) to a binary file. The value written out will be a short int (16 bits on most systems). Your txt2bin.c program will take two command line args: the input filename and the output filename. The input file should be the text file named "sine-1.txt" or such. The second command line argument will be the name of a binary file that is to be created such as "sine-1.bin". Since your program will write out a binary value every time it reads a number from the text file, you will not use any arrays anywhere in this program (or the second program either). Every value must be read in from the text file using fscanf(infile, "%hi", &shortVal ), and then a value written to the binary output file using fwrite( &shortVal, sizeof(shortVal), 1, outFile ) as illustrated in the File I/O demo code in Chapter 1.3.

  After you write out the binary file, close both files and then reopen the binary file for reading. Read the binary values back in one at a time and echo them out to stdout via printf. Your txt2bin program must not write any other text to stdout during execution. After you have finished echoing the values read back from the binary file - close the binary file.

  Below is what a compilation then execution of txt2bin should look like: (the .exe is not needed on the executable - I just named it that way for emphasis/clarity). Your txt2bin program must not write anything to stdout except the numbers you read back in from the binary file you just wrote to disk.

  $ gcc -W -Wall -Wextra -O2  txt2bin.c  -o txt2bin.exe
  $ ./txt2bin.exe sine-1.txt sine-1.bin
  814
  1533
  1652
  1523
  1843
  1990
  2047
  2016
  1924
  1987
  1523
  1211
  1055
  777
  566
  241
  0
  -110
  -508
  -946
  -1129
  -1427
  -1918
  -1771
  -2015
  -2004
  -1883
  -1752
  -1806
  -1600
  -1170
  -826
  -724
  -334
  -212
  0
  474
  862
  940
  1365
  1353
  1917
  1877
  2016
  2047
  2016
  1847
  1687
  1459
  1277
  1027
  818
  152
  0
  -106
  -443
  -693
  -1236
  -1206
  -1595
  -1781
  -1923
  -1987
  -1980
  -2015
  -1983
  -1680
  -1530
  -1459
  -1233
  -508
  -341
  -102
  123
  551
  755
  1163
  1295
  1688
  1873
  1680
  1867
  2047
  1814
  1803
  1646
  1402
  1520
  1030
  790
  571
  126
  -174
  -642
  -829
  -1126
  -1692
  -1699
  -1923
  -2015
  -2046
  -2015
  -1744
  -1981
  -1485
  -1450
  -1134
  -792
  -401
  0
  180
  582
  756
  946
  1452
  1339
  1768
  1924
  1838
  2047
  2016
  1951
  1596
  1415
  1078
  1260
  740
  266
  0
  -250
  -683
  -1118
  -1491
  -1787
  -1970
  -1823
  -2015
  -1855
  -2041
  -1707
  -1592
  -1711
  -1374
  -1266
  -613
  -148
  0
  358
  833
  1218
  1475
  1527
  1781
  1895
  1854
  2047
  2016
  1680
  1553
  1731
  1234
  1174
  541
  367
  -95
  -195
  -591
  -1149
  -1528
  -1770
  -1539
  -1923
  -2015
  -2046
  -1915
  -1882
  -1686
  -1444
  -1545
  -1177
  -929
  -257
  0
  273
  934
  1054
  $
  

  Use Linux tools to verify the correctness of your txt2bin program

  If your txt2bin program was written correctly, you will see a file named sine-1.bin with a size of 372 bytes in your directory. You can verify using the ls command a follows:

  $ ls -l *.bin
  -rw-r--r--  1 hoffmant users 372 Jan 12 21:57 sine-1.bin
  $

  If your txt2bin program is correct you should also notice that the output from it is identical to the contents of the original sine-1.txt file. Linux can help you verify this by using the following commands:

  $ ./txt2bin.exe sine-1.txt sine-1.bin > output.txt
  $ diff output.txt sine-1.txt
  $

  The first command executed your program again but this time the output that normally would have gone to the screen (stdout) was redirected by the carrot operator > into a file named output.txt.

  The second command (diff) looks for any differences between then two files and reports each line that differs from the corresponding line in the other file. Since there was no ouput from the diff command, we know that the two files are identical and that your binary file must be correct since it is of the correct size and produced the correct values when echoed back to the screen.

• The second program will be named project-1.c. Here is a starter file: project-1.c_STARTER

  • Your project1.c will take one command line arg - the name of the binary file. It will read the binary file calculate some basic statistics pertaining to the digital signal represented by the sequence of values in the file. For each FULL wave detected you are to output one line of text with the values seperated by TAB. On each line you are to print the following:
    
    
    • sample # on which this wave ended
    • the value of the above ending sample
    • the number of samples in this wave
    • the peak value in this wave (highest)
    • the nadir value in this wave (lowest)

    Here is a sample execution with correct output for the first sine wave.

    

    Here is a graph of the first sine wave. This graph shows where those output values came from. Note that the first 35 values are not part of a full wave because a full wave must start either on zero or the first point above zero after a zero crossing.

    

    Besides visual inspection you can use Linux tools again to verify that your project-1 program is correct by redirecting the screen output to a file and then diff to verify it matches the expected output.

    $ ./project-1.exe sine-1.bin > output.txt
    $ diff --ignore-all-space --ignore-blank-lines output.txt correct-output-sine-1.txt

    Notice we use a more forgiving form of the diff command which allows you to use spaces or tabs and only looks at the non whitespace.

    Here are your input files and the expected output for each file:

    • sine-1.txt correct-output-sine-1.txt
    • sine-2.txt correct-output-sine-2.txt

    Here is a graph of the data in "sine-2.txt" with indications of wave detection and characteristic values.