viterbi 0.0.2

A Convolutional Encoder and Viterbi Decoder in Python/C++.

Convolutional Encoder and Viterbi Decoder

Author: Min Xu <[e-mail]>

This program implements both a Convolutional Encoder and a Viterbi Decoder in C++.

Running ViterbiCodec

Compile:

make

Run:

./viterbi_main <constraint> <polynomial>... <bits>

Example use:

Decoding:

./viterbi_main 3 7 5 0011100001100111111000101100111011
010111001010001

Encoding:

$ ./viterbi_main 3 7 5 --encode 010111001010001
0011100001100111111000101100111011

Please note there are (constraint - 1) "0" bits padded in front of the original message.

More Features

• It supports reading input from stdin or from commandline arguments. When reading from stdin, it ignores blank lines or comment lines (which start with #).

• It supports any number of generator polynomials.

• It can perform convolutional encoding by providing --encode commandline flag.

• It supports a different notation of generator polynomials by providing --reverse_polynomials commandline flag.

Here are more options to run the program.

Show help message:

./viterbi_main --help

Read input from stdin:

./viterbi_main [--reverse_polynomials] [--encode]

Input format is:

<constraint> <polynomial>... <bits>

Read input from commandline arguments:

./viterbi_main [--reverse_polynomials] [--encode] <constraint> <polynomial>... <bits>

Flags:

--reverse_polynomials
    Reverse polynomials. E.g. 6 (=0b110) becomes 3 (=0b011).

--encode
    Do encoding instead of decoding.

Example usage:

./viterbi_main 3 7 5 0011100001100111111000101100111011
./viterbi_main 3 6 5 111011011100101011
./viterbi_main 7 91 121 1010110100001101001000011101
./viterbi_main 7 91 117 121 111100101110001011110101111111001011100111
./viterbi_main --reverse_polynomials 3 3 5 111011011100101011
./viterbi_main --encode 3 7 5 010111001010001
./viterbi_main --encode 3 6 5 1001101
./viterbi_main --encode --reverse_polynomials 3 3 5 1001101

Error Handling

All inputs are validated, and proper error messages will be output.

• The constraint should be greater than 0.
• A generator polynomial should be greater than 0 and less than 1 << constraint.
• The received bit sequence should be consisted with only 0 and 1.
• The received bit sequence should be of length N * <num-of-polynomials> where N is an integer. Otherwise some bits must be missing during transmission. We will fill in appropriate number of trailing zeros.

Running Unit Test

Run unit test:

make test

Or

make viterbi_test && ./viterbi_test

If the test passes, it will output PASSED and the end of the output; otherwise, the test program aborts with assertion error.

There are a few sample cases in the unit test, some are with manually injected bit errors.

It also generates random messages of length 8, 16 and 32 for several different convolutional codes and test if they can be properly encoded and decoded.

Dependencies

The code is self-contained, meaning it depends on nothing but the C++ standard library. The purpose is to make it easy to be integrated in any project.