News:
25-Apr-2014: Now the build uses -Wall compiler option so all the warnings on various architectures should show up easily.

Notes:
The dvbt TX implementation now supports 2k/8k mode, QPSK, QAM16, QAM64 constelaltions and rates 1/2, 2/3, 3/4, 5/6, 7/8.
Enable Doxygen: cmake ../ -DENABLE_DOXYGEN=ON
FIXED - Use AWGN generator to add noise to signals with huge SNR. Otherwise the signal is not decoded correctly.

Various TODOs beside what can be found in the sourcecode:

Fixes:
 - Peak detector does not find peaks for small search window (+/-4)
 - FIXED: Fix the problem with huge SNR input files not being played correctly.
 - Write docs for demod_reference_signals
 - Fix sampling freq correction calculation.
 - FIXED: Use -Wall on compiler and fix all issues that shows on build.

Rework:
 - Move pilot_generator class in its own file

Features: 
 - Support 8k mode.
 - DONE: Include puncturing matrix on convolutional encoder/decoder.
 - DONE: Support the other rates beside 1/2 on convolutional encoder/decoder.
 - Implement Soft decision Viterbi decoder together with interleaver.
 - Implement BICM.
 - Implement DFE on equalizer.
 - Autodetect transmission params.

Optimization:
 - Use VOLK on vector operations (on each module).
 - Calculate tables for pilot generator (spilot, cpilot, tps).
 - Use PI control loop for post-FFT carrier and sampling frequency correction.
 - Use tags in order to signal NSYNC start from conv. deinterleaver to energy descramble.
 - use std:conj, std::arg, std:: exp instead of C functions
 - Do not input block length directly in grc. Instead use transmission mode. Assert when data from xml differs from data from transmission mode.

Documentation:
 - Document enum types.

QA:
 - Add tests for each block.
 - FIXED: Check Viterbi-SSE2 implementation.
 - Use ETR290 to measure DVB-T parameters.


