aboutsummaryrefslogtreecommitdiffstats
path: root/src/dabOutput/edi/Interleaver.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/dabOutput/edi/Interleaver.cpp')
-rw-r--r--src/dabOutput/edi/Interleaver.cpp122
1 files changed, 0 insertions, 122 deletions
diff --git a/src/dabOutput/edi/Interleaver.cpp b/src/dabOutput/edi/Interleaver.cpp
deleted file mode 100644
index f26a50e..0000000
--- a/src/dabOutput/edi/Interleaver.cpp
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- Copyright (C) 2017
- Matthias P. Braendli, matthias.braendli@mpb.li
-
- http://www.opendigitalradio.org
-
- EDI output,
- Interleaving of PFT fragments to increase robustness against
- burst packet loss.
-
- This is possible because EDI has to assume that fragments may reach
- the receiver out of order.
-
- */
-/*
- This file is part of ODR-DabMux.
-
- ODR-DabMux is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as
- published by the Free Software Foundation, either version 3 of the
- License, or (at your option) any later version.
-
- ODR-DabMux is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with ODR-DabMux. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#include "Interleaver.h"
-#include <cassert>
-
-namespace edi {
-
-void Interleaver::SetLatency(size_t latency_frames)
-{
- m_latency = latency_frames;
-}
-
-Interleaver::fragment_vec Interleaver::Interleave(fragment_vec &fragments)
-{
- m_fragment_count = fragments.size();
-
- // Create vectors containing Fcount*latency fragments in total
- // and store them into the deque
- if (m_buffer.empty()) {
- m_buffer.emplace_back();
- }
-
- auto& last_buffer = m_buffer.back();
-
- for (auto& fragment : fragments) {
- const bool last_buffer_is_complete =
- (last_buffer.size() >= m_fragment_count * m_latency);
-
- if (last_buffer_is_complete) {
- m_buffer.emplace_back();
- last_buffer = m_buffer.back();
- }
-
- last_buffer.push_back(std::move(fragment));
- }
-
- fragments.clear();
-
- while ( not m_buffer.empty() and
- (m_buffer.front().size() >= m_fragment_count * m_latency)) {
-
- auto& first_buffer = m_buffer.front();
-
- assert(first_buffer.size() == m_fragment_count * m_latency);
-
- /* Assume we have 5 fragments per AF frame, and latency of 3.
- * This will give the following strides:
- * 0 1 2
- * +-------+-------+---+
- * | 0 1 | 2 3 | 4 |
- * | | +---+ |
- * | 5 6 | 7 | 8 9 |
- * | +---+ | |
- * |10 |11 12 |13 14 |
- * +---+-------+-------+
- *
- * ix will be 0, 5, 10, 1, 6 in the first loop
- */
-
- for (size_t i = 0; i < m_fragment_count; i++) {
- const size_t ix = m_interleave_offset + m_fragment_count * m_stride;
- m_interleaved_fragments.push_back(first_buffer.at(ix));
-
- m_stride += 1;
- if (m_stride >= m_latency) {
- m_interleave_offset++;
- m_stride = 0;
- }
- }
-
- if (m_interleave_offset >= m_fragment_count) {
- m_interleave_offset = 0;
- m_stride = 0;
- m_buffer.pop_front();
- }
- }
-
- std::vector<PFTFragment> interleaved_frags;
-
- const size_t n = std::min(m_fragment_count, m_interleaved_fragments.size());
- std::move(m_interleaved_fragments.begin(),
- m_interleaved_fragments.begin() + n,
- std::back_inserter(interleaved_frags));
- m_interleaved_fragments.erase(
- m_interleaved_fragments.begin(),
- m_interleaved_fragments.begin() + n);
-
- return interleaved_frags;
-}
-
-}
-
-