Make overuse estimator one dimensional.

webrtc/modules/remote_bitrate_estimator/overuse_estimator.cc

 /*
  *  Copyright (c) 2013 The WebRTC project authors. All Rights Reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS.  All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
 #include "webrtc/modules/remote_bitrate_estimator/overuse_estimator.h"
 
 #include <algorithm>
 #include <assert.h>
 #include <math.h>
 #include <stdlib.h>
 #include <string.h>
 
+#include "webrtc/base/checks.h"
 #include "webrtc/modules/remote_bitrate_estimator/include/bwe_defines.h"
 #include "webrtc/system_wrappers/interface/logging.h"
 
 namespace webrtc {
 
 enum { kMinFramePeriodHistoryLength = 60 };
 enum { kDeltaCounterMax = 1000 };
 
-OveruseEstimator::OveruseEstimator(const OverUseDetectorOptions& options)
-    : options_(options),
-      num_of_deltas_(0),
-      slope_(options_.initial_slope),
-      offset_(options_.initial_offset),
-      prev_offset_(options_.initial_offset),
-      E_(),
-      process_noise_(),
-      avg_noise_(options_.initial_avg_noise),
-      var_noise_(options_.initial_var_noise),
-      ts_delta_hist_() {
-  memcpy(E_, options_.initial_e, sizeof(E_));
-  memcpy(process_noise_, options_.initial_process_noise,
-         sizeof(process_noise_));
-}
+OveruseEstimator::OveruseEstimator()
+    : num_of_deltas_(0),
+      offset_(0),
+      prev_offset_(offset_),
+      e_(0.1),
+      process_noise_(0.01),

[Gaetano Carlucci, 2015/10/06 13:50:25]

We can run some tests to verity this setting of the process noise variance.

[stefan-webrtc, 2015/11/24 16:08:03]

We'll have to tune this, but for now I'll keep it as is. In a follow-up I think
we will have to make it somewhat depending on sample rate.

+      avg_noise_(0),
+      var_noise_(50),
+      ts_delta_hist_() {}
 
 OveruseEstimator::~OveruseEstimator() {
   ts_delta_hist_.clear();
 }
 
 void OveruseEstimator::Update(int64_t t_delta,
                               double ts_delta,

[terelius, 2015/10/12 12:04:41]

s/t_delta/recv_t_delta or something similar?
s/ts_delta/send_t_delta ?

Should they have different types?

[stefan-webrtc, 2015/11/24 16:08:03]

Done.

-                              int size_delta,
                               BandwidthUsage current_hypothesis) {
   const double min_frame_period = UpdateMinFramePeriod(ts_delta);
   const double t_ts_delta = t_delta - ts_delta;
-  double fs_delta = size_delta;
 
   ++num_of_deltas_;
   if (num_of_deltas_ > kDeltaCounterMax) {
     num_of_deltas_ = kDeltaCounterMax;
   }
 
   // Update the Kalman filter.
-  E_[0][0] += process_noise_[0];
-  E_[1][1] += process_noise_[1];
+  e_ += process_noise_;
 
   if ((current_hypothesis == kBwOverusing && offset_ < prev_offset_) ||
       (current_hypothesis == kBwUnderusing && offset_ > prev_offset_)) {
-    E_[1][1] += 10 * process_noise_[1];
+    e_ += 10 * process_noise_;
   }
 
-  const double h[2] = {fs_delta, 1.0};
-  const double Eh[2] = {E_[0][0]*h[0] + E_[0][1]*h[1],
-                        E_[1][0]*h[0] + E_[1][1]*h[1]};
-
-  const double residual = t_ts_delta - slope_*h[0] - offset_;
+  const double residual = t_ts_delta - offset_;
 
   const bool in_stable_state = (current_hypothesis == kBwNormal);
   const double max_residual = 3.0 * sqrt(var_noise_);
   // We try to filter out very late frames. For instance periodic key
   // frames doesn't fit the Gaussian model well.
   if (fabs(residual) < max_residual) {
     UpdateNoiseEstimate(residual, min_frame_period, in_stable_state);
   } else {
     UpdateNoiseEstimate(residual < 0 ? -max_residual : max_residual,
                         min_frame_period, in_stable_state);
   }
-
-  const double denom = var_noise_ + h[0]*Eh[0] + h[1]*Eh[1];
-
-  const double K[2] = {Eh[0] / denom,
-                       Eh[1] / denom};
-
-  const double IKh[2][2] = {{1.0 - K[0]*h[0], -K[0]*h[1]},
-                            {-K[1]*h[0], 1.0 - K[1]*h[1]}};
-  const double e00 = E_[0][0];
-  const double e01 = E_[0][1];
+  const double k = e_ / (var_noise_ + e_);
 
   // Update state.
-  E_[0][0] = e00 * IKh[0][0] + E_[1][0] * IKh[0][1];
-  E_[0][1] = e01 * IKh[0][0] + E_[1][1] * IKh[0][1];
-  E_[1][0] = e00 * IKh[1][0] + E_[1][0] * IKh[1][1];
-  E_[1][1] = e01 * IKh[1][0] + E_[1][1] * IKh[1][1];
+  e_ = e_ * (1.0 - k);
 
-  // The covariance matrix must be positive semi-definite.
-  bool positive_semi_definite = E_[0][0] + E_[1][1] >= 0 &&
-      E_[0][0] * E_[1][1] - E_[0][1] * E_[1][0] >= 0 && E_[0][0] >= 0;
-  assert(positive_semi_definite);
-  if (!positive_semi_definite) {
-    LOG(LS_ERROR) << "The over-use estimator's covariance matrix is no longer "
-                     "semi-definite.";
-  }
+  // The covariance matrix must be positive.
+  RTC_DCHECK(e_ >= 0.0);
+  if (e_ < 0)
+    LOG(LS_ERROR) << "The over-use estimator's covariance is negative!";
 
-  slope_ = slope_ + K[0] * residual;
-  prev_offset_ = offset_;
-  offset_ = offset_ + K[1] * residual;
+  offset_ = offset_ + k * residual;
 }
 
 double OveruseEstimator::UpdateMinFramePeriod(double ts_delta) {
   double min_frame_period = ts_delta;
   if (ts_delta_hist_.size() >= kMinFramePeriodHistoryLength) {
     ts_delta_hist_.pop_front();
   }
   std::list<double>::iterator it = ts_delta_hist_.begin();
   for (; it != ts_delta_hist_.end(); it++) {
     min_frame_period = std::min(*it, min_frame_period);
@@ skipping 20 matching lines @@
   const double beta = pow(1 - alpha, ts_delta * 30.0 / 1000.0);
   avg_noise_ = beta * avg_noise_
               + (1 - beta) * residual;
   var_noise_ = beta * var_noise_
               + (1 - beta) * (avg_noise_ - residual) * (avg_noise_ - residual);
   if (var_noise_ < 1) {
     var_noise_ = 1;
   }
 }
 }  // namespace webrtc