• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  *  Copyright (c) 2013 The WebRTC project authors. All Rights Reserved.
3  *
4  *  Use of this source code is governed by a BSD-style license
5  *  that can be found in the LICENSE file in the root of the source
6  *  tree. An additional intellectual property rights grant can be found
7  *  in the file PATENTS.  All contributing project authors may
8  *  be found in the AUTHORS file in the root of the source tree.
9  */
10 
11 #include "webrtc/modules/remote_bitrate_estimator/overuse_estimator.h"
12 
13 #include <assert.h>
14 #include <math.h>
15 #include <stdlib.h>
16 #include <string.h>
17 
18 #include <algorithm>
19 
20 #include "webrtc/base/logging.h"
21 #include "webrtc/modules/remote_bitrate_estimator/include/bwe_defines.h"
22 
23 namespace webrtc {
24 
25 enum { kMinFramePeriodHistoryLength = 60 };
26 enum { kDeltaCounterMax = 1000 };
27 
OveruseEstimator(const OverUseDetectorOptions & options)28 OveruseEstimator::OveruseEstimator(const OverUseDetectorOptions& options)
29     : options_(options),
30       num_of_deltas_(0),
31       slope_(options_.initial_slope),
32       offset_(options_.initial_offset),
33       prev_offset_(options_.initial_offset),
34       E_(),
35       process_noise_(),
36       avg_noise_(options_.initial_avg_noise),
37       var_noise_(options_.initial_var_noise),
38       ts_delta_hist_() {
39   memcpy(E_, options_.initial_e, sizeof(E_));
40   memcpy(process_noise_, options_.initial_process_noise,
41          sizeof(process_noise_));
42 }
43 
~OveruseEstimator()44 OveruseEstimator::~OveruseEstimator() {
45   ts_delta_hist_.clear();
46 }
47 
Update(int64_t t_delta,double ts_delta,int size_delta,BandwidthUsage current_hypothesis)48 void OveruseEstimator::Update(int64_t t_delta,
49                               double ts_delta,
50                               int size_delta,
51                               BandwidthUsage current_hypothesis) {
52   const double min_frame_period = UpdateMinFramePeriod(ts_delta);
53   const double t_ts_delta = t_delta - ts_delta;
54   double fs_delta = size_delta;
55 
56   ++num_of_deltas_;
57   if (num_of_deltas_ > kDeltaCounterMax) {
58     num_of_deltas_ = kDeltaCounterMax;
59   }
60 
61   // Update the Kalman filter.
62   E_[0][0] += process_noise_[0];
63   E_[1][1] += process_noise_[1];
64 
65   if ((current_hypothesis == kBwOverusing && offset_ < prev_offset_) ||
66       (current_hypothesis == kBwUnderusing && offset_ > prev_offset_)) {
67     E_[1][1] += 10 * process_noise_[1];
68   }
69 
70   const double h[2] = {fs_delta, 1.0};
71   const double Eh[2] = {E_[0][0]*h[0] + E_[0][1]*h[1],
72                         E_[1][0]*h[0] + E_[1][1]*h[1]};
73 
74   const double residual = t_ts_delta - slope_*h[0] - offset_;
75 
76   const bool in_stable_state = (current_hypothesis == kBwNormal);
77   const double max_residual = 3.0 * sqrt(var_noise_);
78   // We try to filter out very late frames. For instance periodic key
79   // frames doesn't fit the Gaussian model well.
80   if (fabs(residual) < max_residual) {
81     UpdateNoiseEstimate(residual, min_frame_period, in_stable_state);
82   } else {
83     UpdateNoiseEstimate(residual < 0 ? -max_residual : max_residual,
84                         min_frame_period, in_stable_state);
85   }
86 
87   const double denom = var_noise_ + h[0]*Eh[0] + h[1]*Eh[1];
88 
89   const double K[2] = {Eh[0] / denom,
90                        Eh[1] / denom};
91 
92   const double IKh[2][2] = {{1.0 - K[0]*h[0], -K[0]*h[1]},
93                             {-K[1]*h[0], 1.0 - K[1]*h[1]}};
94   const double e00 = E_[0][0];
95   const double e01 = E_[0][1];
96 
97   // Update state.
98   E_[0][0] = e00 * IKh[0][0] + E_[1][0] * IKh[0][1];
99   E_[0][1] = e01 * IKh[0][0] + E_[1][1] * IKh[0][1];
100   E_[1][0] = e00 * IKh[1][0] + E_[1][0] * IKh[1][1];
101   E_[1][1] = e01 * IKh[1][0] + E_[1][1] * IKh[1][1];
102 
103   // The covariance matrix must be positive semi-definite.
104   bool positive_semi_definite = E_[0][0] + E_[1][1] >= 0 &&
105       E_[0][0] * E_[1][1] - E_[0][1] * E_[1][0] >= 0 && E_[0][0] >= 0;
106   assert(positive_semi_definite);
107   if (!positive_semi_definite) {
108     LOG(LS_ERROR) << "The over-use estimator's covariance matrix is no longer "
109                      "semi-definite.";
110   }
111 
112   slope_ = slope_ + K[0] * residual;
113   prev_offset_ = offset_;
114   offset_ = offset_ + K[1] * residual;
115 }
116 
UpdateMinFramePeriod(double ts_delta)117 double OveruseEstimator::UpdateMinFramePeriod(double ts_delta) {
118   double min_frame_period = ts_delta;
119   if (ts_delta_hist_.size() >= kMinFramePeriodHistoryLength) {
120     ts_delta_hist_.pop_front();
121   }
122   std::list<double>::iterator it = ts_delta_hist_.begin();
123   for (; it != ts_delta_hist_.end(); it++) {
124     min_frame_period = std::min(*it, min_frame_period);
125   }
126   ts_delta_hist_.push_back(ts_delta);
127   return min_frame_period;
128 }
129 
UpdateNoiseEstimate(double residual,double ts_delta,bool stable_state)130 void OveruseEstimator::UpdateNoiseEstimate(double residual,
131                                            double ts_delta,
132                                            bool stable_state) {
133   if (!stable_state) {
134     return;
135   }
136   // Faster filter during startup to faster adapt to the jitter level
137   // of the network. |alpha| is tuned for 30 frames per second, but is scaled
138   // according to |ts_delta|.
139   double alpha = 0.01;
140   if (num_of_deltas_ > 10*30) {
141     alpha = 0.002;
142   }
143   // Only update the noise estimate if we're not over-using. |beta| is a
144   // function of alpha and the time delta since the previous update.
145   const double beta = pow(1 - alpha, ts_delta * 30.0 / 1000.0);
146   avg_noise_ = beta * avg_noise_
147               + (1 - beta) * residual;
148   var_noise_ = beta * var_noise_
149               + (1 - beta) * (avg_noise_ - residual) * (avg_noise_ - residual);
150   if (var_noise_ < 1) {
151     var_noise_ = 1;
152   }
153 }
154 }  // namespace webrtc
155