TITRE Adaptive Packet Video Streaming Over IP Networks - LaBRI

More documents

Recommendations

Info

classification layer which classifies the object between them. As we said, the result of the classification is a set of AVOs sorted according to their importance in the scene. The classification layer affects a final relative priority score (RPS) to each Access Unit to apply a differentiation mechanism. This priority score reflects both the priority of particular AVO in the scene and the priority of a single frame type (I, P, B or hierarchical stream if any BL or EL). In the rest of this Section, we will interest to the prototype implementation which is based on a neural network algorithm. n Let T = {(X,c)} be a training set of N labeled vectors, where Χ ∈R is a feature vector X = (x 1 ,x 2 ,…,x n ) t and c ∈ T is its class label from an index set T (X is an AVO, and T is the set of available classes (e.g. Diffserv PHB), in our context) . The variables x i are referred to as attributes (e.g. QoS features of each AVO). A class is modeled by one or more prototype, which have { U , U , 2 K } U m 1 as features. A classifier is a mapping function called C defined as C: R n T, which assigns a class label in T to each vector in R n (a vector is an MPEG-4 AVO features). Typically, the classifier is represented by a set of model parameters Λ = { ∆ k }. The classifier specifies a partitioning of the feature space into regions R j = { X∈ R n n : C(X) = j} where U R j ≡ R and I R j ≡ Φ . It also induces a corresponding partitioning of the training set into subset T. There are different methods based on this definition, which allow an automatic classification of Vectors. We present in what below the classification method that used Radial Basis Function (RBF) classification. Figure 4-3 shows the RBF classifier. A vector to be classified is passed to a set of basis functions, each returning one scale value ϕ j j=1, 2,…,l. The concrete choice of the basis function is not critical; the common implementations prefer the radial basic function (the Gaussian “bell” functions). Hence the name: RBF network, with : j j ϕ (x) = e j 2 D − 2 2σ j ϕ 1 + F 1 (x) X ϕ 2 + F 2 (x) Choose Largest F j (x) j ϕ l λ lk + F k (x) Figure 4-3: RBF architecture for classification 68
In RBF, 2 D = X − is the distance between X and U . ϕj(x) measures similarity, i.e. the k U k mapping function. λjk in the Figure is a set of scalar weights that connect each of the receptive fields to the class outputs of the network. The general equation of an output of the neuron j is given by: n ∑ F ( X ) = λ ϕ ( X ) k j = 1 jk j R j The classifier maps the vector X (an Audio Visual Object) to the class with the largest output: n { x ∈ R / F ( x) ≥ F ( x ∀k} ≡ ) j k { (x)} F j A neural network takes in it input the feature vector X, produces computing class outputs , and then, classification decisions are made based on the largest output. The classification cannot be accomplished without knowing the characteristic of the underlying transport mechanisms. 4.1.3 Performance Evaluation This section presents our experiment testbed, and some results got by using our MPEG-4 platform. During experiments, two scenarios are investigated, the first on the standard MPEG-4 framework (without any mechanism of QoS) and the second with the new MPEG-4 framework (with Classification Layer and prioritization mechanisms). We loaded the network by background traffic to generate congestion. Measurements are done to evaluate packet losses and end-to-end one-way delay from the server to the client. 4.1.3.1 System and Network Models In our experiments, we have used an MPEG-4 video composed of three Video Objects. Each object is carried through an elementary stream. These objects are (see Figure 4-4 for more details): (1) Object 1: The logo at the top left of the MPEG-4 Video, (2) Object 2: The background of the MPEG-4 Video, (3) Object 3: The person. Scene logo Background Speaker MPEG-4 Scene Figure 4-4: MPEG-4 scene used for experiment 69
Page 1:
UNIVERSITÉ DE VERSAILLES SAINT-QUE
Page 4 and 5:
Résumé Nous constatons aujourd’
Page 6 and 7:
3 Related Work ....................
Page 8 and 9:
5.5.1.2 Simulation Results.........
Page 10 and 11:
List of Figures Figure 1-1: Structu
Page 12 and 13:
Figure 5-15: End-to-end packet tran
Page 14 and 15:
Acknowledgments The first person I
Page 16 and 17:
[11] Toufik Ahmed, Guillaume Burida
Page 18 and 19:
d’accès. Les interactions de QoS
Page 20 and 21:
Métrique MAX_DELAY PREF_MAX_DELAY
Page 22 and 23:
1.1.2 Proposition dun Protocole d'E
Page 24 and 25:
des paquets RTP. Reste toujours le
Page 26 and 27:
Nous pouvons distinguer trois possi
Page 28 and 29:
1.1.2.5.2 Multiplexage des Flux El
Page 30 and 31:
l’horloge est par défaut égale
Page 32 and 33:
La valeur R TCP représente le déb
Page 34 and 35: Pour ces raisons, nous nous intére
Page 36 and 37: Telles que illustrés par la Figure
Page 38 and 39: 3. DMIF2SIP vérifie si son propre
Page 40 and 41: utilisée optionnellement pour loca
Page 42 and 43: Chapter 2 2 Introduction 2.1 Motiva
Page 44 and 45: conditions. Based on end-to-end fee
Page 46 and 47: Chapter 3 3 Related Work Nowadays,
Page 48 and 49: • multicasting: it takes the stre
Page 50 and 51: expensive, fixed delivery and recep
Page 52 and 53: 3.1.6 Static vs. Dynamic Channels T
Page 54 and 55: environment because they require a
Page 56 and 57: H.261 is called video codec for aud
Page 58 and 59: VideoScene (VS) VideoObject (VO) Vi
Page 60 and 61: The emphasis of MPEG-7 will be the
Page 62 and 63: Temporal scalability involves parti
Page 64 and 65: multitude of streams. It can switch
Page 66 and 67: increased linearly in the absence o
Page 68 and 69: 3.2.2.1 End-to-End Retransmission R
Page 70 and 71: correctly received block. The missi
Page 72 and 73: epresent user traffic. The delay ca
Page 74 and 75: 3.2.5 IP Signaling Protocols for Pa
Page 76 and 77: two H.323 endpoints or between an e
Page 79 and 80: Chapter 4 4 Adaptive Packet Video T
Page 81 and 82: 4.1.1 Video Classification Model Pr
Page 83: attributes can refer to the QoS par
Page 87 and 88: Figure 4-6: Experiment testbed In t
Page 89 and 90: 190 180 170 End-to-end delay MPEG-4
Page 91 and 92: Sync Layer. The SL-packetized strea
Page 93 and 94: conveyed by MIME format parameters
Page 95 and 96: the same value. This timing informa
Page 97 and 98: 4.2.2.1 Fragmentation We propose to
Page 99 and 100: Timestamp: Indicates the sampling i
Page 101 and 102: When G is used as generator matrix,
Page 103 and 104: 4.2.3 Performance Evaluation Intens
Page 105 and 106: 0.3 0.25 our proposal classical app
Page 107 and 108: lost packet at the receiver. Conseq
Page 109 and 110: marking scheme. The server must be
Page 111 and 112: The video server adds audio-visual
Page 113 and 114: Marker (TR3CM) [131] to mark the ba
Page 115 and 116: VO1 VO2 AVO Scenario Audio BL EL1 E
Page 117 and 118: Throughput (Kbits) 3000 2500 2000 1
Page 119 and 120: Throughput (Kbits) 3000 2500 2000 1
Page 121 and 122: 4000 3500 3000 2500 2000 1500 1000
Page 123 and 124: 45 40 Original Quality Received Qua
Page 125 and 126: Chapter 5 5 A Dynamic Packet Video
Page 127 and 128: The TCM algorithm is based on a tok
Page 129 and 130: BEGIN Initialize the classification
Page 131 and 132: User Profile (Token Bucket) Traffic
Page 133 and 134: 5.3.3 QoS Management Algorithm Spec
Page 135 and 136:
Figure 5-7: Hierarchical aggregatio
Page 137 and 138:
• Filters: to put packets into cl
Page 139 and 140:
mechanism associated with the Drop
Page 141 and 142:
Figure 5-16: Packets loss with IP D
Page 143 and 144:
5.5.2.3 Experimental Results Figure
Page 145 and 146:
1 MPEG-4 Based Layer Stream -AF11 M
Page 147 and 148:
0.2 0.18 MPEG-4 Based Layer Stream
Page 149 and 150:
By using our PBNM tool, we allocate
Page 151 and 152:
Id Condition Indicator Action 1 Net
Page 153 and 154:
The proposed marking algorithm bett
Page 155 and 156:
Chapter 6 6 A SIP/MPEG-4 Multimedia
Page 157 and 158:
performed by many ways like SLA (Se
Page 159 and 160:
6.1.1.4 SIP Communication Model SIP
Page 161 and 162:
• It assign a value (network sess
Page 163 and 164:
6.2.1 SIP2DMIF Subsystem The SIP2DM
Page 165 and 166:
Step 1: The MPEG-4 Terminal passes
Page 167 and 168:
1. Set the result C to the empty se
Page 169 and 170:
TCP/UDP port, our implementation us
Page 171 and 172:
Control & Data Plan Video Rate Cont
Page 173 and 174:
characteristics and relevance of ea
Page 175 and 176:
References [1] Kazaa Media Desktop
Page 177 and 178:
[43] B. Schuster, “Fine granular
Page 179 and 180:
[79] D. Bansal and H. Balakrishnan,
Page 181 and 182:
[115] N. Laoutaris, I. Stavrakakis,
Page 183 and 184:
[154] W. Almesberger, J. H. Salim,
Page 185 and 186:
Appendix A A Appendix A: Overview o
Page 187 and 188:
Media aware Delivery unaware ISO/IE
Page 189 and 190:
A.3.3 Scene Description Streams Sce
Page 191:
This appendix presents structure ex
show all

TITRE Adaptive Packet Video Streaming Over IP Networks - LaBRI

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?