An End-to-End Adaptation Protocol for Layered Video Multicast Using Optimal Rate Allocation Jiangchuan Liu, Member, IEEE, Bo Li, Senior Member, IEEE, and Ya-Qin Zhang, Fellow, IEEE IEEE TRANSACTIONS ON MULTIMEDIA, VOL. 6, NO. 1, FEBRUARY 2004

Content  Paper Overivew  Objective  Key Issues  Motivation of Paper  Related Works  TCP-Friendliness  Layered Video Multicast  Rate Adaptation & Allocation  Sender-side functionality  Receiver-side functionality  Simulation Result

Paper Overview

Objectives  Problem :  Multicasting 시, coarse-grained layer subscription levels 과 heterogeneous 한 환경의 수신자가 요구하는 rate 간의 불일치 (mismatches)  To solve this problem  Sender-side : support to dynamic and fine-grained layer rate allocation  Rate allocation/layer 을 하기 위한 지표 ( 기준 ) 를 확립해야 한다.  Application-aware Fairness index  Works 1 : 저자는 rate allocation 에 대해, multicast session 을 받는 모든 수신자의 expected fairness index 를 maximization 하자는 목표를 가지고 optimization problem 을 formulate 한다.  Works 2 : 효과적인 scalable solution HALM(Hybirid Adaptation Layered Multicast) 을 제시한다  HALM Profit  It can be seamlessly integrated into an end-to-end adaption protocol.  This protocol takes advantage of the emerging fine-grained layered coding(2004 년도 ) and is fully compatiable with the best-effort internet infra-.

Key Issues  Two key issues.  Dynamic allocation layer rate allocation can be an effective.  Practical complement : receiver-driven adaptation.  Question…  What are the proper criteria for optimal allocation?  How to derive an efficient algorithm for the optimal allocation?  How to design an integrated adaptation protocol using the optimal allocation?

Motivation of this paper(3/3)  Real-time video transmission has to adapt to dynamic network conditions  For Adaptation :  In traditional Unicast : usually done by the sender, which collects the receiver’s status via a feedback alg. and adjusts its transmission rate  In Multicast : single rate 를 가지기 때문에 다양한 환경의 users(required BW 가 다 틀리기 때문 ) 을 만족시켜줄 수 없다.  이 결과로 multicast 에 fair distribution 을 위해 multi-rate multicast 방식이 제안됨.  Fair distribution 은 각 receiver 는 하나의 세션 내에서 required BW 가 서로 틀림에도 불구하고, 자신의 capacity 에 적합한 rate 로 비디 오를 받는다.  이와 같이 주어진 multicast session 의 member 와 관련되어 fairness 를 높이는 목적을 가지는 fair distribution 방식을 intra-session fairness 라 부른다.

Motivation of this paper(2/3)  A commonly used multi-rate multicast approach is cumulative layered transmission :  raw video  layered encoding : transmission (base, Enhancement layer).  As an example, layers can be mapped to different IP multicast groups  disadavantage:  실제적으로 layered encoder 가 지원하는 layer 는 개수의 제한이 있 고, receiver 들의 다양한 환경에서 오는 required BW 에 대한 control 은 몇 개의 layer 로써 control 하기는 힘들다.  또한 이런 점은 remarkable fairness degradation 을 가져온다.

Motivation of this paper(3/3)  To mitigate this problem,  one possible solution is the use of fine-grained sender adaptation as a complement, i,e., dynamically allocating the layer rates.  First, the source coder should have the ability to control the layer rate.  Second, the sender should know the global state of the receivers.

Related Works H.264/AVC Extension  FGS functionality has been removed from the SVC specification that is still under development (SVC amendment will be finalized by the next(Geneva) JVT meeting beginning of July 07)

Related Works TCP-Friendliness  TCP 는 real-time video delivery protocol 로 사용하긴 힘 들다.  Why? Because these applications usually require a smoothed transmission rate and stringent restrictions on end-to-end delay.  대부분의 internet traffic 은 TCP 인데 video streaming protocols 은 혼잡상황에 민감한 TCP flows 를 많은 영 향을 끼치지 않는 한도 내에서 video traffic 을 보장하기 위해 몇 가지 rate control algorithm 이 필요하다.

Related Works Design of TCP-Friendliness  Note that, short-term adaptation results in bandwidth oscillations, which is not desirable for video transmission.  Thus our objective is to provide an adaptive protocol that will not starve background TCP traffic and, meanwhile, try to achieve a long-term fair share as close as possible.

Related Works Layered Video Multicast  Receiver-driven Layered Multicast (RLM) is a pure end-to- end adaptation protocol  It sends each video layer over a separate multicast group.  A receiver periodically joins a higher layer’s group to explore the available bandwidth.  Congestion detected : join-experiment  Shared learning mechanism : suppress to join experiment by other receivers

Rate Adaptation & Allocation

Hybrid Adaptation Protocol For Layered Multicast(HALM) Sender Functionality(1/2) Layered Encoder Layer l Layer 3 Layer 1 Layer 2 … Layered Video Base layer Enhancement Layer … b1b1 b2b2 b3b3 blbl The layer rates are given by Let denote the cumulative layer rate up to layer, that is, c1c1 c2c2 clcl denote the rate vector of the cumulative layers, discrete set offers all possible video rates that a receiver in the session could receive the maximum rate delivered to a receiver with an expected bandwidth thus will be Expected BW : Receiver 1 Receiver 2 Receiver 3 Receiver 4 The sender will adaptively allocate the layer rates based on the distribution of the receivers’ expected bandwidths. SR Sender Report(SR) :

Hybrid Adaptation Protocol For Layered Multicast(HALM) Sender Funtionality(2/2)  We assume a rate vector is different from the one in the previous control period (in case they are the same, the sender can offset the current vector by a small value).  Hence, the change of the rate vector can serve as an implicit synchronization signal to trigger the receivers’ joining/leaving actions.

Hybrid Adaptation Protocol For Layered Multicast(HALM) Receiver Functionality(1/3)  To be friendly to TCP, a receiver directly uses a TCP throughput function to calculate its expected bandwidth.  Main operation of receiver’s

Hybrid Adaptation Protocol For Layered Multicast(HALM) Receiver Functionality(2/3)  Advantages  First, it is TCP-friendly  because the rate is equivalent to or less than the long-term throughput of a TCP connection running over the same path.  Second, it is scalable  because the receivers’ joining/leaving actions are synchronized  cf) RLM : shared learning  Finally, it is very robust  because the implicit signal will be detected even if some SR packets are lost.

Hybrid Adaptation Protocol For Layered Multicast(HALM) Receiver Functionality(2/3)  Configuration of Loss event parameter  In highly dynamic network environment  network load change during the interval  persistent congestion  To avoid persistent congestion, if the loss rate p exceeds a threshold, a receiver has the flexibility to leave the highest layer being subscribed. Receiver 1 Receiver 2 Receiver 3 Receiver 4 SR persistent congestion RR - Response Report(RR) = - RR serves as a request for RTT estimation

Sender-based Dynamic Rate Allocation(1/3)  Optimization Criteria for Heterogeneous Receivers  Total Throughput ???  Fairness Index ???  with a cumulative subscription policy  the subscription level of a receiver relies on its expected bandwidth and the set of cumulative layer rates.  Fairness Index for a receiver with expected bandwidth as follows: This definition can be used to access the satisfaction of a receiver when there is a performance loss incurred by a mismatch between the discrete set of the possible receiving rates and the expected bandwidth.

Sender-based Dynamic Rate Allocation(2/3)  Nonlinearity can be characterized by a utility function we define an Application-aware Fairness Index  For a multicast session, our objective is to maximize the expected fairness index, for all the receivers in the session by choosing an optimal layer rate vector. where L is the maximum number of layers that the sender can manage.

Sender-based Dynamic Rate Allocation(3/3)  The complexity of this optimization problem can be further reduced by considering some characteristics of a practical layered coder.  Assume there are M operational points the set of operational rates is given by QP value = { x,y.z ….} : a finite set of admissible quantizers R1 R2 R3 RMRM … R1 R2 R3 RMRM … …

Optimal Allocation Algorithms(1/3)  Assume, the expected fairness index can be calculated as follows: Layer l Layer l-1 Subscription level of receiver’s Sender Receiver

Optimal Allocation Algorithms(2/3)  Let : the maximum expected fairness index when c l is set to the mth operational point, R m  Recurrence relation R1 R2 RlRl

Optimal Allocation Algorithms(3/3)  according to the definition of and the recurrence relation, the following inequation holds for all nonnegative and nondecreasing >= 0

Parameter Measurements and Local Coordination  Estimation of Round-Trip Time(1/2)  Obtaining an accurate and stable measurement of the round-trip time is of primary importance for HALM  To find the “true RTT”, we must use a feedback loop  Feedback mechanism  Many receiver’s & high frequency(BW??) : cause implosion at the sender  Many receiver’s & low frequency(BW??) : inaccurate conclusions.  Using two mechanism Closed-loop RTT Open-loop RTT the sender does not give a response to each request but uses a batch process. The open-loop estimation method tracks the one-way trip time from the sender to the receiver and transforms it to an estimate of RTT.

Parameter Measurements and Local Coordination  Estimation of Round-Trip Time(2/2)  Timing diagram for closed-loop and open-loop RTT estimations Note that an RTT estimate can be expressed as is the one-way trip time from the sender to the receiver and is the time from the receiver to the sender. where t 0 and t’ are the current local time and the local time that the request was initiated, respectively.

Simulation Result(1/3)  Simulation Topology & Distribution of cumulative layer rate without joining and leaving. Simulation Time: 1000 sec (long-term : steady-state) HALM init cumulative layer rates = { 256, 512, 1024 kbps}

Simulation Result(2/2)  Bandwidth distribution between HALM and TCP at different switches.

Simulation Result(3/3) Distribution of cumulative layer rates with dynamic joining and leaving DISTRIBUTION OF THE RECEIVED BANDWIDTHS (Kbps). THE RATIO IS OBTAINED BY DIVIDING THE LAYERED STREAM BANDWIDTH BY THE TCP BANDWIDTH LMSA(Layered Multicast Static Allocation) – U (Uniform Distribution) : { 200, 1100, 2000 kbps} LMSA – E (Exponential) : { 256, 512, 1024 kbps}