Jitter Detection for Layered Multimedia Multicast Traffic Siu-Ping CHAN, and Chi-Wah KOK Department of Electrical and Electronic Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, HONG KONG Abstract— This paper investigates a modified version of active buffer management scheme “Jitter Detection” (JD) for gateway-based con- gestion control to stream layered multimedia multicast traffic over IP network. The JD scheme works well with traditional congestion control schemes, such as RED, which are used for non-multimedia traffic. At the same time, the JD scheme improves the quality of service in multimedia networking by detecting and discarding useless packets that accumulated large enough jitter. Such as to maintain a high bandwidth for packets within the jitter tolerance of the multimedia player. The paper also considered multicast with layered coding scheme for multimedia streaming quality improvement. Differentiated from the traditional JD scheme, we proposed to preserve the base layer traffic with the best effort when passing through the gateway. The priority of dropping decisions are given to those higher level layers (enhancement layers) multicast packets. Queuing analysis has been performed to evaluate the efficiency of the proposed scheme. Simulations by NS2 were also performed and the results showed that the proposed scheme can provide better quality than that using RED in gateway-based congestion control for the multimedia traffic. Index Terms— Multimedia Streaming, Layered Multicast, Congestion Control, Jitter Detection, Priority Dropping I. I NTRODUCTION W HEN a multimedia stream is transported over packet-switched networks, it assumes a constant network delay to maintain the timing relationship among packets. It is difficult, however, if not impossible, to maintain a constant network delay. As a result, multimedia traffic will usually experience impairment due to network delay variation, or jitter, that may lead to degradation in the “Quality- of-Service” (QoS). In some multimedia streaming systems, streaming buffer is allocated in the client trying to solve the problem. However, streaming buffer can only help to increase the tolerance of the network jitter for multimedia networking. The situation will become worse when there are a large amount of jitter corrupted multimedia packets (multimedia packets that have accumulated jitter larger than the client’s jitter tolerance) in the network. In that case, although jitter corrupted multimedia packets will be rendered useless when received by the client, they are continuing to consume network bandwidth, and thus increasing the congestion of the network for multimedia packets delivery. This new kind of congestion cannot be solved by traditional active buffer management scheme such as “Random Early Detection” (RED) [1] and DropTail. A novel gateway-based congestion control for streaming traffic known as “Jitter Detection” (JD) is proposed in [9]. It aims at improving the quality of service by detecting and discarding multimedia packets that have accumulated large jitter, so as to maintain a high bandwidth for packets that stay within the multimedia stream’s jitter tolerance. Multicast with a layered coding scheme has been shown to be an effective solution for improving the quality of multimedia streaming through bitrate adaptation. Layered multicast has been a hot topic in the networking community as well as the coding community [3], [4], [5], [6], [8]. The well-known layered multimedia multicast algorithm, RLM [3], sends each video layer over a separate multicast group. The client periodically joins a higher layer’s group to explore the available bandwidth. If packet loss is detected, the client leaves the This work described in this paper has been supported by the Research Grants Council of Hong Kong, China (Project no. CERG HKUST6236/01E). group. However, the RLM algorithm is not TCP-friendly [4], [6], [7]. This can be improved by using equation-based rate control on the client side [6], [7]. In fact, the JD scheme can be applied to solve the congestion problem of layered multimedia streaming. In this paper, we further investigated the application of JD scheme for layered multimedia streaming aiming to improve the multicasting QoS. An analytical analysis, using queuing theory, was also presented to show the efficiency of the proposed scheme. II. J ITTER DETECTION (JD) FOR LAYERED MULTIMEDIA MULTICAST TRAFFIC Figure 1a illustrates the operation of a gateway where two types of traffic are assumed: UDP and TCP. The UDP traffic carry constant bitrate multimedia traffic, and TCP traffic carry variable bitrate services, such as FTP. Input traffic to the gateway are classified into TCP or UDP streaming traffic. TCP traffic is subjected to the RED scheme while UDP streaming traffic is subjected to the JD scheme before enqueued into the output queue. To simplify our discussion, the output queue is assumed to be FIFO. Figure 1b is the pseudo code of the modified JD scheme for layered multimedia multicast traffic. The variable delay is the time delay encountered by a streaming packet in the gateway. delay can be estimated from the current buffer occupancy (CBO) of the output queue together with the estimated available output link capacity (LC). The end-to-end delay jitter counter v tracks the accumulated jitter of the multimedia packets and is defined in Figure 1b and 1c, where v is updated by quantizing the accumulated jitter stored in v plus the estimated jitter experience by the multimedia packets in the gateway into one of the 4 regions in Figure 1b. We use bound equals to fixed th/6 to define the region similar to that in [9] and each region has a width of 1.5 * bound as shown in Figure 1c. Finally, the dropping threshold (threshold) is determined by threshold = jitter - f ixed th 2 (1 + (v +  v 3  ) 3 ). (1) The multimedia packet will be sent to the output queue iff -f ixed th 2 ≤ threshold ≤ f ixed th 2 . (2) III. QUEUING ANALYSIS OF THE PROPOSED SCHEME As illustrated in Figure 2, there are N l UDP flows with arrival rate γi from the packet classifier that go through the JD mechanism. For simplicity, it is assumed that all the packets are layered multimedia traffic from one source, and that there is no other TCP traffic passing through the gateway. The arrival rate of the “base” layer traffic is γ1, while other “enhancement” layer traffic have the arrival rate γ2,γ3, ...., and γN l , respectively. According to the JD scheme discussed in Section II, the N th l flow has the highest level of priority will be dropped first when there is network congestion, while the “base” layer, 1 st flow, is the last category to be dropped. Therefore, we set N l different dropping fraction βi with β1,β2, ...., and βN l for the N l different flows respectively in this analysis. Let βi = P (drop) * (1/N l ) N l −i for i =1, 2, ··· ,N l - 1,N l . (3) 0-7803-8603-5/04/$20.00 ©2004 IEEE.