1 Interleaving Staircase Broadcasting and Receiving Scheme with Loss-Anticipation Delivery Hung-Chang Yang 1 , Hsiang-Fu Yu 1 , Li-Ming Tseng 1 , Yi-Ming Chen 2 {cyht,yu}@dslab.csie.ncu.edu.tw, tsenglm@csie.ncu.edu.tw, cym@im.mgt.ncu.edu.tw 1 Dep. of Computer Science & Information Engineering, National Central University, Taiwan 2 Dep. of Information Management, National Central University, Taiwan Abstract With the growth of broadband networks, Video-on-Demand (VoD) has become realistic. Many significant broadcasting schemes have been proposed to reduce the bandwidth requirement for stored popular videos. One representative approach is the Staircase Broadcasting (SB) scheme. In comparison with other schemes, the SB scheme requires smaller disk buffer and disk transfer rate. However, the scheme cannot be used for provide reliable delivery over lossy channels perfectly. In this paper, we proposed an Interleaving Staircase Broadcasting (ISB) scheme, which guarantees continuous playback, and mitigates the effect of packet losses. Some bounds on the bandwidth consumption, the buffer requirements, and the required disk transfer rate are also developed. Further, in comparison with the Interleaving Harmonic Broadcasting (IHB) scheme, Interleaving Cautions Harmonic Broadcasting (ICHB) scheme, Reliable Periodic Broadcasting (RPB) scheme and Second Chance Broadcasting (SCB) scheme, the ISB outperforms on maximum buffer requirements, and maximum required disk transfer rate. The IHB scheme consumes the least bandwidth. Keywords: fault tolerant, hot video broadcasting, loss recovery, video on demand 1. Introduction With the advancement of broadband networking technology and the growth of processor speed and disk capacity, Video-on-Demand has become realistic [10,12]. Many studies have investigated VoD. One of the important problems of streaming video on demand is the issue of delivering in a scalable and reliable manner. The scalable delivery is to allow streaming video on demand to large numbers of concurrent clients. The reliable delivery is to allow the clients tolerate packet loss and can continuous playback. To alleviate the scalable delivery, one way is to distribute the top ten or twenty so-called “hot” videos more efficiently [5]. An efficient category is the segment based broadcasting [2]. It transfers each video according to a fixed schedule and consumes a constant bandwidth regardless of the number of requests for that video. That is, the number of clients watching a given video is independent of their bandwidth requirements. The schemes [1,3,4,7-9,13-22] share a similar arrangement and substantially reduce the bandwidth requirement for hot video. One of these channels transmits the first segment in real time. The other channels transmit the remaining segments according to a schedule predefined by the scheme. When clients want to watch a video, they wait first for the beginning of the first segment on the first channel. Thus, their maximum waiting time equals the length of the first segment. While the clients begin watching the video, their set-top boxes (STB) or computers begin downloading enough data from the other channels so they will be able to play the segments of the video in turn. The simplest broadcasting scheme is the staggered broadcasting [1] scheme. The server allocates K channels to transmit a video. Its maximum viewers’ waiting time is K L , where L is the video length. The pyramid broadcasting [18] scheme divides a video into increasing size of segments and transmits them on multiple channels of the same bandwidth. It requires less maximum waiting time than the staggered broadcasting under the same bandwidth. The fast broadcasting (FB) [9] scheme divides a video into a geometrical series of 1, 2, 4, …, 2 K-1 . Its maximum waiting time is 1 2 - K L . In comparison with the staggered broadcasting scheme and the pyramid broadcasting scheme, the FB scheme obtains shorter waiting time. An implementation of the FB scheme on IP networks was reported in [24]. Based on the pagoda broadcasting [14] scheme, the new pagoda broadcasting (NPB) [15] scheme divides a video into fixed-size segments and maps them into data channels of equal bandwidth at the proper decreasing frequencies. Accordingly, the NPB scheme obtains shorter waiting time than the FB scheme. The recursive frequency splitting (RFS) [16] scheme further improves the NPB scheme on waiting time by using a more complex segment-to-channel mapping. The harmonic broadcasting (HB) [8] scheme first divides a video into several segments equally, and further divides the segments into sub-segments according to the harmonic series. Yang, Juhn, and Tseng [23] have proved that the HB scheme requires the minimum bandwidth under the same waiting time. However, the study [13] presents that the HB scheme cannot always deliver all video data on time, and the authors have proposed the cautious harmonic broadcasting (CHB) and quasi-harmonic broadcasting (QHB) schemes to solve the problem. However, their clients need larger buffers to store video data. For example, the FB, NPB and RFS schemes require their clients to buffer 50%, 40% and 40% of a playing video. Suppose a video of 100 minutes is played at the rate of 1.5Mbps. Its size is about 1125 Mbytes. The required buffers of a STB are 450-560 Mbytes. The vender of STB must use either hard disks or RAM to store video data. However, both the devices cost about 100 dollars, even larger than acceptable price of a STB. To reduce the cost, some studies investigate the issue on decreasing buffer requirements at client end. The smaller the required buffers are, the lower the system cost is. Hua and Sheu proposed skyscraper broadcasting [4] scheme, which uses a novel data partition strategy and