Segmented patching broadcasting protocol for video data

Abstract

In adaptive segment-based patching scheme, the video is divided into fixed number of segments, which are transmitted over the server channels. For efficient transmission of the video segments, the server channels are classified into two types – regular and patching channels. A regular channel generally transmits fixed number of segments and a patching channel helps transmitting those segments that cannot be provided by any regular channel to the users. The number of segments transmitted by the first regular channel is decided by the number of regular channels that are allocated to the video by the video server. Other regular channels transmit pre-specified number of segments. This scheme estimates the bandwidth of the patching channels based on the requests received at the video server in terms of fixed time intervals, called time slots. The bandwidth estimation in this scheme is less accurate because for multiple requests received in a time slot more than one patching channels are used. Second, the probability distribution considered in this scheme does not satisfy the basic rule, i.e., the sum of all probabilities is not 1. In this paper, we address these issues and propose a new protocol named as Segmented Patching Broadcasting Protocol for Video Data. The average server bandwidth allocated to the patching channels is much less as compared to the adaptive segment-based patching scheme because only one patching channel is sufficient for any number of requests received in a time slot.

Introduction

In multimedia applications especially video related ones, the objects require large amount of storage and for their transmission large bandwidth is needed. The development of computing and communication technologies are making these applications feasible. The video-on-demand (VOD) services are one such application. The digital technologies have increased the data rate by many fold, which is still a scarce resource as the bandwidth demand is increasing for new video applications. So, the bandwidth should be used efficiently. For efficient usage of the bandwidth, it is important to know the popularity of videos. There have been various studies related to the videos’ popularity. In [1], [2], the video popularity has been reported to follow the Zipf distribution with the skew factor 0.271, i.e., 80% of the users’ demand is for about 20% of the most popular videos and 20% demand for 80% videos. This fact helps in designing the video server by maintaining popular videos in the main memory and less popular videos on the secondary or tertiary memory. For popular videos, efficient utilization of the bandwidth may be done using the broadcasting techniques. A broadcasting scheme implicitly assumes very high request arrival rate and does not take it explicitly into consideration. In broadcasting, same type of contents is transmitted after a regular interval. Generally, all users are not interested in viewing the same type of contents and broadcasting is not a good solution. In such cases, multicasting based techniques are more useful. For example, the senior citizens may not like to view those contents that interest children. To handle such situations, the users are grouped into different classes and for each class same type of contents is provided. This type of data transmission is called multicasting. Multicasting helps in providing different types of services each for a class of users, but for low user requests the bandwidth is under-utilized. The bandwidth is a scarce resource and one cannot bear its wastage. In such cases, the patching based schemes are more beneficial. In these types of schemes, two types of channels – full-length and patching channels – provide the video data to user requests. A full-length channel, also called a regular channel, transmits the data after a regular time interval and the missing video data is provided by the patching channels of appropriate lengths. In patching based schemes, a user downloads and stores the future playback data into the system buffer from the on-going full-length channels. The remaining data are provided by allowing a fresh transmission using a patching channel. Some important patching based schemes have been discussed in [3], [4], [5], [6], [7]. There are other classes of schemes, which are based on batching. In batching based schemes, the requests are collected in batches and then a regular channel is used to provide them the video data. In [8], adaptive segment-based patching scheme has been discussed. This scheme reduces the bandwidth requirement as compared to the standard patching based scheme by combining the benefits of batching and patching techniques. This scheme, however, has some drawbacks: (i) The bandwidth of the patching channels has been derived using the multiple patching channels for multiple requests received in a time slot. This fact is less accurate because any number of requests received in a time slot can be serviced by just one patching channel. (ii) The probability distribution does not follow the basic rule, i.e., sum of all probabilities is not 1 (refer Eqs. (1), (4). In this paper, we address these issues and propose a new scheme, called Segmented Patching Broadcasting Protocol for Video Data.

The remaining paper is organized as follows. Section 2 reviews the adaptive segment-based patching scheme as the proposed scheme addresses its drawbacks and provides their solutions. Henceforth, the adaptive segment-based patching scheme will be referred to as adaptive scheme in this paper. In Section 3, the proposed scheme, i.e., segmented patching protocol for video data is discussed. Section 4 presents the results. Finally, in Section 5, we conclude the paper.