Bamba--Audio and video streaming over the Internet

IBM Journal of Research and Development, Mar 1998 by Willebeek-LeMair, M H, Kumar, K G, Snible, E C

Bamba streaming architecture

A block diagram of the Bamba streaming system is presented in Figure 1. The system consists of a client and a server component. The server is a standard HTTP Web server, which contains the stored Bamba audio and video files. The client consists of a Web browser and the Bamba audio and video plug-in software.

The Bamba plug-ins are implemented as a set of dynamic link libraries that interface with the Web browser through the Netscape-defined plug-in API. Netscape has defined a set of plug-in routines that are used to communicate between the plug-in and the browser. Each plug-in library contains an initialization routine within which is declared what Netscape plug-in routines are used by the plug-in. These routines include mechanisms to create and delete instances of a plug-in, manage the plugin display window, control the flow of data streams to the plug-in, etc. In general, the plug-in is tightly integrated with the browser. Note that while Netscape was used in this example, the approach is similar for other browsers.

Bamba files may be embedded in HTML pages by means of a URL pointing to a file on an HTTP or video server. When the URL is requested, the server passes the metadata identifying the Bamba file and containing information about the file type to the client. The file type is used by the browser to launch the appropriate plug-in to play back the Bamba file.

Bamba was designed to stream clips from standard HTTP Web servers without special streaming software on the server. As such, Bamba is limited to the communication mechanisms provided by the HTTP protocol. This approach has certain advantages, the greatest of which is that it is simple and maps gracefully into the existing Web browsing architecture. As a result, content creators can easily produce Bamba audio and video clips and embed them in standard HTML [13] pages, which are then loaded onto and accessed from a standard HTTP server. Since the underlying transport protocol used by HTTP is TCP/IP, which provides reliable end-to-end network connections, no special provisions are required for handling packet loss within the network. In essence, a Bamba audio or video clip is treated like any other HTTP object, such as an HTML or JPEG [14] file. If selected, the Bamba clip is transferred to the client (browser station) as fast as TCP/IP can move it, and the client begins decoding and displaying the Bamba file as soon as the first few bytes arrive.

Since Bamba uses TCP/IP as the underlying communication protocol, the streams can traverse firewalls with no special configuration requirements. In general, systems based on UDP/IP cannot traverse firewalls without explicit permission changes in the firewall to allow passage to the UDP/IP packets. This is because UDP/IP packets are easier to imitate than TCP/IP packets, since the UDP/IP protocol involves no end-to-end handshakes or sequence numbers [15].

Bamba audio and video technology The audio and video technology used in Bamba is based on standard algorithms originally defined within the ITU H.324 standard for video telephony over regular phone lines [16]. The audio standard, G.723, specifies two bit rates: 5.3 Kb/s and 6.3 Kb/s [17]. Bamba uses the higherbit-rate CODEC, which compresses an 8-kHz input of 16-bit samples to a fixed 6.3Kb/s stream. This audio algorithm is optimized to represent speech at high quality over low-bit-rate connections. It encodes speech into 30-ms frames by means of linear predictive analysis-bysynthesis coding [17]. The input signal for the higher-bitrate coder is Multipulse Maximum Likelihood Quantization (MP-MLQ) [17].