7.1 Design Objectives

• The project should be low cost. – At the time of project inception, low-end hardware based devices sold for under $1000. The goal for this project is that the final product could be assembled for under $500. This will make it competitive with both the current generation and next generation of hardware based solutions.

• Components readily available – One of the keys to a successful product is that all of the necessary components be readily available. In addition to creating the widest possible audience, using popular hardware also helps keep the cost low.

• Correct output - The MPEG-1 video standard is quite complex. Modifications to the original encoder must not affect the way it produces output.

• Fast – Although achieving real-time or nearly real-time compression might not seem worthwhile, the constantly decreasing cost of computational power means that a very fast software encoder would rapidly approach real-time performance.

• Simple command-line interface – The encoder should provide less than 12 user defined parameters that can be used to customize the compression process. Although the standard is quite open ended, a small set of options will provide enough flexibility for the majority of applications while making the implementation more manageable.

• Expandable – One of the key advantages that software has over hardware is that it can be changed easily. It should be possible to integrate new features into the encoder in the future.

• Three computers – The MPEG encoding source was modified specifically for parallelization across three PCs. One machine (master) sends portions of the data stream (the intensive B-frames) to the two "slave" machines, encodes the less time-consuming I-frames and P-frames and synchronizes the data for a single output stream.

• Unix-style Operating System (eg. Linux) – Due to the use of RSH to control the "slave" machines an operating system that can support the remote shell is needed.

• Raw data requirements – At present, the input stream is restricted to a series of frames in a raw YUV (4:2:2) format. The encoding algorithm manipulates the individual frames through compression and delivers them as single MPEG-1 video output. Note that audio was not considered for this project.

There are three easily identifiable indicators that can be used to judge the success of this project. The first major milestone is the creation of software that runs on a pre-configured network in one of the department's computer labs. This, in itself, is a significant accomplishment, but the added complexities of correctly configuring the hardware are not a factor. The second milestone involves making sure that the software runs on the specific hardware intended for the project. This implies that the operating system has been correctly installed, the network has been configured, and that the parallelization is fully functional. The final milestone, representing the project’s completion, is when the software has been tuned to work with the hardware to provide the fastest possible encoding process.