The main idea behind network striping is to combine multiple physical links into one logical link to obtain additional bandwidth. The system consists of two main components :
For the current application of video conferencing, the striper will be fed by an MPEG encoder with the data to be transmitted. The encoder will grab the video data from a camera (in this case the Panasonic EggCam), carry out the compression/optimization and call the underlying API provided by the striper mechanism. Two modes of transmission are available, reliable and unreliable. The general idea is that the MPEG encoder might want to transfer the MPEG headers or reference frames using reliable transmission to minimize data loss and other frames may be transferred using unreliable transmission for higher data rate.
The striper keeps the encoder informed of the number of links that are currently active. The encoder will constantly monitor the available bandwidth and change the amount of data, and hence the quality of the picture accordingly.
Features of the Striper :
It has been found according to our measurements that the striper is capable of handling upto 122kbps of data over 4 links under normal conditions. The time required for the detection of failure and for the recovery of the link is found to be approximately 20 seconds during which the striper continues to transfer over the remaining healthy links. The network striper can be used for real time applications where high bandwidth data transfer is required over multiple phone links. The data rate of the modems can be configured as required by changing the respective parameter in the configuration file, if higher rate modems are available. The striper currently handles upto four modem links but can be made to handle more with little modification. Since the striper does not rely on any specifications from the MPEG data format, other applications such as transfer of voice data etc. can be exploited.