Dear Professor Phil
There is something that confuses me. I know that changing channels in an IP video network takes longer because of a fact related to IP addressing, but I not sure of the details. Also, why do these changes happen quickly in some cases but much more slowly in other cases?
Kevin, Houston, TX
The change in channels, in an IPTV network, is slower because of three factors. The first is because multicasting is almost always being used. Therefore, a leave packet must be sent upstream to tell the routers not to copy the current channel; then a join must be sent up stream to locate a copy of the new channel. When the router with the copy of the new channel receives the request, it begins to copy the channel downstream.
The second factor is the need for the play out device to receive an I-frame. Typically frames are presented every 1/30 of a second. But, in and MPEG-2 or H.264 stream, such I frames typically only appear every 12 or 15 frames. So there could be as long as a half second until the receiver has the required frame. I-frames are crucial because the decoding process depends on using that frame as the basis for decrypting all frames that follow until a new I-frame appears.
The third factor is whether encryption was used on the content and a key might need to be retrieved. This can take as little as a few milliseconds to a full second or more.
So, the variability depends on the availability of the I-frame, encryption key, and distance to the copy of the channel that is requested. Several technologies are being developed to address each of these issues.
Phil Hippensteel is a professor of Information Systems at Penn State Harrisburg.