[Interleave Depth]
>>Shame on Martin De Prycker. He should know better than to publish incorrect
statements.
>> Examples:
>> - There ARE variable bit rate CAP chips available today.
Since when last week?
.
>> - CAP requires a SMALLER interleaving depth than DMT.
Do you realize what you are saying here?
A little background for the non technical types:
FEC (Forward Error Correction) is a data error correcting technique used to transmit data through a channel with high energy, short duration (time), noise (this is called Impulse noise). This noise literally blanks the transmitted information on the channel. Its like the wire was cut and the transmitted bits were allowed to drop on to the ground, then the wire is connected again within microseconds, the bits are gone forever. FEC has been used for years in disc drives and CD players. FEC reconstructs the lost data at the receiving end and therefore the lost data doesn't need to be retransmitted again (consuming valuable bandwidth). Now for packet data (internet data), retransmission is no big deal, the protocol supports retransmission, the user never notices a retransmission, FEC is a waste. On the other hand, for video it is a different story. Video information is streamed to the user and errors are very noticeable as blank squares on the screen. FEC can make a large difference in viewing quality. To implement FEC requires an encoder at the transmitter and a decoder at the receiver. The encoder adds additional bits to the transmitted information (about 3% overhead) and may employ a technique called interleaving. Now I'm not going to risk loosing my audience by explaining Interleave depth so I'll give a rule. Interleave depth is a function of both data rate AND noise duration. If more bits get destroyed, interleave depth must get larger in order to recover the data at the receiving end. If the noise duration is short, the interleave depth for 1.5Mbps is smaller than if the channel data rate were 8Mbps. As the noise duration increases in time, so must the interleave depth for either 1.5 or 8Mbps. So why don't we just crank interleave depth on full blast and be done with it. As with most everything in engineering, there is no free lunch. Larger interleave depth takes more silicon but the quality added to the system will win this argument. The main system drawback is that it increases latency. You won't notice this when looking at a video but if you are playing an interactive video game, you will. It can also be a factor for some packet protocols but I'm sure Billy Gates will fix this one. If it becomes a factor in DMT systems, it can be adjusted.
So Luke, are you saying that CAP, with its SMALLER interleave depth, is more susceptible to impluse noise or is a lower bandwidth system (or both) than DMT systems??
Later
Steve |
