Why are both Viterbi and Reed-Solomon used in DVB-T?

Question

From my understanding, DVB-T packets go through two FEC systems, that are, Viterbi, with a data loss up to 50%, and RS, with a data loss up to 10%. Those are called external and internal coding.

I can't understand the need for the second RS coding (in that case, MPEG-TS packets 188 bytes long are added an additional 20 bytes).

More specifically, what happens to packets that are corrupted, say, 55%? Are 50% of the errors fixed by the Viterbi decoder and the remaining 5% from the RS?

Sorry for my dumbness.

Answer 1

The abilities and targets of Viterbi / RS differ considerably: Viterbi coding is done next to baseband/analog level, where each bit has a high probability of being corrupted. This is combated with a scheme, where not all combinations of e.g. '00000' through '11111' are possible, but where every other or 1/3 or 2/3 bits are correction bits calculated from the history of some N previous bits transferred.

This causes a comparably high expansion of data with the possibility of correcting typically one half of individual bit errors. One has to notice that the bit errors can occur for the correction bits as well...

This kind of bit error correction can mitigate errors mostly on AWGN channels and somewhat on Rayleigh fading (simulation model for signal fading due to moving vehicle with multi-path propagation, i.e. same signal coming from multiple paths).

Because the "window" of the Viterbi encoder is small, and when there's a burst error over the complete window (e.g. 7 bits), the encoder is not able to correct any errors. Thus a secondary coder is needed: Reed Solomon (in DVB or CD) coder works with codewords of size 8 bits, i.e. when a single bit in the codeword is corrupted, the complete codeword needs to be fixed.

The idea thus is, that the outer coder can reduce sporadic single bit errors to a manageable level, leaving basically burst errors (long period of unreceived signal) to the inner coding.