The channel model is used to approximate the
way errors are introduced in a data stream when it is transmitted over a lossy medium. The
two models you may use in the Workshop are the Binary Symmetric Channel(BSC) and the
Additive White Gaussian Noise channel(AWGN). Both of these channel models are memoryless,
meaning that the distortion of one bit is independent of all other bits in the data
stream.The BSC defines a finite probability of bit error which is the probability that
the value of a bit in the data stream is toggled. Decoders using the BSC are considered to
be hard decision decoders because they first quantize the received bits before decoding.
The AWGN channel models the
distortion incurred by transmission over a lossy medium as the addition of a zero-mean
Gaussian random value to each bit. Decoders can take advantage of the added information of
"how close" a received value is to a valid bit value(0 or 1 for our purposes).
This type of decoding is called soft decision decoding. Because decoders that use soft
decision decoding take advantage of information that the BSC throws away, soft decision
decoders often have better error correcting capability.
When selecting the channel
model, make sure you are aware that the channel parameter values mean different things
depending on what channel model you use. For the BSC, the parameters are the actual and estimate bit error probability. Likewise, for the
AWGN channel, the parameters are the actual and estimate noise
- Illustration of the two Channel Models:
with bit error probability defined as: Pe
White Gaussian Noise Channel
with noise variance defined as: sigma squared