Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
ISI is similar to aliasing distortion?
- Thread starter pardis123
- Start date
MacGyverS2000
Electrical
- Thread starter
- #3
zappedagain
Electrical
ISI is its own type of distortion. I like to think of it as the history of what went through the channel affecting then channel performance. For example, nost binary data is randomly 1 or 0 so the average is typically 0.5 (half and half), but if you get a long string of 1s (1111111111111) or 0s it can mess up the channel because the average value starts shifting away from 0.5. It gets a lot more complicated than that as you get into the communication theory.
Hope that helps,
Z
Hope that helps,
Z
- Thread starter
- #5
Thanks alot Zappedagain
can I look at it in this way:
we have just one analog signal x(t) which is sampled respecting nyquist theorem.then these discrete values enter to the chanel hc(t).
then, we're in the end side of channel. if we look at our output signal in freq. domain: X(f)is convolved with a train of delta dirac signals and then the results multiplied with the Channel freq. response Hc(f).
the frequency shape of Hc(f)is very important in order that we don't face up Aliasing and signal be retreivable.
now, look at it in time domain,
the same sampled signal that has discrete voltages in the time domain enters the channel. the shape of channel in time domain is important in order that we can retrieve these time samples of the original signal.
--------------------------------------------
Zappedagain, i don't quite understand your point of view.
why are you looking at the input signals to the channel as a random variable and their probability?
even for a deterministic signal, if the channel time-shape is very large, we'll have ISI.
thanks
Pardis
can I look at it in this way:
we have just one analog signal x(t) which is sampled respecting nyquist theorem.then these discrete values enter to the chanel hc(t).
then, we're in the end side of channel. if we look at our output signal in freq. domain: X(f)is convolved with a train of delta dirac signals and then the results multiplied with the Channel freq. response Hc(f).
the frequency shape of Hc(f)is very important in order that we don't face up Aliasing and signal be retreivable.
now, look at it in time domain,
the same sampled signal that has discrete voltages in the time domain enters the channel. the shape of channel in time domain is important in order that we can retrieve these time samples of the original signal.
--------------------------------------------
Zappedagain, i don't quite understand your point of view.
why are you looking at the input signals to the channel as a random variable and their probability?
even for a deterministic signal, if the channel time-shape is very large, we'll have ISI.
thanks
Pardis
zappedagain
Electrical
P,
Sorry for the slow reply. I'm treating it as a digital channel, and the bits are randomly one or zero. A lot of the channel models take this into account. Another analogy is that if you AC couple the signal it will be symmetric about 0V. Some channels (aka modulation techniques) take this into account and add bits to keep it from shifting one way or the other.
Does it make sense to think of your 'channel time-shape' as the frequency response of the channel? If you try to pass faster data you'll get (low pass) distortion.
Sorry for the slow reply. I'm treating it as a digital channel, and the bits are randomly one or zero. A lot of the channel models take this into account. Another analogy is that if you AC couple the signal it will be symmetric about 0V. Some channels (aka modulation techniques) take this into account and add bits to keep it from shifting one way or the other.
Does it make sense to think of your 'channel time-shape' as the frequency response of the channel? If you try to pass faster data you'll get (low pass) distortion.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Locked
- Question
- Locked
- Question