What are the methods to alleviate this error performance degradation factor in wireless communication? To what extent would it be correct to say that pulse shaping at the transmitter and receiver also causes ISI?
Inter symbol interference is essentially caused by the dispersion of the channel. It can be avoided by leaving enough space in between the transmitted symbols. An obvious drawback of this primitive approach is that the throughput will decrease.
A smarter method is to use an equalizer, which acts like an "anti-channel", limiting the dispersion. The first equalizers processed individual samples. More recent approaches take a block of samples and process these collectively. Modern communication systems, e.g. LTE have equalizers that process such blocks in the frequency domain, which is more efficient.
One of the goals of pulse shaping/windowing at the transmitter is to limit the out-of-band radiation. However, doing this shaping within the limits of the transmitted symbol (or block) itself will distort it (*). To avoid this, the symbol (or block) is extended in time and the shaping is perfomed on this extension. Obviously, introducing such an extension will increase the risk of overlap in time with other symbols (or blocks), causing ISI.
(*) for instance in a multicarrier block transmission like OFDM, a straightforward shaping of the transmitted blocks will lead to a loss of orthogonality between the carriers and thus intercarrier-interference (ICI).
Can we say that ISI will also be prevalent in 5G wireless communication for example D2D communication?
In D2D communication two device network-assisted or without network assistance will communication with each other with controlled out-of-band emission. This will done in order to avoid the interference with neighboring cell...
When ever there is a dispersion of the transmitted symbol in time or frequency by the channel, ISI will happen. These could be caused by high data rate or multi path reflections
Hi all,
You all mentioned the multipath which is what most engineers think about (me inclusive), but (as hr. Cuypers also writes) digital communication system are sometimes designed to introduce ISI either by means of a narrower transmit or receive filter than the data rate prescribes in order to increase spectral efficiency or by using modulation with memory. Most non-linear modulation methods such as phase/frequency-modulation utilizing partial response filtering, will introduces correlation from symbol instance to symbol instance. Often these are demodulated efficiently as if it was a linear modulated signal with ISI (making use of the Laurent pulse approximation) by means of equalization.
When designing a digital communication system to zero inherent ISI in the modulator and demodulator, the channel filtering (RX+TX) are typically designed using filters fulfilling the Nyquist Criterion. Here the problem in the receiver is to sample the symbol at the correct time to obtain the zero ISI; if not sampled at the correct time, ISI will be present in the sample.
These “zero inherent ISI” systems are often used on a physical multipath channel introducing ISI. I typically find my self ending up designing T/2-spaced-equalizers for such systems,- these often being more robust even in AWGN compared to phase offset and sampling time estimation. From my point of view, too much attention is on designing “zero inherent ISI” or you could say zero ISI on the AWGN-channel, of course it is a convenient text-book reference which is easy to get going and make simulations on when no timing issues are considered. I think we can design better systems in terms of Spectral efficiency and/or power efficiency by giving up the “zero inherent ISI” requirement, even for constant complexity of the modulator/demodulator.
I am interested in hearing your opinion on the last paragraph ?
Briefly they are:
1. Limitation of channel bandwidth
2. Pulse shape
3. Pulse rate
4. Relative velocity in mobile communications
5. Fast fading
6. Limitations in A/D and D/A
None of the previous answers have got to the essence of the problem, which is a mismatch between the signal spectrum, and the channel frequency response. Things like modulation type, pulse shape and rate, windowing etc affect the spectrum of the signal. Things like transmit and receive filters, and multipath affect the response of the channel. Transmit and receive filters and multipath affect the channel frequency response. While the channel being flat gain and phase over the region where the signal has its significant energy would be sufficient for no ISI, that's neither desirable nor necessary, as it would allow excess noise into the receiver. A total channel response that is a brick-wall Nyquist filter, convolved with a symmetrical pulse, will also be ISI-free. Usually the square-root of that channel response is put on the transmitter, to control the channel spill, and on the receiver, to control noise. While a dispersive channel will cause ISI, a non-dispersive channel with a non-flat frequency response will also cause ISI.
Adaptive channel equalisation is an attempt to flatten the response of the total channel to things that are not known a priori like multipath.
I'm going to discuss one of the case of limited bandwidth.
In communication system if pulse is time limited then such a pulse can't be band limited and vice versa.
since pulse is time limited signal(in PCM) and its spectrum is band unlimited but due to limitation of channel bandwidth its spectra part is suppressed by band limited channel . Now signal spectrum is limited by channel bandwidth resulting there spreading of amplitude beyond its period and causing to interfere other pulse signal.
for more detail follow the book "Modern digital and analog communication system, B.P. Lathi". in Principles of digital data transmission chapter 7.
The channel dispersions could be in time or space. this would cause overlapping of successive symbols in time or space
- Badges
- Science topic