Requirements and challenges - 3 [Under
LTE]
»
E-UTRA/EUTRAN requirements - 25.913, Rel 8 «
3) Information rate loss in retransmissions
When we talked about signal power, we did not mention about amount of information contained therein. If signal power is limited, can we use in more information there ? That way we will have better information rate anyway !
What we are saying here is: can we use better modulation technique to put in more information. Modulation techniques rely on (mostly) phase and/or amplitude differences between possible information signals. If we use higher order modulation technique to put in more information, difference between two possible information signals would reduce, making the transmission more error prone.
For example in QPSK, smallest difference between two information signals is 90o, but in 8PSK, same is 45o. It is obvious that noise and interference are likely to create more problems to 8PSK than to QPSK !
To correct and detect errors, typically additional bits are added; the technique is known as coding. Lower (numerically) coding rate mean better error correction, but it would mean lesser actual information rate. Higher order modulation being more error prone, we would require lower coding rate. But this would defeat its own purpose of using higher order modulation !
This does *not* mean that we should not use higher order modulation schemes, but use it based on quality of transmission. Better the (expected or instantaneous) quality of transmission (medium/channel), higher the order of modulation that can be used with higher coding rate and so better effective information rate.
So choose optimal combinations of modulation and coding schemes - and then also dynamically change it - based on quality of transmission. This is known as Adaptive Modulation and Coding (AMC).
When above error recovery techniques fail, last resort for error correction is "retransmission" of errored information chunks (ARQ). Of course, retransmissions are to be avoided, and in fact, it is possible to do optimisations there too ! One way is to reduce the size of information chunks (on which error correction and coding schemes operate). That way even if there is an error, only smaller sized chunks need to be resent !! But due to overhead of error correction schemes and quality of transmissions not constant, having smaller sized information chunks can actually decrease overall information rate. Better way is to use retransmission schemes that would not reduce overall information rate. HARQ techniques which combine ARQ with error correction schemes help us here.
For example, in "ARQ with FEC", retransmissions are done if FEC (Forward Error Correction) fail. FEC is capable of correcting major errors.
Other pupoular HARQ techniques are HARQ with Chase combining, HARQ with Incremental Redundancy etc..
We will continue the topic with "spectral efficiency" in next article.
References: 3G Evolution: HSPA and LTE by Dahlman, Parkvall, Sköld, and Beming, LTE - From Theory to Practice, Edited by Sesia, Toufik, and Baker, and Release 8 document at 3gpp.
| Copyright © Samir Amberkar 2010-11 | § § |
| Requirements and challenges - 2 « | LTE Index | » Requirements and challenges - 4 |