As for linear ZF or MMSE detector, the coefficient of soft output is that symbol power over symbol interference power plus white noise power.
As for non-linear ZF-SIC or MMSE-SIC detector, the coefficient of soft output is that symbol power over symbol interference power plus not only white noise power but also error propagation power of decision feedback.
Does anyone know how to calculate this factor of error propagation power of decision feedback, and get a correct LLR value.
Many thanks
Jiajun Zhu
Calculating LLRs requires generating a set of vector symbol candidates. Once you have a set of candidates, the LLRs can be computed the same way as a sphere decoder does it. How to generate such set? As an starting point, I'd suggest you to try the following approach:
1. Calculated the the ZF/MMSE equalized vector symbol.
2. Before performing hard decision, find the K closest constellation vectors to your ZF/MMSE equalized vector symbol.
3. Then you will have K+1 candidates from which you can compute the LLRs. (K obviously depends on the size of the constellation set).
Hope this helps.
Hi Daniel Zahavi,
Thank you for your reply.
K-best sphere is to choose K closet constellation vectors to be candidates. However, the lost of candidates can lead to the invalid LLR values.
How to make sure that the LLR values of K-best sphere is trustable.
Many Thanks
Jiajun
Hi,
You had a very specific question regarding soft-decision based on ZF. Normally, LLRs are much more reliable when using a more robust receiver such as the sphere decoder, or QRDM. As you pointed, the candidate set generated by ZF is obviously not as reliable as one you would get by using the sphere decoder, however, that is since the ZF approach is sub-optimal and causes noise-amplification at the receiver.
If you seek a better performance then you should not limit yourself to ZF only.
Please check the following book. It is very useful
http://www.cambridge.org/us/academic/subjects/engineering/wireless-communications/large-mimo-systems
- Badges
- Science topic
- Similar topics
- Engineering
- Telecommunications Engineering