Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

stds-802-16-tg3: support for AAS (C802.16a-02/49)


Reading through C802.16a-02/49, I find that an excellent effort has been made to rewrite and improve the AAS support in a concise manner. It's an effort I hope will find widespread support.

There is however in my opinion one major exception to which I'd like to draw some early attention, so that the interested people have some remaining time to reconsider the matter.

This one exception is the text discussed as:

The substantial difference between OFDMA modes is how logical carriers are mapped to physical carriers.
OFDMA2 uses sub-channels that are composed of adjacent sub-carriers. A separate permutation mapping can represent this. Using this permutation in conjunction with adaptive array makes the array implementation simpler, and is therefore justified.
which finds its suggested implementation on page 212, line 55 in the 23rd suggested modification.

Imho, the main problem with this language is that it once again introduces two entirely incompatible OFDMA modes. Yet instead of clearly indicating the failure to compromise on one harmonized carrier allocation by having an OFDMA and OFDMA2 PHY, the proposal is yet again to stuff this principle incompatibility into an "optional" subclause; a situation similar to what we started out with way too many meetings ago.

There is currently, to my understanding, a single major technical argument for either carrier allocation. The fully distributed carrier allocation is claimed to be preferable for optimum frequency diversity, the contiguous carrier allocation is claimed to be preferable for minimum array processing overhead, as only one weight per contiguous allocation is needed.
In my mind, as somebody who isn't hung up on either implementation, it seems to be, entirely from a technical perspective, worth it to ponder compromising alternatives to either, such as for example a scheme in which 4 clusters of 13 contiguous carriers are distributed with a modified version of the current mandatory permutation algorithm, in order to retain significant frequency diversity and drastically reduce the AAS implementation complexity.   

This may be painful for the proponents of both current carrier allocation schemes, but imho, this is the only alternative to voting one scheme off the table.