stds-802-16: stds-802.16:sub10. Proposal for Duplexing Scheme

[Jack Garrison <gjg@telus.net>]

With reference to the comments by Eli Pasternick, I submit the
following:

Paragraph 1.

Agreed; that a more thorough study is required. For the EHF 802.16
activities that are in progress, TDD has been attacked as being non
functional. Detailed quantitative analysis has demonstrated that these
negative assumptions are incorrect. There also seems to be an assumption
that since EHF point-to-point FDD systems have been proven to work, it
is a given that point-to-multipoint cellular FDD systems will work.
Nothing could be further from the truth. To date, not one functional FDD
frequency re-use plan has been submitted to IEEE 802.16. Aggressive FDD
frequency re-use proposals, such as those described in US Patent
5,838,670, can be demonstrated to fail miserably due to uncorrelated
rain fading.

Paragraph 2.

Submissions to IEEE 802.16 and ETSI TM4 demonstrate that, with proper
frequency re-use planning, TDD does not require intra-cell/sector or
inter-cell synchronization.

Paragraph 3.

For a given symmetric throughput TDD would require twice the bandwidth
and hence twice the power of FDD. However, in the same bandwidth, at
80/20 % directional asymmetric transmission, the added power requirement
is not significant. Further, the asymmetric operation is dynamic, it can
be reversed when required. Additionally, the asymmetry can reflect
future -yet to be defined - system applications - without compromising
the initial system design.

I would not like to be a system engineer setting the fixed FDD bandwidth
asymmetry for an initial FDD deployment. I think my employment might be
short lived. As well, FDD asymmetry might be all right for LMCS/LMDS
frequency blocks where the system operator has control over the
outbound/inbound bandwidth assignments. But what do you do with
asymmetric requirements for paired frequency assignments? Just throw
away the unused inbound spectrum?

Paragraph 4.

With proper frequency re-use, TDD requires only a guard band that
accounts for the same-cell delay time. This represents a very small
percentage of throughput loss.

Paragraph 5.

FDD is equally unproven for cellular broad band wireless access. It is
easy to brag that you have 64-QAM up and running when you have only
deployed 1 sector in 1 cell. Current allocated frequency bands surmised
only FDD. There is no justifiable reason to exclude TDD from these
bands. For every FDD paired frequency assignment, TDD gets two; a not
insignificant parameter when considering viable frequency re-use plans.

Paragraph 6.

Agree.

G. J. Garrison

DRJ & Associates
gjg@telus.net