stds-802-16-tg4: RE: work on Multiple Access Methods for TG4
Below just a few thoughts to get us going.
On polling:
The goal probably is to minimize the transmitted overhead and delay between
a CPE deciding
it has something to send, and the allocation of an appropriate amount of
bandwidth to that
CPE by the BS, while keeping the implementation complexity and cost to a
minimum.
The evaluation is going to have to rely on a few basic parameters.
One of the parameters is the maximum number of active subscribers (i.e.
without any
oversubscription assumptions) and howmany total customers you want to
subscribe.
Now based on that, and assuming we have piggy-backed requests/polling and
persistent reservations,
and frame-durations of around 10ms and we look at typical internet traffic,
the next parameter is the
distribution of the number of new requests to be handled each frame.
With regards to subcarrier based polling, we'll need to get some grip on the
probability
of false alarms and failed detections in this noisy interference-limited
environment.
In my view, CDMA based fast bandwidth request basically requires the same
synchronization
strictness etc.. of OFDMA, and is probably only useful for the 1k mode. So
it has some complexity
issues.
Comparing these two in performance with slotted ALOHA shouldn't be too hard
once we have
the above parameters.
On ranging:
The goal probably is to mimimize the base station bandwidth allocation
necessary to allow
new stations to acquire rough synchronization and power-settings, while
keeping the implemention
complexity and cost to a minimum. Delay is not an issue here.
A basic parameter is the number of new stations expected to join.
Another is the maximum size of the cell envisioned.
On CDMA based ranging, my comment is the same as for polling.
The other ranging approach is basically unslotted ALOHA.
Our preference for the moment lies with slotted and unslotted ALOHA,
especially since doing subcarrier based
polling is extremely inefficient for mesh-topologies.
Regards,
/Nico