With respect to the channel bandwidth issue I would like
to review some of the history in definition of the charter of the 802.20 project
and why the bandwidth "examples" were chosen, address some of the concerns about
future bandwidth extensions that I believe are based on misunderstandings and propose a way
forward.
1. Project History and Reasons for Channel Bandwidth
Selection
a. Timeliness: The project and its timeline were defined
to allow for early deployment of real world systems in licensed spectrum below
3.5 GHz and available for mobile services, rather then as a long term study
item. This requires that the specification address channel bandwidth that are
currently available. This was discussed extensively and the conclusion that
waiting for spectrum reallocation was not consistent with these goals. Future
work could address broader channels when these become available.
b. Economic Viability: The economic viability of the
project, for both service providers and equipment vendors, was predicated on it
being deployable in the near term in existing spectrum allocation and with
channel bandwidth that service providers would be willing to allocate to data
services. This would be feasible if the new system is co-deployable with
existing cellular mobile systems. This was documented in http://ieee802.org/20/SG_Docs/802m_ecsg-02-08.pdf.
which was included as an attachment to the PAR
submission to the 802 Executive Committee (see excerpt
below).
Spectrum: The AI
should be designed for deployment within existing and future licensed spectrum
below 3.5 GHz. The MBWA system frequency plan should include both paired and
unpaired channel plans with multiple bandwidths, e.g., 1.25 or 5 MHz, to allow
co-deployment with existing cellular systems. Receiver sensitivity, blocking and
selectivity specifications should be consistent with best commercial practice
for mobile wide-area terminals.
c. Differentiation from 802.16e: In order to get the
executive committee and NesCom to approve both
P802.16e and P802.20 extensive discussions were held and agreed too. Differences
between the two projects and their approach to the market were identified and
documented 802.16sgm-02/16.
One of the differences was the 802.16e would concentrate on channels with a
bandwidth greater than 5MHz, reflecting its legacy of evolving off Fixed BWA;
and that 802.20 would concentrate (at least initially) on bandwidth below ~5MHz reflecting the goals stated
above including the capability to coexist in existing cellular deployments.
Actually this is also consistent with the approach in ITU-R for systems beyond
IMT-2000, where the higher channel bandwidth first
appear for technologies supporting lower degrees of
mobility.
2. Support of Multiple Channel Bandwidth: As a result of
the granularity discussion it seems the impression has been created that wider
channel bandwidth would be supported by deploying (N) x (Basic Bandwidth) in
order to fill the total channel BW of the wider channel. I agree with Mark Cudak
that that is not desirable (I actually believe that most people that discussed
this did not have that intention). Instead the 802.20 (family of) standard(s)
should support that wider channel BW as a single fat pipe. I see the advantage
of these "fat-pipes" being multiples of some basic bandwidth in facilitating
deployment by displacing an integral number of these
"skinnier-pipes".
3. A Proposed Way Forward:
I would like to propose that 802.20 start the first set
of requirements with bandwidth of 1.25 and 5 MHz, and then developing
requirements for greater than 5 MHz, e.g. 20MHz. This rollout of successive
requirements need not be "calendar based" but can be done as soon as the
technological and business needs are understood - with some more precission then just "wider is
better".
The rationale for the proposal
is:
a.
This allows a first standard to go
out that accommodates existing well defined and understood markets; i.e. markets
based on the needs filled in the wired world by DSL and cable modem
services.
b.
In order to develop the wider
channel bandwidth systems I think we need additional requirements work to
determine whether the wider channels are required to increase the number of
users that can be supported in a cell or whether these are required to
accommodate new services that require more bandwidth per user. I believe that
such concerns would have significant impact on how these wider channel systems
are designed. To my knowledge there is no good existing base at this time on
which to base such requirements. Experience of individualized
wired wide-area wide-band services (beyond DSL/Cable rates) (such as video on
demand as compared to pay-per-view) have all indicated that
commercially these services have not been viable.
The above discussion should also make it clear that we
will need to keep an open mind on how the PHY and MAC for the wider channel
systems would be optimized for best performance.
Finally, in the debate of whether 1.25 Mhz channels with 3-4 Mbps peak are
considered broadband, I would like to point out that in the real wireless world
at least one prominent international cellular service provider defines 384Kbps
DL and 64Kbps UL as broadband capability.
Mark Klerer