[Federal Register Volume 63, Number 163 (Monday, August 24, 1998)]
[Proposed Rules]
[Pages 45038-45040]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 98-22474]
=======================================================================
-----------------------------------------------------------------------
FEDERAL COMMUNICATIONS COMMISSION
47 CFR Parts 36, 54, and 69
[CC Docket Nos. 96-45 and 97-160; DA 98-1587]
Model Platform Development
AGENCY: Federal Communications Commission.
ACTION: Proposed rule.
-----------------------------------------------------------------------
SUMMARY: In the Universal Service Order, 62 FR 32862 (June 17, 1997),
the Commission stated that it would select a federal mechanism to
calculate the forward-looking economic cost of non-rural carriers
serving rural, insular, and high cost areas. The Commission determined
that it would select the ``platform'' (fixed assumptions and
algorithms) of the mechanism in one stage, and that it would select
other parts of the mechanism, including all input values, in a second
stage. Three models have been submitted to the Commission for
consideration as the platform for the federal mechanism: the Benchmark
Cost Proxy Model (BCPM), the HAI Model (HAI), and the Hybrid Cost Proxy
Model (HCPM). In an effort to move towards a result that combines the
best ideas of all parties considering these complex issues, this
document seeks comment on approaches to a model platform that combine
specific aspects from the customer location and outside plant modules
of the models under consideration.
DATES: Comments are due on or before August 28, 1998 and reply comments
are due on or before September 11, 1998.
ADDRESSES: One original and six copies of all comments and reply
comments should be sent to the Commission's Secretary, Magalie Roman
Salas, Office of the Secretary, Federal Communications Commission, 1919
M Street, N.W., Room 222, Washington, D.C. 20554. All filings should
reference CC Docket Nos. 96-45 and 97-160, and DA 98-1587. Parties also
may file comments electronically via the Internet at: http://
www.fcc.gov/e-file/ecfs.html> and ckeller@fcc.gov>. Only one copy of
an electronic submission must be submitted. In completing the
transmittal screen, commenters should include their full name, Postal
Service mailing address, and the lead docket number for this
proceeding, which is Docket No. 96-45. Parties not submitting their
comments via the Internet are also asked to submit their comments on
diskette. Parties submitting diskettes should submit them to Sheryl
Todd, Accounting Policy Division, 2100 M Street, N.W., Room 8606,
Washington, D.C. 20554. Such a submission should be on a 3.5 inch
diskette formatted in an IBM compatible format using WordPerfect 5.1
for Windows or compatible software. The diskette should be accompanied
by a cover letter and should be submitted in ``read only'' mode. The
diskette should be clearly labelled with the party's name, proceeding
(including the lead docket number in this case, Docket No. 96-45), type
of pleading (comment or reply comment), date of submission, and the
name of the electronic file on the diskette. Each diskette should
contain only one party's pleadings, preferably in a single electronic
file. In addition, parties must send copies to the Commission's copy
contractor, International Transcription Service, Inc., 1231 20th
Street, N.W., Washington, D.C. 20037.
FOR FURTHER INFORMATION CONTACT: Chuck Keller, Common Carrier Bureau,
Accounting Policy Division, (202) 418-7400 or Jeff Prisbrey, Common
Carrier Bureau, (202) 418-7400.
SUPPLEMENTARY INFORMATION: This is a summary of the Commission's
document released on August 7, 1998. The full text of this document is
available for public inspection during regular business hours in the
FCC Reference Center, Room 239, 1919 M Street, N.W., Washington, D.C.,
20554. An electronic copy of the complete
[[Page 45039]]
document also may be found on the Commission's Universal Service Web
Page at www.fcc.gov/ccb/universal__service/da981587.pdf>.
Background
1. In the Universal Service Order, 62 FR 32862 (June 17, 1997), the
Commission stated that it would select a federal mechanism to calculate
the forward-looking economic cost of non-rural carriers serving rural,
insular, and high cost areas. The Commission determined that it would
select the ``platform'' (fixed assumptions and algorithms) of the
mechanism in one stage, and that it would select other parts of the
mechanism, including all input values, in a second stage. Three models
have been submitted to the Commission for consideration as the platform
for the federal mechanism: the Benchmark Cost Proxy Model (BCPM), the
HAI Model (HAI), and the Hybrid Cost Proxy Model (HCPM). These models
have been subject to extensive review by Commission staff and outside
parties, and thousands of pages of comments have been filed regarding
their relative merits and problems. Recent ex parte meetings between
Commission staff and the model sponsors suggest that certain areas of
agreement now exist on the optimal approach to designing a platform for
the federal mechanism. In an effort to move towards a result that
combines the best ideas of all parties considering these complex
issues, this document seeks comment on approaches to a model platform
that combine specific aspects from the customer location and outside
plant modules of the models under consideration.
Issues for Comment
2. In a Further Notice of Proposed Rulemaking (Further NPRM), 62 FR
4257 (August 7, 1997), the Commission raised the possibility that the
platform for the federal mechanism may represent a synthesis of
approaches from different sources. Such a synthesis would capitalize on
the strengths of the algorithms and approaches of the models under
consideration. As the Commission stated in the Further NPRM, the goal
of this model development process is to determine the platform design
components and input values that will most accurately estimate
carriers' forward-looking economic costs. With this goal in mind, we
note that a synthesis of the approaches taken in the models under
consideration may result in a model platform with significant
advantages over each of the individual models.
3. The algorithms that identify customer locations and design
outside plant in each of the models under consideration are important
in determining the estimated costs for a wire center or study area. One
approach that might enhance the accuracy of a model's cost estimate
would be a synthesis of HAI's geocoded customer location information,
which identifies customer locations by latitude and longitude
coordinates, BCPM's assumption that customers that cannot be located
precisely are located along roads, HAI's clustering approach, and
HCPM's outside plant algorithms, which are able to design outside plant
directly, or nearly directly, to latitude and longitude coordinates.
This approach could be combined with other aspects of BCPM, HAI, or
HCPM to develop a complete model platform. While we seek comment on
this possible synthesis and on the specific issues set out below, we
note that the Commission may select as part of the federal mechanism
other combinations of algorithms not described herein. We therefore
also seek comment on any other combinations of algorithms on the record
in this proceeding that they believe would most accurately estimate
non-rural carriers' forward-looking economic costs of providing the
supported services starting July 1, 1999.
4. Customer Location Data. HAI uses data provided by PNR Associates
to identify customer locations by latitude and longitude (actual
geocode data) and creates surrogate geocodes for those customer
locations that cannot be identified (surrogate geocode data). HAI then
uses an algorithm, also provided by PNR, to identify clusters of
customers. BCPM and HCPM, on the other hand, identify customer
locations using publicly available data about the number of customers
in each Census Block. BCPM combines the Census block data about
customer location with road network data, and places customers in
microgrids based on the assumption that people are more likely to be
located along roads. In the Further NPRM, the Commission requested
comment on the availability, feasibility, and reliability of using
geocode data to determine the distribution of customers in the federal
mechanism. Many commenters from across the spectrum of the industry
agree that geocode data that identify the actual geographic locations
of customers are preferable to algorithms intended to estimate customer
locations based on information such as census block data. Although
comments on this issue have already been received, this document
provides a final opportunity for parties to comment on how a model
platform may use the most accurate customer location data available,
which in some cases may be geocode data, in the most effective manner.
We also seek comment on how the expenses for obtaining geocode data for
high cost universal service mechanisms should be recovered.
5. As many commenters have noted, actual geocode data appear to be
incomplete, particularly in low-density areas. A model, therefore, will
have to make assumptions about where non-geocoded customers are likely
to be located. Currently, the BCPM developers create surrogate geocodes
on the assumption that those customers in a census block that cannot be
geocoded are distributed along both the internal and peripheral roads
in the Census block. HAI believes that a more accurate assumption would
place surrogate geocodes along the boundary of that Census block.
Another option would be to distribute surrogate geocodes randomly
throughout an entire Census block, rather than just along its
boundaries or roads. Although comments on this issue have already been
received, this document provides a final opportunity for parties to
comment on the algorithm or combination of algorithms that would locate
most accurately those customers without actual geocodes, and on the
empirical basis for such comments. If commenters propose a different
approach than one of those described above, we seek detailed comments
on how such an approach should be implemented.
6. Grouping Customers. After determining where customers are
located using actual or surrogate geocodes, a model platform must group
customers into serving areas to design feeder and distribution plant
efficiently to those customers. In this document, we consider a model
platform that groups customers using a clustering approach because it
appears to have advantages over gridding approaches. HAI has placed the
computer code for its clustering algorithm on the record in this
proceeding. We are also releasing a clustering algorithm and a set of
cluster outputs generated from sample, surrogate geocode data. These
clusters were generated using a clustering algorithm, developed by
Commission staff, that differs somewhat from the clustering algorithm
used in HAI. We seek comment on the relative merits of HAI's clustering
algorithm and the Commission staff's clustering algorithm described in
the ``Test Data'' section, below. We also intend that parties will use
these cluster outputs to test the
[[Page 45040]]
various algorithms for designing distribution and feeder plant that are
discussed herein.
7. Designing Distribution and Feeder Plant. After identifying
groups of customers, a model must design distribution plant from the
digital loop carrier (DLC) or serving area interface (SAI) to the
customers, and feeder plant from the central office to the DLC or SAI.
In order to design distribution plant, both BCPM and HAI create square
or rectangular distribution areas and assume that the customers in each
group are uniformly spread throughout the distribution areas. While
these approaches create a predictable pattern of customer lots to which
the models may design distribution plant, both also appear to distort
the actual locations of customers when such locations can be identified
with specificity. HCPM appears to be capable of designing plant with
less distortion to customer locations. By reducing the size of its
microgrids, HCPM can associate those latitude and longitude coordinates
of each customer with a small microgrid (the version that is currently
available uses grids 360 feet on each side). With customers grouped by
a clustering algorithm, HCPM can build loop plant directly to
individual microgrids in which customers are located. Thus, HCPM could
build plant directly to every customer with an error of no more than a
few hundred feet from the actual or surrogate geocode specified for any
individual customer. We seek comment on a model that synthesizes this
approach with the use of geocode data and a clustering algorithm. We
also seek comment on the appropriate microgrid size to utilize in
building distribution plant to latitude and longitude coordinates, and
on the methods used by HCPM to subdivide microgrids into lots.
8. The feeder modules of both HAI and BCPM use a modified ``pine
tree'' algorithm that deploys main feeder routes in each of four
quadrants surrounding the central office switch, with subfeeder routes
connecting each serving area interface to the closest main feeder. In
effect, HAI and BCPM build an individual subfeeder route to nearly
every serving area (or cluster). The feeder module of HCPM allows for
more sharing among subfeeder routes by using a modified ``spanning
tree'' algorithm. The spanning tree algorithm finds the minimum
distance necessary to connect a set of remote locations to a central
point. As applied to feeder plant, this algorithm connects SAIs to the
switch. HCPM has modified the spanning tree algorithm to consider
explicitly the amount of traffic that must be carried and factors such
as the costs of cable and structures. We seek comment on these
different approaches to designing feeder plant, including on the feeder
algorithm that should be used if the Commission also adopts a model
platform that includes HCPM's distribution algorithm.
9. Test Data. As noted above, to enable parties to evaluate fully
the synthesis discussed herein, particularly the HCPM distribution and
feeder algorithm, the Bureau has made available on the Commission's
World Wide Web site a set of sample geocode data and customer clusters,
and the clustering algorithm used to generate those clusters. In
addition, an interface that converts the output of the HCPM clustering
algorithm to an appropriate input for the HCPM distribution and feeder
algorithms has been placed on the public record. These latter
algorithms overlay a grid on top of each cluster, and then assign each
customer location in the cluster to a microgrid cell within the grid
for the purpose of building distribution plant. A similar interface
could be used for HAI's cluster data point outputs, or any other set of
clustering outputs. The interface and test data are available via the
World Wide Web at http://www.fcc.gov/Bureaus/Common__Carrier/Other/
hcpm. The sample geocode data represent points randomly distributed
within the census blocks of several wire centers. Groups of the sample
geocode data have been identified according to a clustering algorithm
developed by Commission staff. By making a set of sample geocode points
publicly available and grouping them into clusters, we hope to
facilitate evaluation and analysis of this particular synthesis. We
note that these data could also be used to evaluate other potential
approaches.
List of Subjects
47 CFR Part 36
Reporting and recordkeeping requirements and Telephone.
47 CFR Part 54
Universal service.
47 CFR Part 69
Communications common carriers.
Federal Communications Commission.
James D. Schlichting,
Acting Chief, Common Carrier Bureau.
[FR Doc. 98-22474 Filed 8-21-98; 8:45 am]
BILLING CODE 6712-01-P
