Professional Documents
Culture Documents
Algorithmic PON/P2P FTTH Access Network Design For CAPEX Minimization
Algorithmic PON/P2P FTTH Access Network Design For CAPEX Minimization
net/publication/261152038
CITATIONS READS
3 546
5 authors, including:
Michael Jensen
Aalborg University
17 PUBLICATIONS 78 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Processing, Analysis and Visualization of Electrical Grid Smart Metering Data View project
All content following this page was uploaded by Jose Gutierrez on 10 June 2014.
with exact calculation of the final overall trenching and fiber CAPEX calculation in PON/P2P Design
length needed to connect a certain number of end users to the
network is not considered in another optimization literature. CAP EXP ON = T T L · (T cost + D cost) + T F L · Single strand
(3)
IV. M ETHODOLOGY
The main step to perform this analysis is to construct CAP EXP 2P = T T L · (T cost + D cost) + T F L 1 · Bundle+
the trees for each of the studied minimization techniques. +T F L 2 · Single strand
(4)
However, there are a few intermediate steps that must be
explained to fully understand the procedure followed: where:
1) Forming groups of NTs: In PON architecture design,
a 1:32 splitter is assumed to be used. In P2P architecture • T cost and D cost are trenching and duct cost respec-
deployment case, instead of splitters primary connection points tively, which values can be obtained from Table I
(PCPs) are used where the bundle of fiber (96 cables for • TTL and TFL are Total Trenching Length and Total Fiber
uplink/downlink) is divided between the users connected to the Length, respectively
same PCP. The NTs which are closer to each other are served • CP (Connection Point) denotes a PCP and a splitter in
by the same splitter/PCP. In turn, this leads to a reduction in P2P and PON cases respectively
the fiber needed to connect them to the network. This grouping • i, j, k represents the number of CPs, COs and TSPs
formation of NTs is performed using the k-means clustering respectively
algorithm [23]. • L(x,y) is the length of the path between node x and y
2) Tree formation: Typically, network links cannot be built • P is the set of unique trenched paths (x,y) or (y,x)
wherever desired but must follow some existing cable paths,
the road network and other infrastructure. A network graph 1 TSP: is a segment point available to establish connectivity to a number of
G = (V, E) is used, consisting of edges E representing the NTs
169
V. C ASE S TUDY
In this section, the described methodology is applied to a
real geographical scenario to find the best cost-wise solution
for trenching and deploying fiber in both P2P and PON
access architectures comparing SMT and A* algorithms in
the municipality of Lolland, Denmark. The total area coverage
amounts for 889 km2 [4] while its population reaches 44
635 inhabitants, based on the latest statistical data for the 1st
quarter of 2013, obtained from Denmark Statistics [26]. The
total number of households (GIS available data) is 29 175.
Out of the total number of available NT locations, trenching
and fiber reached almost 90% coverage, due to the CO-NT
Figure 3: Total PON trench length
distance limitation in P2P architecture as described in section
II.
170
Cost(e/m) R EFERENCES
Trenching (T cost) 20 [1] Dawid Nowak and John Murphy, nowakd,murphyj @ eeng.dcu.ie, FTTH:
Duct (D cost) 2 The Overview of Existing Technologies
Fiber cable (Bundle) 0.6 [2] Fibre to the Home Council Europe, FTTH Business Guide, Business
Fiber cable (Single Strand) 0.05 Committee, Second Edition, 16/01/2011
[3] Rong Zhao, Wolfgang Fischer, Edgar Aker and Pauline Rigby, Creating
Table I: Sample Cost Values a bright future: Broadband Access Technologies, 2013
[4] Michael Jensen and Jose M. Gutierrez, Upper Bound Performance Esti-
mation for Copper Based Broadband Access, Networking and Security
By substituting the values from Table I into equations 3 Section in Aalborg University, Denmark.
and 4, the total CAPEX is calculated. The cost results are [5] K. Casier, S. Verbrugge, R. Meersman, D. Colle, M. Pickavet and P.
presented in table II. Demeester, A clear and balanced view on FTTH deployment costs, J.
Inst. Telecommun. Prof., vol. 2, no. 3, pp. 2730, 2008.
[6] K. Casier, S.Verbrugge, R. Meersman, D. Colle, M. Pickavet, P. De-
meester, A clear and balanced view on FTTH deployment costs, Pro-
P2P PON ceedings of FITCE Congress 2008
Steiner 26 188 927 24 510 654 [7] Chinlon Lin, Broadband Optical Access Networks and Fiber-to-the-
A* 27 027 705 26 148 842 Home, Systems Technologies and Deployment Strategies, John Wiley &
Sons Ltd, 2006
Table II: Total costs [8] Leonid G. Kazovsky, Wei-Tao Shaw, David Gutierrez, Ning Cheng, and
Shing-Wa Wong, Next-Generation Optical Access Networks, Journal of
As shown in table II, the initial capital expenditure for P2P Lightwave Technology, Vol. 25, Issue 11, pp. 3428-3442 (2007)
architecture, irrespective of the algorithmic design approach [9] J. Prat, Next-generation FTTH passive optical networks: research towards
unlimited bandwidth access, Springer, 2008.
used, is higher compared to PON architecture. This high initial [10] Samuel Varghese. Fabrication and Characterization of All-Fiber Com-
rollout expenditure tips the balance in PON implementation ponents for Optical Access Network PhD thesis, December 2008.
direction. However, a fiber access infrastructure planning is a [11] IEEE Standards,802.3ah, Carrier Sense Multiple Access with Collision
Detection (CSMA/CD) Access Method and Physical Layer Specifications
much more complex process that should not be based solely [12] M. Chardy , M.-C. Costa , A. Faye , M. Trampont , Optimizing splitter
on the economic factor. Various aspects can affect the pref- and fiber location in a multilevel optical FTTH network , European
erence towards a specific architecture, such as geographical Journal of Operational Research, 18 May 2012.
[13] Akira Agata and Kosuke Nishimura, Suboptimal PON Network Design-
scenario, customer prioritization, scalability, upgradability and ing Algorithm for Minimizing Deployment Cost of Optical Fiber Cables,
energy efficiency issues. Looking at the rollout cost difference KDDI R & D Laboratories Inc
between the two architectures and clear clarification of the [14] S.P. van Loggerenberg , M.J. Grobler and S.E. Terblanche, Solving
the Passive Optical Network with Fiber Duct Sharing Planning Problem
current as well as the future demands and commercial targets Using Discrete Techniques
set, P2P architecture can prove to be an overall better solution. [15] Miao Lv & Xue Chen, Heuristic Based Multi-Hierarchy Passive Optical
Network Planning, 2009
VI. C ONCLUSION [16] Branka Lakic & Marek Hajduczenia, On optimized Passive Optical
Network (PON)deployment, Nokia Siemens Networks S.A., R
This work compares fiber and trenching minimization pro- [17] Anton Riedl, A Versatile Genetic Algorithm for Network Planning,
cess, using two different algorithmic approaches in P2P and Institute of Communication Networks, Munich University of Technology
PON fiber access architectures. Based on the calculations [18] Kin Fai Poon and Anis Ouali, A MILP Based Design Tool for FTTH
Access Networks with Consideration of Demand Growth,6th International
together with assumed realistic costs for trenching works, Conference on Internet Technology and Secured Transactions, December
ducts and fiber cabling, which occupy the biggest share in the 2011.
initial rollout, the preferable topology in terms of least CAPEX [19] Koen Casier, Bart Lannoo, Jan Van Ooteghem, Sofie Verbrugge, Didier
Colle, Mario Pickavet, and Piet Demeester, Game-Theoretic Optimization
is highlighted. The case study presented covers the Danish of a Fiber-to-the-Home Municipality Network Rollout, Optical Society of
municipality of Lolland. Real GIS data for the households America, 2009
locations and the road network is used to produce the final [20] S. Kulkarni, M. El-Sayed, P. Gagen, and B. Polonsky, FTTH network
economics: key parameters impacting technology decisions, Proc. Net-
output. works 2008, Budapest, Hungry, 2008
Results prove that minimum spanning tree algorithms are [21] Brandi N. Guidry, Paula P. Carson and Christoph Haon, Economic
always preferable over shortest path algorithms in terms of Implications of FTTH Networks: A Cross-Sectional Analysis, Journal of
Economic and Social policy, 2012
least trenching while for the fiber minimization the vice [22] The Socio-Economic impact of Bandwidth, Analysys Mason Limited and
versa is true, irrespective of the architecture picked. This Tech4i2 Limited, 2013
behavior is attributed to the different principle underlying [23] Christopher M Bishop, Pattern Ricognition and Machine Learning,
Springer Science + Business Media, LLC, 2006
both approaches. Shortest path algorithms always return the [24] Gabriel Robins and Alexander Zelikovsky, Minimum Steiner Tree Con-
minimum weight path connecting two nodes (fiber), regardless struction
of the rest nodes that need to be connected. On the other [25] Masoud Nosrati, Ronak Karimi and Hojat Allah Hasanvand, Investi-
gation of the * (Star) Search Algorithms: Characteristics, Methods and
hand, minimum spanning trees produce an overall minimum Approaches, 2011
weight solution for all the nodes that need to be connected [26] Danmarks Statistik, http://www.dst.dk/.
(trenching). Out of the four cases presented, applying SMT in
PON architecture returns the least deployment cost. This fact
indicates a preference for PON topology from implementation
point of view, when minimization of the total CAPEX is
prioritized.
171