Loading...
HomeMy WebLinkAboutA Method for Automated Planning of FTTH A Method for Automated Planning of FTTH Access Network Infrastructures Tahir M. Riaz , Ole B. Madsen , Jens M. Pedersen , Michael Jensen Department of Control Engineering, Aalborg University , Fredrik Bajersvej 7C , DK - 9 22 0 , Aalborg, Denmark Tel: + 45 96 35 8 7 00 E - mail: tahir @ control.aau.dk obm@control.aau.dk jens@control.aau.dk mj@control.aau.dk Abstract: In this paper a method for automated planning of Fiber to the Home ( FTT H) access networks is proposed. We introduced a systematic approach f or planning a ccess network infrastruct ure . The GIS data and a set of algorithms were employed to make the planning process more automatic. The method explains different steps of planning process . Considering different scenarios a number of plans can be generate d by usi ng the method . It was also possible to generate the plans in a very short time compare to the traditional planning. A case study is considered to illustrate the use and capabilities of the planning method. The method, however, does not fully automate the planning but make the planning process significantly fast. The results and discussion are presented and conclusion is given in the end . Key works: Techniques for Advance Modelling, Planning & Performance of Networks , 1 . Introduction C opper based communic ation networks are serving more than a century now , and had been dominating at all the levels of networks. In the early age of the copper based networks the number of subscribers and geogra phical area of communication were very limited. Later, t he increasi ng number of subscribers and ever spreading geographical area of communication soon brought the copper based networks to its limits. The emergence of optical fiber made it possible to keep the growth - pace up , which also le d to replace most of the backbone networks with optical fiber s . The increasing demand of bandwidth and reliability is now changing the shape of a ccess network . Th e speed of transition from copper to fiber optic based network s has been increased in the recent years . And this phenomena is ex pected to continue in future. In many part s of th e world fiber optic based access networks , commonly known as Fiber to the Home (FTTH), are being planned and implemented . It is expected that th is trend will explode over the coming few years . At current t ime , the planning of FTTH is practiced in more manual fashion. Recently, some effort s for automated planning for only wireless hotspot locations have been introduced in [ 1 ] [2] . Fore FTTH networks t here are no tools or models available which can o ffer auto mat ed planning i n a green - field situation . The model s available mostly focus on expansions and upgrades of already established infrastructures . Furthermore, they are mostly used for backbone networks [ 3 ] . At access network level, however, the planning is g enerally ad - hoc based , unstructured and nonsystematic manner [ 4 ] . This paper presents a p lanning method to make the FTTH pl anning more auto matic by using available GIS data and set of algorithms . GIS is mostly used in communication network planning for jus t visualization purposes [ 5 ]. W e present a planning process for generating long term infrastructure plan s based on a select ion of Structural Quality of Service (SQoS) parameters . SQoS concept has been introduced recently in [ 6 ] ; it aims to ensure QoS param eter at infrastructure levels . These parameters deal with the properties in the underlying ICT infrastructure in order to ensur e sufficient quality of the various upper layer services depending on the infrastructure. To make the base for the network poten tial traces we use a combination of available GIS data represent ing road s , railway tracks, foot path etc . The GIS data may be complemented with the already established ICT infrastructure (duc ts, cables, wells, masts etc). The trace data is regarded as a gr aph representing a network framework, containing the target physical infrastructure as a n embedded sub network . In order to determine the Network Termination s ( NTs ) similar GIS data is collected, representing addresses, buildings, technical installations, type of business, existing and potential industrial areas etc. The GIS data may also be complemented with already implemented NT s either as end users or as established nodal points in an implemented infrastructure, as well as predetermined Central Office positions or mandatory cross connection and exchange points. The planning method is based on a set of steps looking for a target network carried out on in an iterative manner, until the required criteria is met . The structure of t his paper follows as : in beginning, an introduction is given , describing shortly t he evolution of telecommunication networks and the importance of fiber based access network infrastructure. In section 2 , the architecture for fiber based access network will be explained . In the ne xt secti on the planning method will be presented. In this section we will describe each planning step in det a i l . To illustrate the planning method a case study of the municipality of Hals will be discussed in Section 4. The results and discussion are prese nted in section 4 .1, and finally a conclusion is given in Section 5 . 2. Network Architecture We chose the FTTH a rchitecture for fiber based access network . In this architecture no active equipment is pl aced between Access Node (AN) and user pre mises or NTs. The architecture is divided into two subclasses : Dedicated FTTH ( D - FTTH ) , and Passive Optical Network FTTH (PON - FTTH) [ 7 ] . In the D - FTTH architecture a dedicated fiber cable is used from AN to NT s . In contrast to D - FTTH, the PON - FTTH architec ture use s passive equipment between AN and NT. The passive equipment do not require e lectrical power. PON - FTTH uses wavelength multiplexing techniques to save the amount of fibe r and thus gets some cost advantage. On one hand PON - FTTH gives a cost advantag e but on the other hand it compromise s with efficiency. Both architectures are used and having their advantages and disadvantages. In this paper we will avoid this deba te and consider only D - FTTH , which is considered more future proof and reliable . 3. Planning Method The proposed planning method consists of several steps which are carried in iterated manner . In the starting steps the find ing of potential trace s for placing fibers, and normalizing GIS data for traces are performed . When the GIS data is normalized, the parameters are set and d ifferent algorithms are applied to generate the plans. The flow diagram of the planning method is depicted in Figure 1. The explanation of each step in the flow diagram is given in the following sections. Initialize Planning GIS data for traces and existing Infrastructures Selection of suitable traces and existing infrastructures Selection and grouping of nodes GIS data for Node locations Site survey of Selected trace network Indentifying and adding further potential lines Normalizing trace network (Normalizing Algorithms) Is network normalized? Setting/Modefying parameters of Node/Line locations, distances etc. Applying Planning Algorithms to generate Planning Models Plan repository Comparison of planning models from plan repository If plan is Acceptable? Ready Plans and documents Final Documentation and Map Presentation End Planning Figu re 1 : Flow diagram for the planning method . 1) GIS data , selection for potential traces and node locations In order to model the network we need the information of potential traces and node locations. To cater this we use GIS data. GIS data is vastly used for many planning purposes . There are two types of GIS data : raster maps, and vector maps. The r aster maps are commonly referred to bit - mapped , such as TIFF, JPEG etc. They are usually used for visualization purpose s . The v ect or maps are most interesting for the automated planning method. It is a strong form of representing spatial data in digital from. It uses coordinate pairs ( x, y ) to represent the feature s of interest on the landscape . The v ector maps use geometrical segmen ts such as points by coordinate pair ( x, y ) and lines (series of point coordinates) to represent objects [ 8 ] . The use of GIS vector maps is good for extracting the most interesting properties of a network, such as node degree, crossing points, segment len gths and distance s between segment points or any other related . An example of such a vector based roa d network is shown in Figure 2 . Segment Segment point Degree 1 Degree 3 Figure 2 : An example of GIS vector map used for planning. To obtain the potential location s for traces , it is important to collec t a set of vector based data . Since the representation is fairly simple, a variety of infrastructures can be integrated together. The collection can present the potentials for laying fiber in. E xisting communication infrastructure s are recommended to include so that unnecessary replication s of the infrastructure are avoided. The node locations for NT s and ANs must be a l l ocated. It is not trivial to find the location for NTs ; t he addresses for h ouses and buildings based GIS data are represented by x and y coordinates . T hey can be used to represent the NTs locations . A ddition ally , if a building has several apartments then the number of NTs should be calculated according to the number of apartments in the building. NTs are sorted according to their classes, i.e. ordinary households, business, public institutions etc. The grouping is useful in many ways; certain class es of groups can be given priority for the redundancy demands. 2) Site s urvey an d additional paths When the potential trace network and NTs are selected , the next step is to analyze them carefully . The selected trace network still can lack some small chunk s of closely located paths. Th ese can be identified in GIS maps. On otherwise, there could be some path s which are not suitable to use for the potential trace network can be removed. A site survey of the location is very useful in this regard to get the maximum potential of trace network. If paths are identified they can be added /rem oved from the potential trace network. 3) Nor malizing the potential trace network The potential t races network obtained from previous steps is need ed to normaliz e . By adding multiple networks it quite likely that paths will be replicating each other . In order to increase the efficiency of planning algorithm a new simplified representation can be generated by reducing the smaller segments by merging them into big segments. This will reduce the complexity of network in a greater extent. An example is il lustrated in Figure 3 . (a) (b) Figure 3 : Simplifying network by reducing the number of segments Some paths may be isolated from the network or complex representation s such as roundabouts and parking loops . They shoul d be further simplified. The step is repeated several time s until a fairly normaliz ed form of network is achieved. 4) Parameters selection Before generating a complete plan important parameters must be set. These parameters depend on the requirement s o f network . The value of the selected parameters is updated/changed to achieve an acceptable planning model. The most important parameters are used in this paper are :  d istance from A N to NT s  d imensioning parameters for AN s  d imensioning parameters for du cts  m anholes distance The first parameter deals to set the distance from AN to NTs. The distance from AN to NT s is set approximately a maximum fiber span distance. This is normally considered 10 to 20 km, depending what type of fiber cable s is used. The n umber of NTs per AN node is set usually according to an active equipment at AN handling maximum possible NTs ’ connections . The dimensioning parameter for ducts is approximately the maximum number of fibers allowed in ducts. The distance parameter is also s et for manholes. The maximum distance between manhole s is usually from 600 to 800m. This is due to the limitation of blowing fibers in ducts . It should be noted that we are not limited to the parameters described above. Only the most importan t and used par ameters are included here. 5) Planning a lgorithm This is the main part of the planning method . When trace network is normalized and parameters are set , th e trace network is processed further by a set of graph algorithm s . The algorithm performs calc ulations for laying fiber cables and ducts base d on the specified parameter s . C alculations performed deal with:  Selection of traces  Finding the locations for manholes  Documentation of SQoS parameters  Generating maps The planning algorithm selects the traces for fiber which gives a shortest/ cheapest path. Algorithm also documents SQoS parameters. Finally, maps are generated using a GIS tool. The detail of algorithm will not be given here in this paper due to the IPR s . 6) Plan repository and comparis on s After applying the planning algorithm the out put goes to plan repository. The plan repository is a database which holds different version s of plans based on different scen arios and parameters used. A comparison is then performed to find an acceptable plan. The criteria “ acceptable ” is meant to satisfy the requirements or goals of network plans . 7) Final d ocumentat ion and p resentation The plans should to be well documented. A good idea could be to make a document template system. The documentation process can be automated so that plans documents can be generated as well. We will not go into the detail of document ation in this paper but stress i t s importance. A presentation form of the plans is usually required for decision makers or interested parti es to sell. Including raster GIS data will provide a more natural way of presentation. Presentation could be in both paper and electronic maps form. 4. Case Study The planning method is applied on several places in Denmark. To illustrate the planning m ethod a case study of FTTH planning for the municipality of Hals will be presented here . The municipality of Hals is a small rural community in Denmark, with approximately 3500 private households and 11500 inhabitants spread over 191 square km. There are 4 9 public institutions and 604 workplaces (business or agriculture), of which 346 have more than one employer. Seven of these businesses are considered most important. Additionally , there are as much as 3400 summer houses. The private households, public ins titutions and business are mainly situated in v illages while the agriculture is more spread out. Figure 4 shows the distribution of NTs in Hals municipality. T he existing infrastructure is representative to the infrastructure in similar rural areas around the country. There are 7 central offices owned by the former national telecom, TDC. The existing places of central offices were chosen for the access network nodes. O nly 5 ANs a re decided which we re located in rather big towns. The goal is to give an over all fiber optic access to all the inhabitants of municipality with in a number of steps inside maximum of 15 years. The aim of implementing in different step i s to give the access first to prioritized NTs and then gradually to all the NTs. Figure 4 : Distribution of NTs in Hals municipality. 4.1 The r esults and discussion The proposed method was applied for the FTTH network planning for the municipality of Hals. It has been possible to generate several versions of plans using a computation power of today’s standard PC. This also gave the possibility to use several scenarios an d make the plan accordingly. Thus the method can support the scenario based planning [9] in a quite robust manner. The main motivation to use the Hals ca se study was the availability of GIS data for all the road networks and the information of NTs and their classes . The GIS information of other infrastructure was not possible to add due to our limited resources t o obtain the data. The survey was not conduc ted either but for the real world planning it is recommend . The data obtained and then normalized following the planning method steps . After the normalization process the data is ready to use for planning algorithm step. The maximum distance parameter from AN to NTs was set to 10km. Figure 4 shows the calculations for number of fibers in each trace. When it is known how much fibers are in a trace it is then it is easy to dimension the ducts . Access Nodes Fiber Cable Gandrup 1746 Hals 4171 Hou 6375 Uls tad 925 Vester Hassing 1902 Total 15119 Table 1 : Estimated fiber [Km] in Access Netwrok . The total number of manholes calculated is 1494. Three planning scenarios considered for implementation steps which are : operational step , tactical step, and strategic step. T he operational step correspond s to 3 years w h ere only the important business and public institutions get the fiber access. In the tactical step which corresponds to 5 years includes all the business and agriculture. L astly, strategic step which corresponds to 15 years covers all the households. The plans are generated according to the different scenarios presented. The total estimated fiber for the all step is show n i n Table 1 . Figure 5 : Nu mber of fibers in Access Network to all the NTs One of the examples of plan presentation is show n in Figure 3, which illustrate the density of fiber connection on traces. For the presentation a GIS tool, Mapinfo, was used . The results shown here only give an overview h ow the planning method is used. The methods employed variety of independent software tools in order to support the planning process. A unified stand alone or an add - in tool may be a best choice to maximally automate the planning process. 5 . Conclusion We have presented a systematic method for an automat ed planning of FTTH networks. We have also shown which way the GIS data can be used and useful to plan the FTTH networks . The method was applied on a case study of municipality of Hals. T he results a re quite promising, and us ing th is method a number of plans have been generated in a very short time ~ few minutes, depending o n the available computation power . If this planning process was done by using the traditional method the planning tim e could go to several days, depending upon the man force used. A benefit of d oing is a number of scenarios can be used which would give more choice to decision makers. Yet we included only some the most im portant parameters, such as distance from AN to NT, dimensioning parameters and manhole distances. The model can be extended by using more parameters. Reference: [1] A. McGibney, M. Klepal, D. Pesch, R. Mathur, “ Indoor Channel Model for link BER estimation”, Pro. Of IT&T Conference 2004, October 20 - 2 1, Ireland . [2] R. Olexa, “Implementing 802.11, 802.16, and 802.20 Wireless Networks – Planning, Troubleshooting, and Operations” , Elsevier Inc. 2005. [3] B. Sanso, P. Soriano, “Telecommunications Network Planning”, Kluwer Academic Publishers, 1999. [ 4 ] T. M. Riaz , J. M. Pedersen, O.B. Madsen “SQoS based planning using 4 - regular Grid for Optical Fiber Networks” , Proc. of Proc. of IEEE/ICACT 2005, Korea. [ 5 ] T. G. Robertazzi, “Planning Telecommunication Networks” IEEE press, New York, 1999. [ 6 ] O.B. Mad sen, T. P. Knudsen, J. M. Pedersen, “SQoS as the base for the Next Generation Global Infrastructure”, Proc. of IT&T Conference 2003, Oct o ber 22 - 23 2003, Ireland. [ 7 ] R. Ramaswami, K. N. Sivarajan, “Optical Netw o rks” Academic Press, San Diego, CA, 2002. [ 8 ] N. Schuurman, “GIS: A short Introduction ”, Blackwell Publisher, 2004. [9] T. P. Knudsen, J. M. Pedersen, O. B. Madsen, Sce nario B ased Infrastructure P lanning” , Proc. of IEEE/ICACT 2005, Korea.