COMPARISON OF CLASSICAL AND WDM BASED RING ARCHITECTURE
Laurent BLAIN(1), André HAMEL(2), Tivadar JAKAB(3), Alain SUTTER(4)
(1) France Telecom CNET LAB/SAR/SRI Technopole ANTICIPA 2 avenue Pierre Marzin F-22307 Lannion Cedex France
Phone: +33 96 05 28 14 Fax: +33 96 05 12 52 E-mail: blain@lannion.cnet.fr
(2) France Telecom CNET LAB/RIO/ARO Technopole ANTICIPA 2 avenue Pierre Marzin F-22307 Lannion Cedex France
Phone: +33 96 05 34 26 Fax: +33 96 05 32 26 E-mail: hamel@lannion.cnet.fr
(3) Technical University of Budapest Department of Telecommunications Sztoczek u. 2. H-1521 Budapest Hungary
Phone: +36 1 463 1010 Fax: +36 1 463 3266 E-mail: jakab@hit.bme.hu
(4) France Telecom CNET PAA/ATR
38-40 rue du Général Leclerc F-92131 Issy les Moulineaux Cedex France
Phone: +33 1 4529 4188 Fax: +33 1 4529 6069 E-mail: sutter@issy.cnet.fr
Abstract
The coloured section ring architecture* is based on wavelength division multiplexing technique. It is retutilising standard SDH equipment and the well known linear multiplex section protection automatic protocol and provides a full transparent 100% protected two-fibre ring architecture for SDH applications.
The architecture gives the possibility to define a logical order of nodes in the ring different from the physical (cabling) one and to multiply nodes in the same ring system.
The efficiency of the new ring architecture is demonstrated in comparison with path protected and multiplex section shared protected ring architecture.
* Coloured Section WDM Ring Architecture technique is covered by a patent owned by France Telecom.
Introduction and Motivation
Transmission networks of the present and the near future are based on SDH (SONET) technology. Many SDH based networks are in service or under installation all over the world. The self-healing ring architectures are effective solutions for full protected
transmission networks. PNOs successfully operate path protected (PP) rings and multiplex section shared protected (MSSP) rings in their networks. However, the ring link capacity often limits the optimal utilisation of expensive add-drop multiplexer (ADM) capacities in two-fibre ring architectures, because the half of the ring capacity in MSSP rings or in many cases more than the half in PP rings are reserved for protection. Figure 1 shows two examples of that situation, one for PP and
one for MSSP ring, respectively. Figure 1 Examples for saturated links in classical rings
Both rings are STM-16 systems. ADM-16s in the rings are supposed to have 16 STM-1s tributary capacities.
An adjacent (cyclic) demand pattern is given for a PP ring, each demand is 4 STM-1s. In such a case only 25%
of the ring link capacities is used for working transmission and the half of the capacity of each ADM can not be used because of the saturated links. The ineffective utilisation of ADM capacities originates from the protection technique. MSSP ring architecture provides better performance for that type of demand patterns.
In the MSSP ring example the major demand is given for a non-adjacent node pair. In spite of having 75% free ADM capacities in the other non-adjacent node pair no demand can be realised between them. The links are saturated because of the improper node order and the saturated links block the utilisation of free ADM capacities.
To overcome this limitation coloured section wavelength division multiplexing based ring architecture is proposed [1].
Coloured Section Ring Architecture
The Ring Architecture
The basic idea of the coloured section (CS) ring architecture is to take advantage of the linear multiplex section protection (MSP) protocol available in standard SDH ADMs and the wavelength routing in a two-fibre bi- directional ring to increase the transmission capacity.
In the CS ring SDH ADMs access the ring via optical add-drop multiplexers (OADM).
The purpose of an OADM is to insert a transmit signal from an SDH ADM and
extract a received signal to an SDH ADM at a particular wavelength (Figure 2). The OADM is transparent for other wavelengths not concerning the node. The transmit and receive signals can be at two different wavelengths, however there are technological limitations towards the total number of wavelengths in a wavelength multiplex.
In usual SDH rings ADMs are connected to their two neighbours using multiplex sections. In coloured section rings a particular wavelength is dedicated to each multiplex section (therefore a multiplex section is called a coloured section). This solution gives the possibility to connect two ADMs regardless of the physical order of nodes on the ring cabling infrastructure and a logical ring structure can be established (Figure 3).
The Protection Mechanism
The protection against a fibre cut is provided using the well known standardised MSP protocol. The transmitted signal is split and permanently bridged to both the working and protection systems. The decision on which signal to use is made by the receiver end analysing the signals at the receive terminal. The non- revertive single-ended protection switching
is performed in the electrical domain. No transfer of extra information is required simplifying the procedure considerably. In CS ring architecture the MSP implicates that every optical line interface is duplicated in the SDH ADMs (Figure 2). Working and protection signals are transmitted through the OADM in the opposite directions via the two fibres. Thus, a connection between two nodes uses two divers routed fibre pairs on the complementary arcs of the ring (Figure 4).
Figure 2 Node architecture in coloured section ring
Figure 3 Wavelength routing in coloured section ring
New Capabilities - New Levels of Optimisation
New Capabilities of Ring Architecture Based on to the wavelength routing a logical order of nodes, different from the physical cabling infrastructure can be realised in CS rings. This capability gives the possibility to reduce transit traffic via the intermediate nodes and links and improves the utilisation of SDH ADM capacities. Not only the connection order of nodes can be modified in CS rings, but nodes can be multiplied in the same ring as well. In case of mainly concentrated traffic patterns (e.g. in hubbed network structures) the duplicated insertion of hub nodes can improve the utilisation of SDH ADM capacities as considerably.
New Levels of Optimisation in Ring Dimensioning
In the basic SDH ring dimensioning problem a set of nodes (a network cluster) with fixed connection order and the transmission demands are given, the capacities of the ring links and the ADMs are specified. Generally, the target of the planning process is to fulfil the demands with minimum network cost (e. g. in a simplified representation with minimum number of ADMs).
As consequences of new capabilities of CS rings new dimensions of the planning problem can be identified.
Since in the basic ring dimensioning problem the connection order of nodes is fixed according to the cabling infrastructure, dimensioning CS ring the order of nodes on the logical ring layer is target of optimisation.
For example let have a cluster of N nodes, where all of the nodes are supposed to access to the same ring system. Taking into account the symmetry in a ring structure nodes can be interconnected in (N-1)!/2 different ways. Suppose that the number of available wavelengths in a wavelength multiplex section is limited in 8, so the number of nodes in one CS ring system is limited in 8 as well. That means 2520 different variation for the order of nodes.
Taking into account that a node can be multiplied in the same CS ring the number of nodes having a possible access to a given ring system is not fixed as well. Generally, the number of nodes to be multiplied strongly depends on the total amount of originated and terminated demands of the nodes and the demand structure itself.
For example in hubbed networks optimal CS ring solutions can be achieved by duplicating the hub nodes in the same ring in many cases.
These features make the CS ring dimensioning problem more complex than the classical SDH ring dimensioning .
Comparison Case Studies
The aim of the presented comparison case studies is to provide basic information about the capabilities of coloured section ring and to evaluate the architecture in comparison with classical PP and MSSP rings for real network cases.
The comparisons are performed on number of equipment basis and on cost basis. On equipment basis the efficiency of capacity utilisation and the impact of increased tributary capacities of SDH ADMs on CS ring dimensioning are studied. Cost studies are elaborated to compare the total network cost of PP MSSP and CS rings and to evaluate the extra costs for optical ADMs defined as the cost gap between MSSP and CS network costs.
The applied cost approach tries to keep the generality of the study. The cost model is simplified, only installation costs are concerned. The total network cost is defined by the total cost of installed SDH ADMs, because the management and operational costs can strongly depend on network environment. The fibre cost is not included in the total network cost.
Figure 4 Linear MSP in coloured section ring
Generally, the installation and lifetime costs together are more accurate fore comparison. However, the simplification of the cost approach does not have a significant impact on the final conclusions. The ignored management and operational costs of CS ring architecture are under the same cost for classical rings. In the CS rings there are no more equipment to manage, linear MSP is one of the simplest protection techniques, extra OADMs are simple and reliable equipment built up of passive optical elements. The CS rings need less fibre in all cases, so the simplification of cost approach looses some extra savings on behalf of CS rings. It is fact on the other hand, that WDM solution needs more spare parts for optical interfaces because of the different wavelength of transmitters.
The SDH ADM costs are derived from the average market prices and are given in relative units. Since the configuration of SDH ADMs in classical and CS rings are different the equipment (there are duplicated optical interfaces in the latter), the cost specifications are based on a functional equipment model. The elements of the model are:
- optical interfacing (shortly will be referred as optical costs)
- tributary interfacing, local cross-connecting and multiplexing (shortly will be referred as electrical costs).
An SDH ADM with two optical interfaces and with 16 STM-1s tributary capacity is supposed to cost one relative unit. Extra costs are assigned to extra optical interfaces and to increased tributary capacities. It is difficult to specify the relative cost of an optical ADM (even in the used relative units), because the equipment is not commercialised yet. Instead of that the difference between MSSP ring and CS ring cost is divided by the number of CS ring nodes on the logical ring layer and presented as cost gap for OADMs in each node. (If the cost gap is negative there are no savings on CS ring architecture comparing with MSSP ring.)
To dimension CS rings a simple heuristic was elaborated and implemented in frames of an integrated planning tool . The method tries to find proper order of nodes for each ring (if the cluster can be realised only with more than one ring system) and assign as much demand to the rings as possible in a greedy way taking into account node multiplication possibilities. The dimensioning of PP and MSSP rings was carried out by a complex software tool based on mixed integer-linear programming and simulated annealing [2].
Two scenarios are specified: Scenario 1 is to compare PP, MSSP and CS rings and Scenario 2 is to study the impact of different ADM tributary capacities on the CS ring dimensioning.
The comparison of classical SDH rings and CS ring are very sensitive for the ratio between the optical and electrical costs. Four sub-scenarios are specified with different optical-electrical cost ratios as 1:1, 1:2, 1:3 and 1:4.
The network cases are derived from realistic networks. The dimensioning is made under the following constraints and limitations:
- Maximum 8 wavelengths in a CS ring system are supposed. It limits the number of nodes in a CS ring in 8.
- No optical amplification is taken into account, the geographical size of network clusters are limited in order to meet the optical budget. (Only metropolitan area networks are concerned.)
- Eight clusters with larger demand are taken from French and Hungarian metropolitan area networks with two demand patterns for each cluster in a 4-6 year perspective.
- The number of nodes by cluster varies from 4 to 8.
- Demands in STM-1 units are only concerned, total amount of demands by cluster covers the range from 31 STM-1s to 75 STM-1.
- A significant part of the demands relates to concentrated traffic, since the clusters are taken from dual-homing network structures.
- In Scenario 1 PP, MSSP and CS WDM STM-16 two-fibre ring architecture based on ADM-16s with 16 STM-1s tributary capacity were studied.
- In Scenario 2 CS rings realised with ADM-16s with 16, 24 and 32 STM-1s tributary capacity are concerned.
The results of Scenarios 1 and 2 are summarised in Table 1 and Table 2, respectively. Besides the result of dimensioning simple theoretical lower bounds for the number of SDH ADMs are given as well.
A simple theoretical lower bound for the number of SDH ADMs needed to realise a network cluster can be defined as follows:
Let denote the graph representing the network G(N,E), where N=1,2,..n is the set of nodes with n element, and E=1,2,..e is the set of edges with e element.. The set of demands to be realised on the network is D={di,j}, where i and j are the source and the termination of the demands. The minimal number of SDH ADMs is needed to realise a node i is mi, and can be determined with the following formula:
mi = (( di j
j n
= ,
∑
1- 1) div t) + 1 (1)
where t is the tributary capacity of the ADM and div represents the integer dividing operator. A lower bound of number of needed SDH ADMs (let denote M) for the whole network cluster is given summarising (1):
M = mi
i n
∑
= 1=
i n
∑
= 1((( di j
j n
= ,
∑
1- 1) div t) + 1) (2)
If the number of installed SDH ADMs is equal to the lower bound given by (2), the efficiency of SDH ADM capacity utilisation is maximal. The 100% efficiency does not mean that the ADMs are fully loaded. It means only that the given network can not be realised with less number of equipment.
Denote, that more accurate lower bound can be established by comparing the cuts of the demand structure and the capacity of ring links. However, it can be seen from the results, that the above defined lower bound is good enough for the studied network
cases.
Analysis of Results
As can be depicted from Figure 5 the efficiency of PP rings is between 60% and 90%, the efficiency of MSSP ring varies from 50% to 80%.
(Denote, that it is not general that PP performance is better that MSSP.) Based on the protection capacities separated in wavelength and the flexibility of logical ring
definition the CS ring provides 100% efficiency for the studied network examples. With other words the two 16 STM-1s capacity links connecting nodes to each other and the possibility of optimising the logical node order give enough flexibility to achieve full utilisation of 16 STM-1s SDH ADM tributary capacities in the studied network examples.
Having the above results a question can be arisen: whether that promising performance can be kept increasing the SDH ADM tributary capacity? Figure 6 shows the number of SDH ADMs with different tributary capacities (16 STM-1s, 24 STM-1s and 32 STM-1s) for different network cases. In most cases the number of SDH ADMs decreases increasing the tributary capacity, however, the efficiency of capacity utilisation is the not maximal for some networks. If there is no difference between the number of equipment for a network case either the total demand of nodes less than the increased tributary capacity (in smaller network cases, e.g.
network case 9), thus, the application of SDH ADMs with higher tributary capacity is not relevant. Or the high transit traffic limits the better
fill of SDH ADMs (for network cases with 7 or 8 nodes, e.g. network cases 10 and 12), and it is nor worth to use larger ADMs.
Comparisons on equipment basis are informative concerning the flexibility and performance of coloured section ring architecture, however the different equipment configuration in classical ring and CS ring makes difficult to draw more
0 10 20 30 40 50 60 70 80 90 100
1 2 3 4 5 6 7 8 9 10 11 12
Network Cases Efficiency of Installed ADM Capacity Utilisation [%]
PP Ring [16]
MSSP Ring [16]
CS Ring [16]
Figure 5 Efficiency of SDH ADM capacity utilisation in different self- healing ring architecture
0 2 4 6 8 10 12
1 2 3 4 5 6 7 8 9 10 11 12
Network Cases
Total Number of SDH ADMs
Trib.: 16 STM-1s Trib.: 24 STM-1s Trib.: 32 STM-1s
Figure 6 Optimal SDH ADM tributary capacities of coloured section rings in different network cases
general conclusions. Some cost comparisons are elaborated for that purpose.
The network cost comparison based on the previously discussed simplified cost model and covers the comparison of PP, MSSP classical rings and CS WDM- based ring architecture. Four different assumptions are made for optical-electrical cost ratio (1:1, 1:2, 1:3, 1:4).
In the illustrated sub-scenario, where the optical-electrical
cost ratio is 1:2, the CS ring solution is less expensive in eight network cases from the twelve - 66% of the studied network cases (Figure 7). In the sub-scenarios with lower optical cost rates (1:3, 1:4) the number of less expensive CS realisations are 75% and 83%, respectively. Even in case of quite expensive optical functionality (1:1) the CS ring architecture is less expansive for the half of studied networks.
The cost of optical ADMs are not included in the network cost. To evaluate whether the savings are large enough to cover extra cost for OADMs the difference between the MSSP ring cost and CS rings cost is determined. In results of Figure 8 the SDH ADMs in the MSSP rings are with 16 STM-1s tributary capacity, in the CS rings are with 24 STM-1s tributary capacity.
The parameter of the comparison is the optical- electrical cost ratios.
Nearly in all cases the savings are enough for OADM (if OADM is supposed to cost 0.1 relative unit), and for the major part of network examples there are significant savings in the network costs. For SDH ADMs with 16 or 32 tributaries in CS rings the results are slightly worst for the studied networks in general, but can be better in some specific network cases.
Evaluation of Coloured Section Architecture
Based on the description of the architecture and the analysis of comparison case studies advantages and drawback of the coloured section ring architecture can be summarised as follows:
Advantages
Increase of ring capacity
The use of wavelength routing allows an increase in ring capacity. By applying MSP for protection, no capacity is used inside the SDH STM-n frame for protection, thus leading to a full use of the capacity for working transmission. In addition to that, being able to build the logical ring according to the demand pattern reduces the transit traffic. The choice of the node connections should be made in order to maximise the direct traffic between the neighbouring nodes and consequently to reduce the traffic transiting in nodes.
Existing ring upgrade
When the traffic increases and the installed networks are saturated, the main solution to provide additional transmission capacity are:
0 2 4 6 8 10 12 14 16 18 20
1 2 3 4 5 6 7 8 9 10 11 12
Network Cases
Total ADM Cost
PP Ring [16]
MSSP Ring [16]
CS Ring [16]
Basic SDH ADM relative cost components:
two optical terminations: 0.33, electrical part with 16 STM-1s tributary capacity: 0.66
Figure 7 Comparison of networks costs of different self-healing ring architecture
-0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
1 2 3 4 5 6 7 8 9 10 11 12
Network Cases Relative Cost Gap for One Additional OADM
Rel. opt. cost: 0.5 Rel. opt. cost: 0.33 Rel. opt. cost: 0.25 Rel. opt. cost: 0.2 SDH ADM tributary capacity: 24 STM-1s
Figure 8 Cost gap for optical ADMs in different network cases realised with coloured section ring architecture
- To install new SDH rings
- To upgrade to more powerful TDM ring (e.g. from an STM-4 ring to an STM-16 ring)
- To introduce WDM on the existing rings, while keeping the already installed equipment (only OADMs and MSP must be added).
The last solution, presented in this paper, has the benefit of reutilising the existing equipment. This is an asset for smooth network evolution.
Compatibility with SDH management
The coloured section ring is compatible with SDH ring management system. No new protection protocol is needed and therefor no upgrade of management software. Once the logical node order has been specified, the coloured section ring is considered by the management system as a standard SDH ring.
Drawbacks WDM restraints
The use of a defined wavelength for each multiplex section on the ring implies the use of specific optical interfaces in the SDH ADMs with transmitters at a selected wavelength. Except in case of wavelength conversion, this is the situation for every solution using WDM technology. The application of specific optical interfaces leads to operational constraints. The spare parts used for replacement of optical interfaces in case of failure must now be specific (with the same wavelength as the replaced optical interface). Multiplying the number of spare parts bring additional cost to solutions using WDM technology.
Incomplete protection
The MSP only applies to protect coloured sections. Therefore, a total failure of a node (concerning the OADM or the SDH ADM) includes the loss of the traffic terminating in the node (that is unavoidable) but also the traffic transiting in the node. This problem does not exist with the classical SDH rings. The coloured section ring architecture does not preclude the use of path protection. The use of path protection in conjunction with the solution presented here solves the problem of transit traffic loss in case of a node failure. However, the application of path protection annihilates the major advantages of the original solution by using a spare in the frames for protection. Still, as the non-direct traffic can be limited by the proper choice of node connection order (logical ring), this solution can be attractive even in conjunction with path protection. Or in an other solution path protection can be reserved for very important traffic.
Optical power budget
The use of WDM includes additional constraints on the optical budget. OADMs add insertion losses that decrease the power budget. But the most important constraint is linked to the protection section propagating on the complementary arc which length is in the worst case equal to the ring perimeter minus the shortest distance between two neighbouring OADMs. This problem does not exist for small rings (e.g. rings in urban zones), but extra optical amplification is required for longer rings, causing extra costs.
Conclusions
Coloured section ring architecture provides a new method to increase self-healing SDH ring capacities. With the applied protection technique and the new levels of optimisation as logical order of nodes and multiplication of nodes the ring capacity is increased and a more effective utilisation of SDH ADM capacities can be achieved than in classical two-fibre SDH rings.
The cost of coloured section rings is very sensitive for the cost of optical tributaries because of the applied protection technique. However, in a wide cost range the coloured section ring is less expensive than the classical path protected or multiplex section shared protected ring in many network applications. The presented cost comparison shows that CS rings can be realised with a lower cost in a metropolitan network area in most cases. In case of large total originating and terminating demands of nodes the application of SDH ADMs with 24 STM-1s tributary capacity in coloured section rings is cost effective in many cases.
Because of the good compatibility with the standard SDH equipment the coloured section ring seems to be a good alternative in the near future to upgrade existing SDH rings or install new ones at a low cost.
Acknowledgement
The comparison case studies to evaluate coloured section ring architecture were elaborated partly in frames of EURESCOM P413 "Optical Networking" project by the contributors of France Telecom and Hungarian Telecommunications Company.
The contribution of Tivadar Jakab to the presented work was supported partly by France Telecom CNET partly by the COPERNICUS 1463 ATMIN project.
References
[1] L. Blain, F. Chatter, A. Hamel, A. Sutter and V. Tholey: " Increased capacity in a MSP protection ring using WDM techniques and Optical ADM: the coloured section ring " accepted in Electronic Letters [2] ALOA Algorithmes pour L'Optimisation d'Anneaux Description des fonctionnalites d'un prototype
d'optimisation de reseaux SDH en Anneaux. Document d'Etude DE/ATR/ORI/95 France Telecom CNET
Annex
Table 1 Comparison of different SH ring architecture
Clusters Lower bound
PP ring MSSP ring CS ring # of
nodes
total demand
# of ADMs
# of rings
# of ADMs
# of rings
# of ADMs
# of rings
# of ADMs
logical structure
multiplied nodes 1 4 35 5 3 8 2 6 1 5 different yes 2 4 65 10 5 12 4 11 2 10 different yes 3 5 37 8 3 10 3 10 2 8 different no 4 5 68 11 5 13 5 13 2 11 different yes 5 6 41 7 3 12 3 12 1 7 different yes 6 6 45 8 3 12 3 11 1 8 different yes 7 6 70 11 5 16 4 15 2 11 different no 8 6 75 12 5 15 5 15 2 12 different no 9 7 37 7 2 12 2 10 1 7 different no 10 7 69 10 5 19 3 14 2 10 different no 11 8 31 8 3 16 2 12 1 8 different no 12 8 65 12 5 20 3 16 2 12 different no
Table 2 Comparison of CS ring architecture realised with different ADMs
Clusters ADM with 16 STM-1s tributary capacity
ADM with 24 STM-1s tributary capacity
ADM with 32 STM-1s tributary capacity Lower
bound
Network dimension
Lower bound
Network dimension
Lower bound
Network dimension # of
nodes
total demand
# of ADMs
# of rings
# of ADMs
# of ADMs
# of rings
# of ADMs
# of ADMs
# of rings
# of ADMs
1 4 35 5 1 5 5 1 5 4 1 4
2 4 65 10 2 10 8 1 8 5 1 6
3 5 37 8 2 8 5 1 5 5 1 5
4 5 68 11 2 11 8 1 8 6 1 6
5 6 41 7 1 7 6 1 6 6 1 6
6 6 45 8 1 8 6 1 6 6 1 6
7 6 70 11 2 11 9 2 9 6 2 9 8 6 75 12 2 12 10 2 10 7 2 10
9 7 31 7 1 7 7 1 7 7 1 7
10 7 65 10 2 10 8 2 10 7 2 10
11 8 37 8 1 8 8 1 8 8 1 8
12 8 69 12 2 12 9 2 12 8 2 12
