Introduction to SDH/SONT Professor Richard Harris Objectives You will be able to: Describe the basic frame format of SDH/SONT Discuss architectural issues associated with networks comprising SDH elements SDH Ring structures and options Dynamic reconfiguration methodologies Discuss mathematical models for SDH network design. Semester 2-2005 Advanced Telecommunications 43.466 Slide 2 Presentation Outline Revision of PDH technology The SDH Hierarchy Frame Formats Traffic Management with SDH Network architectures SDH network design methodologies Semester 2-2005 Advanced Telecommunications 43.466 Slide 3
PDH - Revision Plesiochronous Digital Hierarchy - PDH The existing ( old ) digital multiplexing/ transmission systems are not properly bit synchronised since clocks in different parts of the network run at different rates. The differences in clock rates (hence exact bit rates at different locations) are allowed for by bit stuffing and/or data stream buffers. Since the differences in clock rates are tolerably small and are accounted for, these systems are said to be Plesiochronous rather than Asynchronous. Semester 2-2005 Advanced Telecommunications 43.466 Slide 4 SDH - Synchronous Digital Hierarchy SONT - Synchronous Optical Network Originally proposed by Bellcore Later standardised by the ITU in recommendations G.707, G.708 and G.709. It has become known as SDH. SDH - Synchronous Digital Hierarchy SONT was proposed to take advantage of high speed digital transmission in optical fibres. SONT and SDH are similar in many respects but they are actually not quite identical. Semester 2-2005 Advanced Telecommunications 43.466 Slide 5 Issues addressed by SONT / SDH Standardised multiplexing format Optical standard for interconnection of optical equipment. Administration, Operations and Maintenance (OAM) are all part of the standard. Interworking with existing signals. Able to accommodate future applications including BISDN broadband rates. Semester 2-2005 Advanced Telecommunications 43.466 Slide 6 2
SONT Frame Structure STS- Frame Format Transport Overhead 3 octets Payload area Synchronous payload envelope (SP) 87 octets Section overhead 3 octets Line overhead 6 octets Path overhead octet 90 * 9 = 80 Octets 80 Semester 2-2005 Advanced Telecommunications 43.466 Slide 7 SDH Frame Structure General format of the STM-N frame structure 270 x N octets Section overheads 9 x N octets STM-N Payload Area 26 x N octets 9 octets Overhead Path overhead octet & Pointers 2430 N Semester 2-2005 Advanced Telecommunications 43.466 Slide 8 SDH/SONT Frame Structure A note on interpretation of the diagrams 2 3 4 90 9 94 80 2 3 9 4 90 94 80 The example shows the first 200 (approx.) Octets of a STS- Frame Semester 2-2005 Advanced Telecommunications 43.466 Slide 9 3
SDH Signal Hierarchy - STS - Synchronous Transport Signal level STS- = 5.84Mbps STM - Synchronous Transport Module STM- = 55.52Mbps Why the discrepancy? The lowest signal for ITU level 4 signal is 39.264Mbps STS signals can be multiplexed to produce the following signal levels STS-, STS-3, STS-9, STS-2, STS-8, STS-24, STS-36, STS48. (The table in the next slide shows the equivalent ITU data rates. Semester 2-2005 Advanced Telecommunications 43.466 Slide 0 SDH Signal Hierarchy - 2 SONT Designation ITU Designation Data Rate (MBPS) Payload Rate STS- 5.84 50.2 STS-3 STM- 55.52 50.336 STS-9 STM-3 466.56 45.008 STS-2 STM-4 622.08 60.344 STS-8 STM-6 933.2 902.06 STS-24 STM-8 244.6 202.688 STS-36 STM-2 866.24 804.032 STS-48 STM-6 2488.32 2405.376 Semester 2-2005 Advanced Telecommunications 43.466 Slide SDH/SONT Frame Formats The basic SONT building block is the STS- frame, which has 80 octets transmitted once every 25µsec. Check that this is equivalent to 5.84Mbps!! We can view the basic frame as a matrix of 9 rows with 90 octets each 9 x 90 = 80 octets. Transmission is one row at a time from left to right and top to bottom The first 3 columns (3 x 9 = 27 octets) of the frame are assigned to overheads. 8 octets for line overhead. 9 octets for section overhead. Semester 2-2005 Advanced Telecommunications 43.466 Slide 2 4
A Synchronous Broadband Network The main components of a synchronous broadband network are: Terminal Multiplexer (TM) Add/Drop Multiplexer (ADM) Digital Cross Connect (DXC) Network Management System (NMS) The network elements of SDH have primarily been designed for optical fibre transmission, but is equally applicable to digital microwave radio (DMR). Semester 2-2005 Advanced Telecommunications 43.466 Slide 3 Add/Drop Multiplexers Node A T Node B ADM Configurable Node C T Customer Tributaries The Add/Drop Multiplexers are used to add or drop traffic to the stream between nodes A and C. Within the ADM there is a small digital cross-connect facility which allows the traffic to be dropped or inserted, passed through or rearranged within the high speed stream. This is known as traffic grooming. Control may be local or remote through the network management system. Semester 2-2005 Advanced Telecommunications 43.466 Slide 4 Digital Cross-Connect Switches Service Service Fibre cut Route A Route B Route A Route B Service Service Network protection is achieved with DXCs and a percentage of excess bandwidth in the transmission system between nodes. Note that the network management system decides which services should have priority and downloads the appropriate switch maps to the DXCs. Semester 2-2005 Advanced Telecommunications 43.466 Slide 5 5
Link Rerouting B W-DCS C W-DCS DS3 DS3 DS3 W-DCS DS3 A Replace Link B-C With B-A-D-C Old Path of DS Demand New Path of DS Demand W-DCS (Where DS is USA term for 55Mb/s service and DS3 is the term for 45Mb/s service) Semester 2-2005 Advanced Telecommunications 43.466 Slide 6 D Path Rerouting B W-DCS C W-DCS DS3 DS3 DS3 W-DCS DS3 A Reroute A-D DS Demand Over Spare Facilities Old Path of DS Demand New Path of DS Demand W-DCS Semester 2-2005 Advanced Telecommunications 43.466 Slide 7 D Advanced Network Architectures ADM ADM ADM 622Mb/S Metropolitan Network ADM 2Gb/s DXC DXC DXC Intercapital Network 622Mb/S TRM Distribution Network TRM 55Mb/s ADM HUB Semester 2-2005 Advanced Telecommunications 43.466 Slide 8 6
Ring Structures for SONT/SDH The ability of SONT/SDH to be deployed in ring architectures rather than as strictly point-to-point or point-multipoint architectures, has become the defining feature of SONT/SDH to date. The incentive for building SONT rings was to provide a means of standardizing the traditional : protection switching in a cost-effective manner. The self-healing ring, like : diverse protection structure, is totally automatic and provides 00% restoration capability for a single fibre cable cut and equipment failure. Semester 2-2005 Advanced Telecommunications 43.466 Slide 9 Ring Structures (Continued) As technology advances and competition drives the prices for higher-rate systems towards those of lower-rate systems, SHRs tend to become even less costly to deploy than low-cost :N protection systems. Using these rings, thus improves network survivability and availability, while reducing cost. Hence, SONT self-healing rings are expected to form the major network infrastructure in future B-ISDN. Semester 2-2005 Advanced Telecommunications 43.466 Slide 20 Distinguishing Attributes There are three main features that characterise all SONT rings, each with two alternatives. These basic distinguishing attributes are listed in the table below: Attribute Number of fibres per link Direction of the signal Level of protection switching Choices 2-fibre 4-fibre Unidirectional Bidirectional Line switching Path switching Semester 2-2005 Advanced Telecommunications 43.466 Slide 2 7
Possible Ring Configurations Obviously, there are eight different SONT ring configurations arising from these attributes. To designate all these different types of ring architectures, various abbreviations are used. The abbreviations include: Uni-directional Line Switched Ring (ULSR) Bi-directional Line Switched Ring (BLSR) Uni-directional Path Switched Ring (UPSR) Bi-directional Path Switched Ring (BPSR) Semester 2-2005 Advanced Telecommunications 43.466 Slide 22 Practical Rings In actual practice, however, only three of these eight types of rings have been built on a large scale, including: fibre UPSR fibre ULSR fibre BPSR Most local exchange carriers have tended to favour 2- fibre rings of the unidirectional sort with either line or path switching. Inter-exchange carriers, on the other hand, have favoured 4-fibre BPSR Semester 2-2005 Advanced Telecommunications 43.466 Slide 23 SONT/SDH Self Healing Rings (a) Unidirectional SHR 2 (b) Bidirectional SHR with 2 fibres. 2 4 3 4 3 Counter-rotating ring Protection uses separate fibre Key: Working Protection Point-to-point traffic arrangement Working and protection use the same fibre (reserve half bandwidth for protection) Semester 2-2005 Advanced Telecommunications 43.466 Slide 24 8
Sample Rings (c) Bi-directional SHR with 4 fibres. 2 4 3 Point-to-point traffic arrangement Protection/Restoration uses separate fibres Key: Working Protection Semester 2-2005 Advanced Telecommunications 43.466 Slide 25 SONT/SDH Rings Compared Two-fibre Two-fibre Four-fibre Unidirectional Bidirectional Bidirectional Usually seen in: Rings Cities Rings Cities Rings Regional and beyond Symmetrical Delays? No Yes Yes Multiple failures Usually a problem Usually a problem Usually not a problem Bandwidth efficiency Medium Medium High Initial cost Medium Medium High xpansion costs Low Medium Low Complexity Low High Medium Semester 2-2005 Advanced Telecommunications 43.466 Slide 26 Interconnected Rings Although, SHR s are highly survivable the number of nodes on a ring, is limited by its capacity requirement and the number of hops between any two nodes. Hence, in order to utilise SONT self-healing ring technology in large networks, it is important to investigate efficient methods of interconnecting rings to overcome the problems of a large single ring. Desired features of an interconnected ring network include preservation of survivability performance of single rings, efficient routing, simplified network control mechanism, and appropriate control over problems, such as congestion. Semester 2-2005 Advanced Telecommunications 43.466 Slide 27 9
xample Ring Network SDH/SONT Ring network What do you see as the advantages and disadvantages of a network arranged in a ring fashion? Semester 2-2005 Advanced Telecommunications 43.466 Slide 28 Hierarchical self-healing network Two-level single-homing hierarchical SHR network Two-level dual-homing hierarchical SHR network Semester 2-2005 Advanced Telecommunications 43.466 Slide 29 Interconnected Ring Network Design The major issue in designing survivable SONT selfhealing ring networks is how best to utilize the unique characteristics of SHRs to meet different demand requirements in a cost-effective manner. For instance, the way rings are interconnected and the type of the rings used has an impact on the overall architecture, cost and survivability. Semester 2-2005 Advanced Telecommunications 43.466 Slide 30 0
Designing HSHR Networks The problem of designing HSHR networks can be stated as follows: Based on the given information of a network, which includes a set of nodes, the geographical distance, traffic demand between each pair of nodes, and a cost function f(x,y) of a link with length x and capacity y, We need to find an optimal Hierarchical Self Healing Ring accommodating each node and minimising the total cost of the network. Semester 2-2005 Advanced Telecommunications 43.466 Slide 3 Fibre-based Loop Network Design In the previous slide, we dealt with the design of interconnected ring networks, which can be used for designing large SONT survivable transport networks. Here we discuss the design of a fibre-based loop network i.e. an access network. Fibre facilities have been actively deployed in the feeder segment of local loop networks to reduce operating costs of present copper-based networks and to provide a fibre-optic infrastructure that will support new high bandwidth telecommunications services, such as broadband integrated switching services. Semester 2-2005 Advanced Telecommunications 43.466 Slide 32 Fibre-based Loop Network Design The design problem for a loop network is how to interconnect a set of customer locations through a ring of end offices so as to minimize the total tariff cost and provide reliability. The input elements of the problem include a set of end offices, a set of digital hubs (switches), and a set of customer locations that are geographically distributed on a plane. ach customer location is connected directly to its own designated end office, which in turn, needs to be connected to exactly one selected hub. Semester 2-2005 Advanced Telecommunications 43.466 Slide 33
Fibre-based Loop Network Design Then, the selected hubs are connected by a ring. ach hub has a fixed cost for being chosen and each link has a connection cost for being included in the solution. The objective is to design such a network at minimum cost. In other words, the aim is to connect all the end offices to at least one hub, in a most cost-effective way. Semester 2-2005 Advanced Telecommunications 43.466 Slide 34 Problem Formulation Problem: Consider min z Cij Xij i, j = Subject to: C j= T i= X = ij WX i ij W where x ij is binary variable equal to if end office i is assigned to switch j. The first set of constraints guarantees that each end office is associated with a switch. The second set ensures that the switch capacity constraint is not violated. Semester 2-2005 Advanced Telecommunications 43.466 Slide 35 Problem Solution Greedy Heuristic with Tabu Search can be used for solving this network problem. This heuristic method, presented in [2], assumes that switches may be of different types and defines the capacity of a given switch as the number of OC-3 ports that may be used by the clients. The main objective of the Greedy Heuristic is to find a good solution quickly i.e. to design a minimum cost network subject to all the constraints described above. This method incorporates features of the well-known Steiner tree problem and the travelling salesman problem. Semester 2-2005 Advanced Telecommunications 43.466 Slide 36 2
Conclusions Ring structures are the simplest method for ensuring the minimum level of protection for traffic flowing on high capacity links. Design methodologies can be complex and time consuming to implement and heuristics prevail due to the nature of the problem: Similarity to the travelling salesman problem, etc. Semester 2-2005 Advanced Telecommunications 43.466 Slide 37 Clever ways to change topology! Network on the left is a star network. Flip internal connections in central node and get a? Semester 2-2005 Advanced Telecommunications 43.466 Slide 38 Models for Dynamic Reconfiguration There have been a number of different models proposed for dynamic reconfiguration in networks: Harris (DSPN model). This is a different technology but the model appears to be relevant to the SDH context. Doverspike, Pack and Jha: Based on a stochastic model for demand at the DS0 and DS(.5Mb/s) levels. The system uses state dependent routing of Krishnan and Ott. Herzberg: Simple LP model based around a simplification to the Gopal et al and the DSPN model and uses stochastic demand elements. Gopal, Kim, Weinrib: Model begins as an NLP to optimise a traffic weighted blocking formula. Heuristics used to solve problem. Semester 2-2005 Advanced Telecommunications 43.466 Slide 39 3
Herzberg Model - Define: g = Group capacity size (g=30 if 2Mb/s trunks) C i = Available bandwidth of link i=,2,...l N p = Number of OD pairs j=,,n(n-)/2 A j = Offered traffic to OD pair j P j = Number of chains for OD pair j X j,p = Amount of bandwidth assigned to OD pair j through chain p Semester 2-2005 Advanced Telecommunications 43.466 Slide 40 Herzberg Model - 2 One possible objective function to use is to minimise the weighted traffic losses and this is done by Gopel et al. in their ITC paper. quivalently, Herzberg maximises the carried traffic through the network as follows: N p max Yj( Aj, X j) Subject to: j= Capacity Constraint N p i δ pj X j= Ci i=, 2, L jp Bandwidth constraints Pj x M jx j = X jpm p= j=, 2, N n j p on OD pair X 0 and integer jp Semester 2-2005 Advanced Telecommunications 43.466 Slide 4 Revised Herzberg Model - Gopal et al. developed an heuristic approach to the solution of this model. Herzberg exploited the nature of the Y functions and represented them as piece-wise linear functions using the coefficients given as Y jk where this represents the amount of traffic the k-th capacity unit assigned to OD pair j will carry, viz: Y jk = A j [B(A j, (k-)g) - B(A j, kg)] where B(A,n) is the rlang Loss Formula: Semester 2-2005 Advanced Telecommunications 43.466 Slide 42 4
Revised Herzberg Model - 2 x N M P p j j max Yjk X p= j= p= Subject to: x N p M j Pj j= k= p= M k= p= Pj p= X x j Pj X X jkp δ jkp i jkp X jkp = M = 0or j C jkp Semester 2-2005 Advanced Telecommunications 43.466 Slide 43 jkp i Doverspike and Pack Model In their paper to Networks '92, Doverspike and Pack describe the SONT Switched Bandwidth Network or SSBN. The SSBN is a dynamic bandwidth strategy that aims to integrate Dimensioning Network operations Customer control Network restoration Network planning. The SSBN aims to "automatically and quickly provision bandwidth, use intelligent network status based routing methods, and dynamically reconfigure the network to provide survivability and service restoration features". Semester 2-2005 Advanced Telecommunications 43.466 Slide 44 Today's Interoffice Network Design Point-Point Load Forecast Originating DS0 Forecast Originating DS Forecast (Unmultiplexed) Originating DS3 Forecast (Unmultiplexed) Grade of Design Trunk Network Service (GOS) DS0 Link Capacities Design DS0 Network Design DS Network DS Link Capacities (Multiplexed) DS3 Link Capacities (Multiplexed) Design High Rate Network (565MB,.2GB, etc.) High Rate Link Capacities Design Physical Network (Cable, Radio) Deterministic Network Design Stochastic Network Design 5
Tomorrow's Interoffice Network Design Process Point-Point Load Forecast Originating DS0 Forecast Originating DS & Multiplexed DS Forecast Design Trunk Network Grade of Service (GOS) DS Link Capacities Level of Design DS0 & DS Network Performance DS3 Link Capacities (Multiplexed) Design High Rate Network (565MB,.2GB, etc.) High Rate Link Capacities Design Physical Network (Cable, Radio) Deterministic Network Design Stochastic Network Design Demand Model At the DS level, the demand for SSBN can be characterised by DS requests for service. Their demand modelling is similar to circuit switched telephony in that it requires: Arrival rate Holding time distribution. However, it should be noted that arrivals are not Poissonian, service times are long (in the order of years perhaps!). Steady state conditions are unlikely to be achieved since the arrival rate changes before the end of a typical holding time! (The special service demand model is described in a separate paper which I have obtained from Doverspike.) Semester 2-2005 Advanced Telecommunications 43.466 Slide 47 Routing Strategies The aim here is to select a path through the digital cross connect switches to service a DS demand request. Their system is based upon a modification of the Krishnan and Ott state dependent routing system. (Described shortly.) Network Survivability The proposed SSBN method is designed to use path rerouting as described earlier. (You will see that this is more efficient than link rerouting and I used it in my DSPN model also.) Semester 2-2005 Advanced Telecommunications 43.466 Slide 48 6
Overview of State Dependent Routing (Krishnan and Ott) For each link or trunk group k in the network determine a marginal cost f k (j) of adding a call to that trunk group when j of its trunks are already busy, for j=0,, s k where s k is the number of trunks in the group. This cost represents the effect of the added call on the probable blocking of future calls on the group, and is defined to be the additional number of calls blocked on the group if the present call is accepted. Note that 0 f ( j ) and f s k k( k)= corresponds to the loss of a blocked call. Semester 2-2005 Advanced Telecommunications 43.466 Slide 49 Ideal SDR Method - Determine the cost for an arriving call in the current network state by considering each of the possible chains over which the call could be routed. If the minimum path cost for the call is < then route the call on that minimum cost chain. Otherwise, reject the call. x k = occupancy of trunk group k D 4 C 5 VAB = f( x) VACB = f2( x2) + f3( x3) V = f ( x ) + f ( x ) 2 3 A B ADB 4 4 5 5 Semester 2-2005 Advanced Telecommunications 43.466 Slide 50 Ideal SDR Method - 2 In the original work by Krishnan and Ott the cost function was a ratio of rlang Loss formulae, viz: B( sk, yk) fk ( j) = Bjy (, k ) y k is the load induced on link k by a nominal or reference routing scheme in which arriving calls are allocated amongst their admissible paths in a random fashion. It has been shown that these costs approximate a policy iteration method in a Markov decision process. Semester 2-2005 Advanced Telecommunications 43.466 Slide 5 7
Revised SDR Method At the 2th ITC in Torino, Krishnan modified the cost function to take into account the specific OD pairs in the network, viz: f ik (j) = f k (j)g jk Where g jk was calculated from parameters of the nominal traffic allocation scheme mentioned earlier. Practical implementation of the SDR scheme involves obtaining network status information at 5 minute intervals and hence the scheme has become known locally as DR-5. Semester 2-2005 Advanced Telecommunications 43.466 Slide 52 8