Ida Pepe
Telecom Italia Mobile, Largo Tassoni, 323, 00186 Rome, Tel. 39 6 3900 9162 - Fax 39 6 3900 9185 - E-Mail ipepe@tim.it

1. Introduction

Traffic transport costs form a very significant component of the network costs sustained by mobile telephony operators.

A reduction in transmission costs can thus lead to a substantial recovery of competitiveness that, especially in so competittive a context as mobile telephony, must constitute a primary objective for its operators.

A fundamental role in the reduction of the costs of a telecommunications network is played by the entire process of planning a mobile telephony network, since it must follow efficiency criteria both as regards the network elements to be set out in the territory and as regards the design of the future architecture of improved networks in relation to developments envisioned for the planning period.

For the planning and design of its own transmission network, TIM had Esri develop a special s/w system called SNAP (Synchronous Network Architecture Planning).

2. Description of the TIM network architecture

During 1997 TIM chose to create its own flexible transmission network layer, based on the use of high-speed SDH connections (155 Mbit/sec) leased by the fixed-network operator, and on the use of configurable network nodes (DXC: Digital Cross-Connect, and ADM: Add-Drop Multiplexer), having functions that involved the permutation and multiplexing of the flows to be transported (2Mbit/sec).

The TIM transmission network breaks down into two main components:

1. the national network, which substantially serves the relations between the switching nodes, MSC (Mobile Switching Center). What it is is a grid network based on the use of DXC equipment in the main nodes, whose locations practically coincide with the locations of the mobile telephony network (TACS and GSM) MSCs, and on the use of 155 Mbit/sec links between these.
2. the regional networks, which serve the local relations between BTS/SRB (radio stations) and their BSC/MSC. These networks are based on the introduction of peripheral SDH nodes (secondary HUBs), which constitute the new collection points for the mobile telephony network SRB/BTS. This makes it possible to reduce the length of the low-speed queues (2 Mbit/sec) between the SRB/BTSs and the collection node (secondary HUB), and to create the connection between the collection node (secondary HUB) and the BSC/MSC (primary HUB) taken care of by the SRB/BTS by means of high-speed SDH flows (155 Mbit/sec).

Traditionally the mobile telephony operator does not have its own transmission network, and the connections are made by leasing direct connections at 2 Mbit/sec between each pair of nodes in the mobile telephony network to be connected.

The economic convenience for the operator of making up its own transmission network layer comes, within the context of telecommunications regulations in Italy, at the time when the possibility of leasing high-speed connections (PDH at 34/140 Mbit/sec) or SDH at 155 Mbit/sec) is introduced at costs more economic than connections at 2 Mbit/sec (the current cost of leasing a connection at 155 Mbit/sec, which can transport sixty-five 2 Mbit/sec flows, is on the average around one-third the cost of leasing sixty-three 2 Mbit/sec connections). Therefore, instead of leasing 2 Mbit/sec direct connections between the network nodes, the operator leases high-speed connections (in TIM’s case SDH connections at 155 Mbit/sec) between pairs of national transmission nodes.

Besides the leasing of more economic transmission resources (relative to the transport capacity made available) the operator can obtain a further reduction in overall network running costs by suitably sizing the nodes (in terms of equipment ports configuration) and the number and type of trnsmission links employed to connect each pair of nodes.

Both the definition of the network structure and its sizing are influenced by sundry factors, such as the locations of the network nodes to which the transport service is to be furnished (the national nodes are placed relative to the locations of the user nodes in such fashion as to save in the connector connections), the efficiency of use of the transport resources (a high use efficiency of the capacity made available by the high-capacity links reduces unit costs sustained by the operator for the transport of the individual 2 Mbit/sec circuit), the traffic protection policies adopted and the level of quality of the service provided.

All these factors can be translated into precise constraints in the definition of the structure and in the sizing of the transport network, and as such can be worked out by suitable procedures that can be implemented by software. This has made it possible to support network planning activities with computer-tech instruments, which reduce processing times and therefore increase the number of network solutions that can be compared, thus achieving the objective of the overall optimization of the network:

3. A functional description of the support software for the TIM transmission network planning and design

The SNAP system offers support to the main phases of the network Planning process at both the national and regional levels. Furthemore, considering that it currently also performs the function of database in the TIM transmission network, it is an aid during the whole circuits Provisioning phase by means of functions that enable verification of the feasibility of the circuits on the network that are already active and to prepare the orders for the creation of the circuits themselves.

The planning and automated running of the network has thus enabled the operator to obtain, besides the economic advantages arising from the optimization of the structure, a better control as well over the quality of the service offered and a greater flexibility and independence in the network configurations.

The SNAP system was created to respond to specific requirements expressed by TIM, such as:

1. the definition of the network structure. The system furnishes the possibility of storing in memory and of displaying on georeferential graphics maps the network nodes (both the customer networks nodes and the transmission nodes) and the connections between them (network topology).
2. network sizing: the system is able to define the routings plan; that is, to offer functions for the automatic search for the optimum paths for the routing of circuits on the network structure defined, and to determine the sizing of the nodes and of the transmission connections between them.
3. cost evaluation: SNAP makes it possible to compute the costs of the network solution found, by dynamically acquiring the rate scheme changes imposed by the supplier of the transmission connections.
4. network running and circuits provisioning: the system, by furnishing the operations involved in the creation and updating of the network database, provides support to the activity of provisioning the circuits by means of functions that make it possible to verify the circuits activation design on a network in operation and to prepare the orders for their creation.

The two main demands: for georeferentiation of the network data and for its optimization were met by creating a complete integration of the network database and of the mathematical models by using Esri Inc’s GIS (Geographic Information System) technology. It is just the use of GIS instruments that enables the user to create and update the network database directly on the video to the benefit of a more efficient and user-friendly use of the system.

The structure of the database, based on MSAccess (in which is?) maintained all the information on the network and on the national territory, consists of two main sections:

• network elements: a section embracing the network structure (nodes, connections and the parameters necessary for starting up the mathematical algorithms for the network study).

The data contained in this section of the database
is fed in by import from outboard files and editing on special data cards; furthermore, the data is also in part updated, modified and filled out downline of optimization, thus forming the archive to which access may be had for the display or verification of the results of the algorithms.

- Geographic coverages: a section embracing the cartography of Italy broken down into towns, sectors, districts and areas of service of MSC exchanges. This database, for consultation only, acts as support for the pre-processing necessary to the startup of the network-optimization functions.

Figure: block diagram of the system and of the information flows between the subsystems that form the network database

(scheme)

• Display - Pre-processing
• Report - Geographic coverages - network elements - Optimization
• DATA BASE
• Data import/export - Editing

It is obvious from the figure that the database together with the network elements can be generated and updated on the basis of data fed in directly by the user or of data resulting from the network optimizaiton/planning phase.

This fact suggested the implementation decision to create the system in modular fashion, breaking it down into two subsystems that take care, the one of administration of data and graphics analysis in the territory, and the other of network optimization, handling the two database sections described (geographic coverage and network elements).

Both subsystems use furthermore a common data base, which was conceived to be partly or wholly shared with other company sectors.

The data administration and graphics analysis subsystem makes available the following functions:

• it makes it possible to create or modify the physical network and the logical network (made up of the listing of the traffic relations to be served) by means of imports from files or graphics interfaces on Italian cartography, organized in special template/datacards for insertion, display and modification of the network elements.
• it furnishes the input data and starts up subsystem 2 of "network optimization and simulations interface", whose results it integrates into the (tabular and graphics) representation of the network, enabling the operator to accept them and to make them stable, after any necessary partial modification.
• on the basis of the physical structure of the network and of the routings (found by the algorithm or imposed by the operator) it semiautomatically creates a file of circuits among users, reporting the details on the correspondence on high-speed physical resources.
• it furnishes query and statistical analysis functions on the network data available, offering a series of reports that makes it possible to keep the development of the network and its filling status under monitoring.

The network optimization subsystem, on the other hand, enables the planning and optimization of the transmission network, bringing about its sizing in terms of nodes and of transmission connections between each pair of these.

The network optimization achieved through the SNAP system is carried out on two levels of processing: nationally or regionally and locally, it utilizing two mathematical algorithms created ad hoc for the two analysis situations.

The first mathematical model achieves a minimization of the cost of the resources to be commited by means of an analysis of the routings of the traffic relations; the second model instead is in a position to optimize the access network architecture, identifying the number and positions of the secondary HUB nodes to be inserted, with the function of acting as points of concentration for the circuits between SRB/BTS and MSC.

In the first case, for the purposes of sizing the network, the system acquires as input the structure and the state of occupation of the resources present on the network and defined by the operator in an earlier planning phase, and a forecasting model for the future structure towards which it is wished to have the network evolve during the plan period, worked up on the basis of forecasts of the development of transmission needs and of variations in the telecommunicat-ions regulations scenario as regards the supply of services. At this point, the optimization subsystem is in a position to carry out the sizing of the network elements (transmission nodes and connections between each pair of nodes) through the definition of a "routings plan", which establishes, for each logical relation, (pair of customer nodes to be connected) the routes to be followed on the network for the routing of the logic relation circuits.

The algorithm for routing the traffic relations (between users connected to a transmission network of given topology) works in successive steps.

More in detail, the subsystem acquires data on the nodes of the network under examination: number, positions, typology (the user nodes are origin or destination of transmission traffic relations: logical relations, while the transmission nodes are points of consolidation and transit of circuits between users) and the architecture of the physical connections between nodes (accompanied by additional information: rate distances between nodes, flows already leased or existing and constraints on directrixes in the leasing of means of given capacities. On the basis of this input data and of parameters configured by the operator that are necessary to define the rate schemes to which the mobile operator is subject, the mathematical model is in a position to determine, among the routes compatible with the network structure furnished and the constraints imposed , the one that minimizes the distance traveled for the routing of traffic or else the one that minimizes the unit cost of transport for the routing of the traffic.

On demand too a secondary route is sought (a protection route or a route for the sharing of the load), that is alternative to the optumum route (primary), with the constraint of minimization of the elements (nodes and sections) common to the two routes.

At the end of this search the algorithm furnishes for each point-to-point connection between nodes the indication of the best mix of transmission links and of different speeds to be leased that permits transport of the traffic at the least cost.

In support of the planning of the access network a second mathematical model for positioning the HUBs, or nodes dedicated to the gathering and and bundling of local transmission needs distributed over the territory, is used. The level of access is constituted of n local subnetworks, each made up of a given number m of access transmission nodes for access to the national network and of a principal node, typically located with the centralized entity (e.g. MSC). On each of the m nodes there converge the transmission needs of all the TIM network entities located in the territory, corresponding more or less to an administrative province, and from each node, through high-speed connections (1500 Mbit/sec or 34 Mbit/sec), these needs are transported up to the principal node.

The positioning criterion that was implemented in the mathematical model (created?) pursues the minimizing of the total interconnection costs between radio stations (BTS or RSB) and node of access. Minimizing of the costs is obtained by positioning the access nodes at the center of gravity of the areas of aggregation of BTS/SRB (most thickly concentrated in zones having high urbanization) and by sizing each node, and therefore the total number of HUBs to be introduced for each local subnetwork, in such fashion as to have flows of 2 Mbit/sec head up there in such number that the saving consequent on the reduction of the distances is not less than the cost of the connection with high-capacity flows for the connector node itself and for the principal node.

More in detail, the secondary HUBs positioning algorithm acquires the number, geographic position and channel size of the SRB/BTS, and the positions of the switching nodes, MSC, having the responsibility for each radio station, and identifies the number (possibly susceptible of definition by the operator too) and the position of the secondary HUBs under a minimum overall cost criterion for the connections -- assumed to be star connections -- between SRB/BTS and the secondary HUBs. and between these and the primary HUB connected with MSC/BSC. The algorithm permits that both the positions of some secondary HUBs and some associations between SRB/BTS and secondary HUBs be imposed and therefore not be optimized.

The figure shows the result obtained by the use of the positioning algorithm for the local TIM subnetwork in the Emilia Romagna region of Italy.

(figure)

Analysis of the results obtained by the optimization/planning activity described is immediate owing to the definition of an interface level that automates the handling of the inputs and outputs of the two algorithms.

This level takes charge in fact both of the various calculation procedures, such as for example the calculation of the rate distances between each pair of nodes of the physical network connected, of the costs of leasing circuits on the basis of the sizes deriving from the routing of traffic on the physical network, of the overall network costs, obtained as the summation of the network element costs and the cost of leasing the connections, making it possible to carry out cost comparisons between different network structures (which envision ring, tree or complete grid connections even though these architectures are not optimized but are defined by the operator)

In addition, permitted too is the management and presentation of the results of a specific session of the algorithm for identifying the secondary HUBs operating on a local subnetwork for which just one primary HUB is present (the SRB/BTS considered are those whose MSC/BSC having jurisdiction are connected to a single primary HUB) just as the integrated handling is permitted of more than one local subnetwork having more than one primary HUB, giving the operator the possibility of making any modifications and lumpings together in the overall set of results identified.