Using expert spreadsheets on a personal computer can help you save on long-distance circuit costs.
The largest cost of corporate telecommunications is that of long-distance circuits. The multi-vendor environment and reliable multiplexing hardware now make significant cost savings attainable in the operation of these circuits and network services. Savings of 20 percent or more can be made through proper selection, combination and operation of these circuits. This article will tell how personal computers with expert spreadsheet systems can be used to reduce circuit costs.
An expert spreadsheet system contains a set of interconnected program-driven spreadsheets that provide an interactive technical and mathematical consultant between the telecom analyst and the computer. Its knowledge base includes detailed technical expertise in tariffs, in traffic engineering, in economic optimization methods and in telecomunications strategies. Such expert systems are becoming available as expert assistants in many judgmental professions; it is no surprise that they are available for telecom facility design.
The accompanying diagram summarizes schematically the uses to which an expert spreadsheet system can be applied. The reader will find it suggestive at this point, and more understandable after finishing this article.
To show how this new software fits into the emerging telecom environment, this article will analyze a number of recurring PBX and network problems that many network managers regard as keys to lower cost. These classical telecom-management problems are usually regarded as difficult or impossible to solve at a reasonable cost. In analyzing these problems, the economic principles of long-distance circuit cost reduction will also be described.
It will become apparent, first, that the multi-vendor environment presents new cost-cutting operational solutions for these problems. Second, that these classical problems can be directly analyzed at low cost with an "expert telecom spreadsheet' system on a personal computer. The classical problems are:
Methods for selecting low-cost vendors and for combining and interfacing circuits of different vendors to reduce circuit costs.
Calculating the effects on PBX circuit costs of changing the call-routing pattern, particularly when multiple routing possibilities are present and when vendor circuit combinations allow multiple routing for many traffic destinations.
Evaluating bulk-priced circuit installations.
Assessing the tradeoffs between private lines and network service when many sets of differently tariffed private lines are installed, particularly for varying call demand.
Calculating the effects of new traffic estimates on PBX trunk-circuit groups and network circuit design. Also, calculating the effects of tariff changes on PBX circuit configuration and network design.
Analyzing the uncertainties of corporate network design. What are the optimal network switch locations? How many switches should there be? What are the optimal network switch-to-switch links?
Head-end hop off versus tail-end hop off as routing methods for off-network calls.
Is it profitable to bring a new location into a corporate network?
First, we shall address the new options provided by the multivendor alternatives. If telecom facility managers can utilize appropriate analytical procedures to organize the information for spotlighting the cost-savings opportunities, they can make substantial cost reductions. This is the role of expert spreadsheet systems that contain a set of interconnected program-driven spreadsheets that fit on personal computers.
The economics of long-distance-circuit optimization in a multi-vendor competitive environment are vastly different than in the old regulated environment, for four basic reasons:
There are many more circuit alternatives in terms of vendors, and these vendors differ greatly in tariffs, geographic availability and type of service (private line, measured use, or network service).
These vendor circuits can be combined in many ways to make cost-effective channel routes with lower costs than single-vendor circuit offerings. This was not true in the old regulated environment.
High-tech multiplexing equipment provides methods to further reduce the single-channel cost, if bulk numbers of channels are extracted from a single wideband line and are operated together between two points. Such hardware is now available in proved and reliable equipment, including TI, with advanced modulation, statistical channel interpolation, channel compression, statistical multiplexing, and wide-bandwidth digital lines. Since these alternatives can be combined with the multi-vendor combinations described above, the total number of alternatives is further increased.
The competitive environment operates to reduce prices and produce a diversity of vendor prices and circuit offerings for private lines, for network services, and for specialized data services.
The net result of these factors is that there are many more possibilities for routing voice and data traffic from a PBX to its final destination. Some of these possibilities have significantly lower costs than some of the other alternatives that currently are available to users.
Instead of setting up trunk groups with a single routing destination, it is now expedient to set up trunk groups with many routing possibilities. Each trunk group will interface to other trunk groups for routing the calls to various traffic destinations, or to external network services that distribute the calls or data messages to their final destinations. In this way, the corporate user sets up his own subnetworks. How one allocates the traffic to the various final destinations among the various trunk groups affects the call traffic volume that can be carried, because of bottleneck effects. Thus, capacity limitations on the various trunk groups affects the number of circuits that are needed, in accordance with the mathematical relations of traffic engineering.
The mathematics of the allocation process of traffic to trunk groups turns out to be mathematical programming--the minimization of a complex cost function subject to capacity constraints. While this type of mathematical optimization process is well known in economic theory, there are two additional difficulties in telecommunications facility management: the non-linear form of the costs and the necessity for combining this analysis with traffic engineering.
The Complexities Can Be Handled with Spreadsheets
However, these complexities can be handled conveniently with newly developed expert spreadsheet systems that incorporate the economics of circuit and tariff costing with mathematical optimization techniques. Increasingly, these systems have the ability to generate the telecommunications-management alternatives and to combine these with automation of all the calculations. The telecommunications analyst must provide guidance to the system in making many selections, since the number of alternatives is large and their complexity requires human decision making and intuition.
In order to describe the capabilities of expert systems for voice telecommunications, it is useful to focus on a generic type with which significant experience has been obtained--expert spreadsheet systems. It will therefore be understood that the capabilities and descriptions that follow are generic in nature, and not limited to a specific product. Yet all the capabilities described have been attained in practice.
Telecom spreadsheets provide the telecom manager the capability to know and analyze call traffic, to select the most-effective circuit vendors, to compute tariff costs automatically for individual vendors and for combinations of vendors, to automate traffic engineering and circuit economics calculations, to evaluate long-distance-circuit configurations for each PBX, to determine the most-effective call-routing pattern, and to design the least-cost corporate network that meets user demands.
The total design task of voice telecommunications is so complex that it requires a number of individual spreadsheets. These spreadsheets transfer their results to one another automatically. Each is internally programmed and performs its own specialized part of the analysis and design. By thus dividing the overall analysis into separate modules, the task of the user is simplified. Furthermore, the user can make choices and selections at many junctions in the design and evaluation process.
By automating the details of tariff data, traffic engineering or logical layout of the analysis, the user can focus on the real choices: selection of vendors and service availabilities, exploiting customer calling patterns, least-cost routing, grade of service, and cost reductions. The summary data on traffic that is needed for the circuit selection and design process is available from SMDR (station message detail recording), which produces an accounting record and statistical analysis of all voice and data message transmissions through a PBX.
Problems Can Be Worked Out Easily, Economically
The expert-systems approach to the eight classical problems of voice-facilities management are important in their own right, but also as a vehicle for describing the economic principles of long-distance-circuit cost minimization. Each of the eight problems can be worked out easily and economically with an expert system in spreadsheet form. A description of the approach to these classical voice-facilities management problems will indicate the general procedures for circuit cost reduction in a multi-vendor environment.
Several types of circuit trunk groups can be considered in today's competitive vendor environment: point-to-point private lines, measured-use network services, and private lines interfaced to a measured-use network service at a gateway. One can consider many vendors, including Bell operating companies, for both the private lines and the network services. The user can create his own subnetwork, for example, by interfacing one vendor's private lines to another vendor's measured-use network. The call cost will depend on the specifics of the vendor tariff combinations and on the mileages, mean call lengths and monthly usage. These considerations are all taken into account in calculating the measured-use costs. All this is automated in an expert spreadsheet system.
To identify the most-economical combination of circuits and measured network services for each PBX, the expert system must analyze the offered traffic at each PBX. The most cost-effective combination of circuits will depend on the unique call-traffic demand at the specific PBX. Methods must be provided to match this traffic with vendor-provided facilities. It must also generate combinations of such facilities--these are the user subnetworks or circuit facility trunk groups (CFTGs).
Expert Spreadsheet Helps Select Suitable Vendors
The expert spreadsheet system should scan and select the low-cost vendors for the traffic demand, consider and evaluate bulk-priced channels, evaluate high-tech channels derived from wideband circuits, and generate low-cost combinations of private-line vendors and/or network-service vendors at appropriate gateways.
Combinations of vendors can be generated by consolidating the traffic elements into groups, evaluating tariff alternatives for vendors, and then identifying the matching combinations of vendors. This must be done for each of the traffic destination groups (TDGs). The result of this combination process is a CFTG--a user-defined subnetwork.
In more-general terms, one can summarize the methods for generating cost-effective circuit combinations as follows: First, tabulate all the traffic elements at each PBX. These traffic elements should be consolidated into a reasonable number of offered TDGs, each of which can be individually routed. Then, scope out the tariff alternatives for private-line groups. Similarly, scope out the tariff alternatives for network services. These are then matched to form a user-defined subnetwork.
Large changes in long-distance-circuit costs can result from variation of the call-routing pattern, particularly when a number of user subnetworks are in place. This sensitivity is enhanced when time of day, time of month and seasonal routing are added.
The expert spreadsheet system performs the following functions: For a given routing pattern, it evaluates the total long-distance cost (per month); differential calculations are used in which the sensitivity of the total long-distance cost (per month) to changes in the routing pattern are obtained; and an optimization algorithm is used to obtain the best routing pattern, using the results of such differential calculations.
Long-Distance Circuit Costs Depend on Routing
Analysis will show the significant dependence of long-distance-circuit costs on the routing pattern in a multi-vendor environment. The optimum circuits and network services should be operated using the most-efficient call-routing patterns at each PBX. The cost difference between an efficient and an inefficient routing plan is very considerable. New methods should be used to derive the best call-routing pattern. What is required is an allocation pattern in which the offered traffic from each TDG is allocated among the user subnetworks. A fairly good substitute for allocation routing is a sequential overflow pattern, from each TDG to each CFTG, matched to the optimum allocation routing.
A number of new optimization results have been found to apply to multi-vendor circuit cost minimization. These include the importance of user-defined subnetworks for cost reduction, the fact that optimal routing is even more important for cost reduction than it was in the old regulated environment, and that the optimal routing pattern is obtained by allocating a certain fraction of the call traffic from each traffic destination group to each of the circuit facility trunk groups (This is not equivalent to an overflow pattern, in which one attempts to place all the traffic on one trunk group, with the overflow going to another trunk group, and its overflow to another, and so on).
It is possible to obtain a routing overflow pattern, called sequential overflow, which is a fairly good substitute for routing by allocation. The optimal routing-allocation pattern requires the use of mathematical programming. This analytical method is ideal for economic circuit optimization. Its repeated use in an iteration cycle will yield the best combination of vendors, of trunk group sizes, and the best routing-allocation pattern.
Three Basic Steps to Achieving Optimization
This optimization is performed by a design cycle that consists of three basic steps: selection of vendors, of vendor combinations and of user-defined subnetworks (this involves selection of the traffic destination groups and of the CFTGs), evaluation of the cost and grade of service of the circuit facility trunk groups at a PBX, and optimization (use of mathematical programming techniques to modify the routing and to adjust the trunk-group sizes).
One now iterates the cycle by going back to the selection step to change the vendor combinations and to reorganize the solution, in an attempt to decrease the cost further. In this way, the design cycle is iterated until further cycling makes no changes.
Bulk-priced circuits are those in which many channels, all running between two specified points, are derived from a single facility. Examples include T1, channel compression, multiplexed wideband lines, and so on. An expert system can utilize bulk circuits to set up user-defined subnetworks at low cost per channel, but with a larger number of channels than one would normally use. However, if such subnetworks can be used to carry traffic offered by a number of TDGs, the bulk capacity will all be gainfully employed.
The evaluation of such CFTGs involves the following calculations: Offered traffic for each hour of the day, by each TDG; routed traffic during each hour to each CFTG; routing tables; overflow probabilities; cost tables--measured-usage cost and fixed cost; and a summary of monthly costs and grade of service for the PBX with the specified set of CFTGs.
Offered Traffic Is Available on the Spreadsheet
The offered traffic, which summarizes the user's call demands on the PBX, is available on the spreadsheet in the form of a table that gives the total call traffic dispatched to each traffic destination for each hour of the day. The traffic volumes are given in CCS or erlangs. Also necessary are data on mean call length and total monthly traffic. Primary data of this type is available from SMDR summaries or from call-detail record summaries from the current circuit vendor. When the evaluation spreadsheet is run, the offered-traffic table can be automatically read into it.
An expert system balances one type of facility against another by a series of iterative calculations in which the cost differential from step to step of the iteration guides the cost-minimization process. The process is primarily the performance of repetitive spreadsheet calculations in which the trunk-group sizes are successively modified.
The inputs to a spreadsheet evaluation of a PBX with a given set of CFTGs includes the offered-traffic table giving the traffic offered by each traffic destination group for each hour of the day, the tariff-cost tables giving the measured-use cost (cost per traffic hour) for each TDG and each CFTG, and the fixed monthly cost for each trunk group.
The elements of the private-line versus measured-use trade-off analysis that the expert system makes practical comprise the entire span of the economic-optimization process--selection, evaluation, optimization, summary of cost and service, and cycle iteration.
Two types of cost tables must have previously been prepared in earlier spreadsheets--a measured-use cost table, giving the cost per hour of traffic for carrying traffic from each TDG on the measured-use network portion of each CFTG, and a fixed-cost table giving the monthly fixed cost of the private-line portion of each CFTG. The economic-optimization procedure configures the circuits to minimize the total monthly cost by trading off the fixed cost against the measured-use cost.
Traffic, Tariff Change-Effects Can Be Calculated
The effects of changes in traffic or in tariffs can be calculated by an expert system using differential evaluation--calculations corresponding to specific input differences. In an expert spreadsheet system, the differential evaluations are routinized and easy.
Two types of traffic tables are used. A basic input to the optimization procedure is the offered traffic--the traffic offered to the PBX within each traffic destination group. This offered traffic determines the activity of the PBX. The PBX then allocates this traffic to each of the CFTGs, in accordance with the routing pattern. The circuit trunk-group sizes then determine the traffic carried by each CFTG, in accordance with the traffic-engineering relations. It is the carried traffic that determines the grade of service.
Thus, the offered traffic, the routing pattern and the traffic-engineering relations, which depend on the trunk-group sizes, jointly determine the actual carried traffic. Mathematical programming is used to optimize this calculation and balance these elements against each other. The net effect is to minimize the total circuit cost by substituting one trunk-group for another, adjusting the trunk-group sizes, and selecting the optimal allocation of offered traffic through the routing pattern.
Increasingly, telecom managers confront the need to tie together all of the PBXs into the corporate network. A least-cost design for this corporate network should be configured to minimize the internal call costs and the external call costs of corporate users. These are the factors relevant to the design of minimum-cost networks.
A network is defined to consist of a number of switches, each interacting with its own set of node PBXs. The switches can communicate with other nodes only through their switches. The basic input data is the "from-to' node-to-node traffic matrix of the network.
To design the minimum-cost network, one must calculate switch-to-switch traffic flows in order to cost out the circuit requirements. The basic requirement is to calculate the total circuit cost under many design variations, since the minimum under all design variations must be found. The variations include the number of switches, switch locations, switch-to-switch link connections, trunk-group size, switch-to-node association, and vendor-selection alternatives.
The circuit-cost minimization under all these variations is carried out by several spreadsheets. The reliability of the network must be simultaneously assessed. A summary of cost, reliability and service is then made.
Design Steps for Minimum-Cost Networks
In general, the design of minimum-cost networks includes the following steps: First, calculate the switch-to-switch traffic flows. Then, explore the effects of variation of switch locations and numbers. Then, calculate variations in link connections and trunk-group sizes. Next, scope out the variation of vendorselection alternatives. The user must also consider the reliability consequences of the network design. Finally, summarize the total monthly cost and the resulting service level.
These are simply two different CFTGs--two different user subnetworks. The expert system can be used for a differential evaluation of two routing patterns, one with head-end hop off as a CFTG, the other with tail-end hop off replacing it. The difference in cost can be directly obtained.
The expert system can be used to break this problem into the following substeps: Feasible switch alternatives are evaluated, the cost of serving the node from the switch is calculated, the additional cost of traffic in the network due to this node is calculated, and the original cost of the node is calculated and deducted. The net is the profitability of bringing the node into the network.
The automation of these calculations and analyses fits with the evolution of the role of the telecom manager; it permits the manager to concentrate on strategy selection and cost reduction. It is not only the calculation burden that is removed, but also systemization and organization of the analysis.
Expert Spreadsheets Are Increasingly Available
Increasingly, expert spreadsheet systems are becoming available that cover all aspects of the economic optimization of long-distance circuits for a PBX or a network. In today's multi-vendor competitive environment, these automated expert spreadsheets are being used to select circuit vendors and vendor combinations for circuit trunk groups, to prepare traffic tables and tariff-cost tables, to evaluate circuit configurations for a PBX, to calculate its monthly cost of operation and grade of service, to optimize circuit configuration and call-routing patterns, and to evaluate multi-switched voice and data networks in terms of monthly cost of operation and grade of service.
The economics of circuit selection in a competitive vendor environment is a much-more complex optimization process than in the old situation, but it abounds with cost-saving possibilities. Expert spreadsheet systems provide the capability to realize such opportunities.
Photo: Telecom Expert-System Applications
|Printer friendly Cite/link Email Feedback|
|Date:||Apr 1, 1986|
|Previous Article:||Freeze-frame television can be a cost-effective choice for images that don't require motion to be understood.|
|Next Article:||Key-variable management for encryption of data.|