Background of the invention Link pricing in communications networks has traditionally been re- garded as a technical issue. The monetary constraints for market entry were based on the need of investment return levels. In a static market with tariffs, monopolies and slowly changing and hard to measure competitor pricing this has been a reasonable means of conducting business.

The new availability of market information has enabled novel means of pricing capacity in a network. The new constraints are not the in- vestments, which may be taken as sunken costs, but the easily available competitor pricing and the ability of using a multi-hop path to replace a com- munications conduit.

The current sales of capacity on links are based on a number of methods, including the following ones: 1. Cost based pricing with investment return assumptions. In this mechanism one takes the investment and other costs, as- sumed time to use, and requirements on investment return. These can be used to derive the price needed to meet the requirements.

2. Competitor analysis. One can price the links based on knowledge, or rumours on competitor pricing, setting the current price based on the latest competitor price, perhaps together with a discount.

3. Extrapolation from products of differing capacity or fram- ing. In bringing a new capacity or a framing mechanism to the market the price on the same link for different capacities can be extrapolated, and/or the price for a different framing can be taken as is as a basis for pricing.

4. Concepts of customer, wholesale and excess pricing. Pric- ing can depend on the customer as well. A different method can be used for customer, wholesale and excess pricing. Customers

may for example be priced using method 1, wholesale customers using method 2 and excess at the highest price that the market will bear for total sale of excess.

5. So called IRU, or irrevocable right of use pricing is a com- mon method of selling capacity. In such the capacity use is priced as in the other methods, but the payment consists of a large initial lump sum and a fixed long term lease.

6. Long term deals. The pricing for a link can be done for a very long term. In such a scenario the pricing is re-evaluated on a regular basis.

7. Split links, each side of oceanic link identically priced by different operators. A single link may be sold as a combination of two half-links, with each being bought from different parties. In such a case the different parts may be priced differently.

The concept of"safe pricing range"is taken to mean the pricing range within which the price is competitive in comparison with the other available alternatives, and not below the price in which the product or service in question can be bought to replace in part or in whole other products.

By a"network"we mean a set of links, interconnected in whole or in part, connected at points called nodes, with the links having at least one additive capability. An example of a trivial additive capability is a constant value, such as 1, for each link. In this document, the network is taken to mean weighted graph representing a telecommunications network.

A"link", or"edge"is taken to be a link in a telecommunications network.

The primary additive constraint is taken to be price, unless explic- itly defined. Other constraints may exist. When the price of the link is dis- cussed, any other primary additive constraint could have been used.

The"cumulation function"is taken to be addition.

A"vertex"is taken to be a telecommunications device or other end-point.

The shortest-path problem is to find the path in a weighted graph connecting two given vertices x and y with the property that the sum of the weights of all the edges is minimized over all such paths.

Robert Sedgewick has studied the shortest-path problem in his book"Algorithms in C", Addison-Wesley 1990.

Examples of networks consist of telecommunications networks in the form of a fiber optical capacity in the form of a whole fiber or a part of such, for example in the form of wavelengths, electromagnetic means of transmission in a conductive material, or without one. These networks may use an arbitrary means of transmission and modulation on the conduit, such as frame relay, synchronous or asynchronous transmission or Internet proto- col packets.

A network calculation may exist for a means of building a network, such as the right of ways for creation of networks, cable conduits, fiber with- out transmission devices, so called dark fiber, wavelength transmission mechanisms without end devices, so called dark wavelengths or colors.

A "constraint" is a property associated or attached to a link, which is used in the calculations, such as the calculation of desirable paths through the network. A constraint is additive, if it makes sense to add the constraints along a path to form a new constraint for the whole path.

Typical additive constraints are the price of the sequence of links traversed, latency of delivery of packets, the total transmission delay and the duration of transmission of a certain amount of data to the destination.

The calculations in this document require that they be applied to networks with non-negative path lengths in the cumulative constraint.

If a path cannot be reliably bought due to lack of liquidity at the set price, the path is not replacing a link. Thus, a path may have insufficient liquidity.

A function of cumulative constraint is not critical for the calculation, for reasons of simplicity a function of addition is assumed to be in use.

In the rest of this document the term"price"is used as the con- straint, but it is trivial to use some constraint than price.

The description in this document is for a network in which the con- straints are symmetric, i. e. the same value holds in both directions along a link. This is done for clarity of expression. For a person skilled in the art the operation in an asymmetric case is trivial.

A"bound"is a limit for a value. In our case the bounds considered are both the upper and lower bounds, or range, of the price that may be set for a link.

Both upper and lower bounds can be extended to cases, where sets of links forming a path are replaced, where thus link A and/or link B is a set of links.

A link's"irreplaceable range"is between the upper and lower bounds of a link.

The upper bound of the irreplaceable range for a link's additive constraint is formed by the sum total of the constraint for the cheapest set of links that in conjunction provides connectivity between the original link's end points. In addition to connectivity the other beneficial properties of the link, such as bandwidth and likelihood of failure must be met. Pricing the link below the upper bound of the price ensures that the link cannot be replaced by a set of links, cheaper in total price.

The lower bound concept is described on the following.

Link is assumed worthless unless it forms in whole or as a part a replacement for another link when searching for the lower bound for the additive constraint (price of a link). The search is characterized as the finding of a link for which this link is the most valuable replacement.

On the following the links are explained in more detail : Link A is the link, which is assumed worthless unless forming a part of the cheapest replacement for an- other link.

Link B is the link to be replaced by A and the other links required to form connectivity.

Links A'.. A" are the links needed to form connectivity of B with A as a component of the replacing set.

If A is at its highest value, when providing a part of the connectivity of B, the lower bound for the constraint of link A is constraint of B subtracted by the sum total of constraints of the links A'.. A".

Summary of the invention The novelty in the invention is to determine both an upper and a lower bound for the price of any link or all links of a network. When the upper bound and the lower bound are determined also the safe pricing range is defined. Said bounds are determined by utilizing shortest path algorithms in which the prices of the links are used as weights.

The invention includes three calculation methods based on the shortest path algorithms. The first and the second calculation method are intended to find the upper bound and the third calculation method is intended to find the lower bound. The results of calculations can be improved by taking into account the liquidity, which is associated with the availability of each of the links.

The preferred embodiment of the invention is a system including a unit for calculation of the upper bound and another unit for calculation of the lower bound. The database system gives the user access to the current safe range of pricing. The data in the software consists of bids and offers of tele- communications capacity retrieved from web based bandwidth trading mar- ketplaces, other sources of telecom pricing information and the user's local sources of prices of telecom capacity. Tariffs of capacity prices may also be used as a source of information.

The benefits to the user include the ability of pricing effortlessly their own products at a competitive level and being able to estimate the value of other acquisition and if needed purchasing propositions at a rapid pace.

Brief description of the drawings The invention is described more closely with reference to the ac- companying drawings, in which Figure 1 shows a flow chart of non-replaceability bound calculation.

Figure 2 shows an example of calculating non-replaceability bound.

Figure 3 shows a flow chart of non-replaceability bound calculation without regard to liquidity.

Figure 4 shows an example of calculating non-replaceability bound without regard to liquidity.

Figure 5 shows a flow chart of supplanting bound calculation.

Figure 6 shows an example of calculating supplanting bound.

Figure 7 shows a system in accordance with a preferred embodiment.

Figure 8 shows a flow chart describing the operation of the system in ac- cordance with the preferred embodiment.

Detailed description of the invention The invention in question is used to define the safe pricing range for a link of telecommunications capacity in a telecommunications network.

Pricing may be done outside the safe pricing range, for example, when the quality of the product differs substantially from the quality used to determine the safe pricing range.

Next we describe three calculation methods related to the present invention. The first and the second calculation methods are both termed "non-replaceability calculation"and the third one is termed"supplanting bound calculation".

Then we describe the preferred embodiment of the invention which is a system giving the user access to the current safe range of pricing.

This system uses said calculation methods.

Lastly we describe alternative embodiments of the inventions. In certain conditions these embodiments may be refinements of the invention.

Some simplifications have been made in this description, but to one ordinary in the skills in the art the modifications and additions described should be readily apparent in nature and in implementation.

Non-replaceability bound calculation A"non-replaceability bound"forms an upper bound for the price of the link. The idea behind this calculation is that the elements in the network can replace the other elements in the network, if combined in a manner that provides connectivity between the endpoints of the element to be replaced.

Non-replaceability can be defined as pricing a link X (A, B) in a manner, where the link cannot be replaced by a cheaper set of links provid- ing connectivity between the endpoints A and B. Non-replaceability is a de- sired form of operation, if there is demand for both the link X (A, B) and a replacing set of links between A and B.

Information on the other prices of links in the network can be used in multiple ways. The simples is to find the shortest path connecting the end- points. It is more difficult to only account the paths with sufficient liquidity. If a path cannot be reliable bought due to lack of liquidity at the set price, it is not truly a replacing one.

FIG. 1 shows the non-replaceability bound calculation. The calcu- lation results in the upper bound for the price of the link i. e. the non- replaceability bound.

FIG. 2 shows an example of calculating non-replaceability bound.

In the example the non-replaceability bound is calculated for the price of the link between B and D.

In the following description we refer to both the figures so that -FIG. 1 is referred by numbers 101 to 110 and -FIG. 2 is referred by numbers 201 to 207.

The non-replaceability bound calculation comprises the following steps: 101) Collect the properties of all links.

102) Transform results of collection to a topological network with a single metric of the additive property in question for each link (201 shows the network).

103) Repeat steps 104-109 for each unprocessed link x (202 shows a selection of link x).

104) Assume price of zero for x (203 shows how the real price is disregarded).

105) Make matrix A of all shortest paths in the network (204 shows the matrix A at this point for the network of 203).

106) Assume price of infinity for x (205 shows the assumed price).

107) Make matrix 8 of all shortest paths (206 depicts the matrix B in this case).

108) Subtract A from B to get matrix C (matrix C shown in 207).

109) Find the non-replaceability bound from matrix C by execut- ing steps 109a-109d (each positive element in matrix C denotes that the link is along the preferred path for the as-

sociated end points as long as it is priced at below the value of the element). a) Create an empty set Q. b) Add elements with the biggest positive value to set Q and remove them from matrix C. These elements are termed"the highest elements", or the elements are termed"the highest elements with liquidity"if they have sufficient liquidity (in matrix C the highest elements are circled; see 207). c) Check if the liquidity of set Q is at or above desired value by evaluating the combined liquidity of the elements added to set Q. If the liquidity of set Q is at or above desired value, the lowest element of set Q is the result i. e. the non-replaceability bound.

Otherwise repeat from step 109b. d) Set price to"the highest elements","the highest ele- ments with liquidity"and"the lowest element"with the additional constraint of a maximum hop count for the shortest path in B. When finding the shortest path the hop count is required to be below the maximum hop count.

110) End the calculation.

Thus, in the above example the non-replaceability bound is calcu- lated for the price of the link between B and D. In this calculation the highest elements of set Q are two 4-values (see 207). Therefore the lowest element of set Q is also 4 i. e. the non-replaceability bound is 4.

In step 109c the liquidity checking can be disregarded by setting the liquidity of set Q to 0.

In step 109d the hop count consideration can be disregarded by setting the hop count constraint to oo.

In the following we describe another non-replaceability bound cal- culation.

FIG. 3 shows the non-replaceability bound calculation which dis- regards liquidity. The calculation results in the upper bound for the price of the link. This upper bound is termed"the maximum non-replaceability bound".

FIG. 4 shows an example of calculating non-replaceability bound without regard to the liquidity. Also in this example the non-replaceability bound is calculated for the price of the link between B and D.

In the following description we refer to both the figures so that - FIG. 3 is referred by numbers 301 to 307 and - FIG. 4 is referred by numbers 401 to 404.

The non-replaceability bound calculation, which disregards the li- quidity, comprises the following steps: 301) Collect the properties of all links.

302) Transform results of collection to a topological network with a single metric of the additive property in question for each link (401 shows a sample network).

303) Repeat steps 304-306 for each unprocessed link x (402 shows an example of the selection of x) 304) Assume price of infinity for x (403 shows the disregard of the real price).

305) Compute the shortest path between the points of x (404 shows the shortest paths).

306) Obtain the shortest path which is the maximum non- replaceability bound for link x (the result is circled in 404).

307) End the calculation.

Thus, in the above example the maximum non-replaceability bound is calculated for the price of the link between B and D and its value is 4 (see 404).

Supplanting bound calculation The maximum price for a link in a path, which is termed"supplant- ing path", forms a lower bound for the price of the link. This lower bound is termed"supplanting bound".

A supplanting path can be defined as pricing a link X (A, B) in a manner, where it can be used to supplant an other link Z (C, D), in combina- tion with sets (potentially empty) I (C, A) and J (B, D) of links at or below the price of Z. If I is empty, C = A, if J is empty, B = D. The supplanting path for Z is thus the set {I, X, J}.

To find the supplanting bound we have to find the shortest sup- planting path for a single link.

FIG. 5 shows the supplanting bound calculation. The calculation results in the lower bound for the price of the link i. e. the supplanting bound.

FIG. 6 shows an example of calculating the supplanting bound. In the example the supplanting bound is calculated for the price of the link be- tween B and D.

In the following description we refer to both the figures so that -FIG. 5 is referred by numbers 501 to 510 and -FIG. 6 is referred by numbers 601 to 607.

The supplanting bound calculation comprises the following steps: 501) Collect the properties of all links.

502) Transform results of collection to a topological network with a single metric of the additive property in question for each link (an example of a network is shown in 601).

503) Create the connectivity matrix D, with each element being the price of the link between the end points, if such exists, or 0 if no link between endpoints exists (matrix D is depicted in 602).

504) Repeat steps 504-509 for each unprocessed link x (an example of selection of link x is shown in 603).

505) Assume price of zero for link x (see 604).

506) Set element corresponding to x to 0 in matrix D (605 shows the masking of x in matrix D) 507) Make matrix A of all shortest paths in the network (606 shows the shortest path matrix A) 508) Subtract matrix A from matrix D to get matrix E (607 shows the result matrix E)

509) Select the non-replaceability bound from matrix E. The non- replaceability bound is selected similarly as described above and shown in 109 in FIG. 1. Set all elements of E that do not have x in the shortest path to 0. Set price to highest element in E (The bound is circled in the example of 607). Set price to the highest element in E with sufficient li- quidity. Set price to the lowest element in E contained in the set that has sufficient liquidity. Set additional constraint of a maximum hop count for the shortest path in A. When finding the shortest path the hop count is required to be below the maximum hop count.

510) End the calculation.

Preferred embodiment The preferred embodiment of the invention is a system giving the user access to the current safe range of pricing. The data in the software consists of bids and offers of telecommunications capacity retrieved from web based bandwidth trading marketplaces, other sources of telecom pricing information and the user's local sources of prices of telecom capacity. Tariffs of capacity prices may also be used as a source of information.

FIG. 7 shows the system in accordance with the preferred em- bodiment. The system comprises the following means: 701) Information source (one or more) contains information of the constraints and the topology of the network.

702) Data gathering unit gathers information on the topology and the constraints from one or more information sources.

703) Data store includes means of storing information on con- straints and topology.

704) Upper bound pricer performs the pricing on the upper bound.

705) Lower bound pricer performs the pricing on the lower bound.

706) Upper price bound data store includes means for storing re- sults of the calculation.

707) Lower price bound data store includes means for storing re- sults of the calculation.

708) Data preparation is a mechanism for transforming and re- trieving data to a displayable form from the data stores.

709) The data display device or system.

FIG. 8 shows a flow chart describing the operation of the system in accordance with the preferred embodiment. The operation of the system comprises the following steps: 801) Start operation of the system to find the current safe range of pricing.

802) Data gathering unit (shown in 702) contacts an information source (shown in 701).

803) Data gathering unit gets link price data from the information source.

804) Link data is entered in the data store.

805) If there are more link items in the source step 803 is re- taken.

806) Data from data store is analyzed by the upper bound pricer 807) The upper bound prices are entered in the upper price bound data store.

808) Data from data store is analyzed by the lower bound pricer 809) The upper bound prices are entered in the lower price bound data store.

810) Data from upper and/or lower bound data stores is prepared for presentation.

811) Data is presented on any media such as, display terminal, paper report etc.

812) End operation.

The preferred embodiment can be simplified and/or adjusted by removing one or more means of the system shown in 706,707 and 708; or by removing one or more steps of the flow chart shown in 807,809 and 810.

However, these removals disable statistical handling of the gathered data and/or presentation of data.

Alternatively the means of shown in 708 and 709 and the associ- ated steps of the flow chart shown in 610 and 611 can be removed to create a system that generates data but does not display it.

Additional embodiments There are multiple additional embodiments of the invention which are not required for functionality, but which may be refinements of the inven- tion in certain conditions. The additional embodiments can be divided into the following groups: 1. Embodiments concerning utilization of a topological network.

2. Embodiments concerning utilization of an adjacency matrix.

3. Embodiments concerning use of additional constraints.

4. Embodiments concerning use of calculation results in addi- tional calculations.

The embodiments of each group are described starting with the first group.

In the first group a physical telecommunications infrastructure is modelled as a topological network. The trivial means of doing this, mapping each node to a node in the graph representing the topological network and each link to a link in the graph can be extended in multiple ways.

The embodiments of the first group are the following.

1.1 The system is adapted to combine or split products from one link type to another. The result set will be of higher quality the more links can be had between the nodes. The demand for a cer- tain capacity between points at a certain bandwidth can be deliv- ered using a link of same capacity and different framing type, if both of the ends of the link can convert from one to another. In cases, where both of the ends of the link can convert, the link may be added to the graph.

In a more general case a whole sub-network capable of con- version at its edges may be mapped into the graph as a link, or a set of links.

1.2 The system is adapted to combine or split products from one type to another using tables, curves and/or formulas. In some cases the transformation of a link, or a sequence of links or a sub- network requires a mechanism of transformation to obtain the bandwidth and latency after the transformation to the desired fram- ing or other mechanism.

Such a transform may be for example a table, curve or a function. A transformation may use the analytic model of the trans- mission control protocol (TCP) to model a sequence of links as a single link so that the links are equipped with different latency and bandwidth properties.

1.3 The system is adapted to acquire products of higher link capacity potentially leaving some of this capacity unused. The full set of links with sufficient bandwidth or above can be mapped into the topology, as all of the superior bandwidth links replace a lower bandwidth link perfectly. In some instances it may be beneficial to force at least one of the links on any and all paths to be of the de- sired capacity, as in some instances the superior capacity may not be what is truly required.

1.4 The system is adapted to acquire products of higher ca- pacity and attempt to benefit from the excess capacity. One can also add the superior capacity products into the graph, and in pric- ing them, make assumptions on the resale ability or other means of benefiting together with the cost and effort involved in benefiting of the excess bandwidth.

1.5 The system is adapted to acquire less capacity that joined together forms higher capacity. For some uses one may use multiple lower capacity links in tandem to provide higher ca- pacity. The links may span the same nodes, and be parallel, or may span the same end-or mid-points and form paths of differing sets of points.

1.6 The system is adapted to use multiple offers back to back to span the duration. The desire for bandwidth contains as its pa- rameters a start time, an end time, endpoints, and the technical characteristics of the desired conduit. The spanning in time can be done by a single offer for a link at a given time, or multiple offers spanning the desired duration. The path through the network may even change during the lifetime of a desired capacity through the network.

Change the topology by selling, swapping, leasing, trading or otherwise divesting network assets temporarily or permanently current assets.

The changes due to the customer demands may be perma- nent in nature. A link or set of links can be sold, if it is not used at all, or if it can be inexpensively replaced.

1.7 The system is adapted to change the topology by acquir- ing assets suitable for the current traffic and the new traffic. The change may be also of the form of acquiring network assets, as- suming that that is a less expensive way to add capacity into the network compared to the continuous leasing of capacity.

1.8 The system is adapted to map a sub-network, virtual cir- cuit or other combination of network assets or subset of network and link assets into a link. The network assets other than links can be mapped into a link like construct enabling computing their worth or utility as described in this invention.

In the second group a physical telecommunications infrastructure is modelled as an adjacency matrix.

The embodiments of the second group are the following.

2.1 The system is adapted to use the trivial existing products filled as connectivity matrix as the point of comparison. In addition, the system is adapted to list all links with desired capacity or above, assuming that the link is either fully used up, or available.

2.2 The system is adapted to add only links with sufficient remaining capacity. Alternatively for each link the utilization rate can be a known fraction or a stochastic quantity describing the utilization of the link. In this case all links with sufficient remaining capacity are added into the matrix.

In the third group the additional constraints can be utilized to fulfil more complex demands, for example to prefer or select a path and its properties.

The embodiments of the third group are the following.

3.1 The system is adapted to prefer or select non-wanted nodes. Some of the nodes in the network may be considered un- wanted and they either - are avoided, when an alternative path exists, - are avoided even when no alternative path exists or - are avoided, except when connecting to the said node.

3.2 The system is adapted to prefer or select non-wanted links. Some of the links in the network may be unwanted, and they are to be avoided either in all cases, or only if no alternative path exists.

3.3 The system is adapted to prefer or select link preference.

There may be different preferences on the link usage. These may be on technical grounds or other grounds, such as the ownership of the link. If preferences exist, the preferred links are to be used in a fashion, where the preference is mapped as a constraint low- ering multiplier.

3.4 The system is adapted to prefer or select node prefer- ence. Nodal preferences are to be implemented as link prefer- ences are. The trivial desired result is that of sufficient capacity, for duration, between the endpoints, satisfying the other technical constraints. Other results, such as aiming for superior quality or fit-

ting other desired characteristics, can be obtained with small modi- fications to the running of the methods.

In the fourth group the result paths are completed with additional constraints relating to the transmission latency or the duration of the trans- mission, wherein the result paths are obtained from the non-replaceability bound calculation or the supplanting bound calculation.

The embodiments of the fourth group are the following.

4.1 The system is adapted to reduce the transmission la- tency. The transmission latency can be reduced by purchasing ex- tra bandwidth for one or more links which belong to the result path.

By using a suitable formula for the transmission, one can see the effects of the extra bandwidth.

4.2 The system is adapted to meet certain latency bounds. In this embodiment it is assumed the result paths are calculated and the latency bounds of each result path are set. The latency bounds may concern the whole result paths or one or more links which be- long to a certain result path. The result paths are completed with the values of links belonging to a certain result path. A value of the link may present link's ability to reduce the latency between two nodes. During the new calculation the result paths not meeting the certain latency bounds are rejected. Beneficial use cases of this embodiment relate to, for example, the TCP or voice over Internet peripheral (VolP).

4.3 The system is adapted to meet certain transmission dura- tion bounds. In some cases the most important requirement may be meeting the transmission duration bounds. If there are many paths meeting said bounds, the cheapest path of them is selected.