Subscriber call routing process system.

Authorization

0001A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.

Field of the Invention

0002This invention relates to the field of telephony and, more particularly, to a system where calls are routed using intelligence resident outside of the public switched telephone network.

Background of the Invention

0003A telephone system, in its most basic form, comprises one individual dialing a telephone number assigned to another individual. With such a system, there is a one-to-one relationship between the dialed number and the destination. The telephone system is also designed to provide special calling services using inbound N00 telephone numbers. These numbers currently utilize the area codes 800 or 900. Particularly, businesses arrange for the use of a particular 800 number to allow customers, at no cost, to call and transact business. Conversely, with a 900 number, the customer is charged for the service.

0004During the early 1980's, the long distance carriers removed 800 number translation and routing from switching systems and instead placed it in a call routing database, referred to as a "service control point" (SCP). This architectural implementation allowed a single nationwide N00 number to be routed to any one of a plurality of subscriber destinations. For example, a caller may wish to place an order with a company, inquire about the status of an account, or obtain information about the company's products or services. The SCP must be configured to make the determination as to which of the destinations should receive a particular call, i.e. where the call should be routed. The general type of subscriber service desired may be selected by dialing a particular number, with different numbers corresponding to different services. Alternatively, the subscriber may have a single number and utilize prompting whereby additional digits are entered by the caller to specify the particular service desired. All of this information is used by the SCP to make the routing decision.

0005In response to a call routing inquiry from a carrier switch node, the SCP makes the routing decision based on some combination of specified criteria, namely, originating area code and/or exchange, time of day, day of week, date, and any caller entered digits. These call characteristics are compared to call screening criteria resident in the SCP in the form of a "routing tree" to determine the destination to which the call should be routed. The call routing response results in the call either being routed to a specific location or allocated among a set of locations using preselected percentages. The subscriber may change the call screening procedure and call allocation percentages by inputting these changes to the SCP in the form of a new or revised routing tree. These changes become effective sometime after they are made depending upon the particular system used. These changes can be used to cope with unexpected variations in customer calling volume, the lack of available agents at one or more subscriber destinations and other factors. However, this is a rather slow response to a dynamically changing system.

0006In managing how calls are routed among several alternate subscriber destinations, caller identity, the carrier network, the SCP, and subscribers resources are all used. The subscriber strives to satisfy various performance criteria. One of extreme importance is how long it takes for a call to be answered by an agent. If the time is too long, then the perception of quality drops. If callers are held in queue, then carrier charges may increase. Finally, an abandoned call can adversely effect revenue in a sales application if a customer does not call back.

0007One solution is to provide a sufficient number of agents to handle the maximum number of calls expected. However, this is an economically impractical solution. An optimal solution is to utilize a minimum number of agents while routing the calls to satisfy specified performance objectives.

0008The present invention is intended to solve one or more of the problems discussed above in a novel and simple manner.

Summary of the Invention

0009In accordance with the invention, there is disclosed a subscriber call routing processing system which makes routing decisions outside of the public switch telephone network.

0010It is one object of the invention to provide call routing decisions based on specific identity of the caller. A caller ID, comprising all ten digits of the phone where the call is originating, can be stored in a database along with information about the caller relevant to the business relationship the caller has with the subscriber. Along with influencing call routing instructions, some of this information can be provided to the carrier network for delivery to the destination which receives the call. This information can be used by the resource to assist in handling the customer.

0011Additionally, this feature allows information collected and stored about a particular customer to influence a call routing decision. The customer database, for instance, can be configured to keep track of how many calls are made from a particular customer during given periods of time. This information can be used to route the call in a specific way, such as by transferring calls to different resources depending on the number made.

0012In accordance with another aspect of the invention, calls can be routed based on the availability of agents or other resources for handling the calls. Calls can be load balanced across multiple subscriber resources at the same or different locations to minimize queuing and maximize resource productivity. Based on screening performed on call characteristics and caller attributes, several subscriber resources may be able to handle the call. Additional criteria provided by the subscriber specifies how to choose from among these alternate resources. This criteria includes precedence level information used to rank the alternate resources and performance information used to determine the maximum time callers are allowed to wait in queue for each resource at each destination.

0013In accordance with yet another aspect of the invention, subscriber provided call routing allows the status of network provided facilities to influence a call routing process. An operational support system in the carrier's network provides alarm data about the switches and facilities used to connect customer calls to the subscriber's resources. Depending on the availability of the carrier's switches and facilities, some of the subscriber's resources may not be able to receive all of the calls they are capable of handling. Calls routed to a subscriber resource in excess of the network's ability to handle calls results in the caller experiencing a fast busy tone or intercept announcement. Network availability information is used in the load balance process to provide an improved call routing decision.

0014In accordance with a further aspect of the invention, the call screening criteria and routing instructions can be changed dynamically. The manner in which decision algorithms are organized permits changes to be made to dynamic variables within a matter of seconds. Also, a subscriber is able to respond to problems either in the public switched telephone network or subscriber network very quickly and minimize the impact on lost sales and customer service.

0015In accordance with still another aspect of the invention, subscriber provided call routing allows many different dialed numbers that share the same destinations to be routed by the same routing rule. By doing so, any dynamic changes made by the subscriber become effective immediately and simultaneously for all dialed numbers routed by the same routing rule. Thus, load balancing is able to incorporate all of the calling demand being placed on the resources, and simultaneously load balance calls for all of the dialed numbers.

0016In accordance with yet a further aspect of the invention, subscriber provided call routing allows call routing queries to originate in more than one carrier network for the same routing rule. The different dialed numbers that share the same resources may be transported over different carrier networks. Calls for the same dialed number may also be transported over more than one carrier network and delivered to the same subscriber resource. Each carrier's own routing tree can only make routing decisions that are independent of those made in other carrier networks. These routing decisions cannot provide a system offering the benefits of subscriber provided call routing using routing intelligence resident in a single point outside of the carrier's network to provide an effective dynamic updating and load balancing system involving multiple carrier networks.

0017More particularly, there is disclosed herein an advanced intelligent network comprising a routing processor, activity monitor, event processor, and caller database. The routing processor uses algorithms, in the form of routing rules that select the business user's preferences, for delivering N00 calls to a proper destination. The event processor receives status information on resources and relays that information to the activity monitor for both route processing and call center operation monitoring. The activity monitor controls the flow of information and uses that information in conjunction with routing algorithms to determine the variable information within the routing rules used by the routing processor that selects the final destination. The caller database provides information about the caller to choose the preferred destination for the call. Such information also may be sent to the destination to assist the agent in handling the call.

0018The process flows through the subscriber network as by a call inquiry exiting the telephone network SCP looking for a destination. The routing processor fills in the destination and returns a call routing response based on the routing rule initially created by the subscriber and dynamically updated by the activity monitor. The call is routed, via the public switched telephone network, to the desired resource and status information is passed back through the event processor to the activity monitor to assist in subsequent routing decisions and to provide agent statistics information.

0019Further features and advantages of the invention will be readily apparent from the specification and from the drawing.

Brief Description of the Drawing

0020<ul id="ul0001" list-style="none"><li>Fig. 1 is a diagrammatic/block diagram illustrating basic elements in the public switched telephone network and the subscriber network in accordance with the invention;</li><li>Fig. 2 is a more detailed block diagram illustrating the elements of the advanced intelligent network processor of Fig. 1 and its relationship with the public switched telephone network;</li><li>Fig. 3 is a table illustrating the general configuration for a routing rule;</li><li>Fig. 4 is a flow diagram illustrating operation of the routing processor of Fig. 2;</li><li>Fig. 5 is a flow diagram illustrating operation of the routing rule task of the flow diagram of Fig. 4;</li><li>Fig. 6 is a flow diagram illustrating operation of a program for the load balancing algorithm implemented in the activity monitor.</li></ul>

Detailed Description of the Drawing

0021With reference to Fig. 1, a diagrammatic/block diagram illustrates a subscriber call routing processing system used in accordance with the invention. The system is designed to allow a subscriber the ability to control routing of inbound N00 (currently area codes 800 or 900) telephone calls to a desired destination or resource based on the subscriber supplied routing criteria. This embodiment of the invention uses area codes 800 and 900 but can easily be extended to support any inbound service and any numbering plan.

0022The system utilizes the facilities of the public switched telephone network (PSTN) 100 provided by one or more network carriers. These carriers include, for example, Sprint, MCI, AT&amp;T and any local exchange carrier. A caller originates a call at a telephone 102 by dialing a number assigned to the particular subscriber. The call is connected to one of many carrier switch nodes 104 contained in the PSTN 100. The particular carrier switch node 104 is generally that in closest proximity to the caller 102. The carrier switch node 104 temporarily suspends the call setup process owing to the receipt of an N00 number and launches a routing query to the carrier's call routing database or service control point (SCP) 106. Based on the number dialed, a call translation process is invoked in the SCP 106 using routing trees configured for the subscriber. Within a subscriber network 112, there are included a plurality of resources, identified as Resources A, B and C, 108, 109 and 110, respectively, to which the calls should ultimately be routed. Conventionally the routing decision is made by the routing tree at the SCP 106. The SCP 106 controls switching in the carrier switch nodes 104 to connect the caller 102 to one of the resources 108-110. One example of such a conventional system is disclosed in Frauenthal et al., U.S. Patent No. 4,737,983.

0023In accordance with the invention, the subscriber network 112 also includes an advanced intelligent network (AIM) processor 114, a user terminal 116 and a customer database 118 for making routing decisions outside of the PSTN 100. The SCP 106 is modified so that instead of directly implementing decisions from its own routing tree, it launches a routing query to the AIN processor 114 in the subscriber network 112. Based on the information provided by the carrier within the routing query, information about the caller contained in the subscriber's customer database 118, information provided by the carrier's operation support system 120 about the status of the facilities, and information about the status of the resources 108-110 which may handle the call, the AIN processor 114 returns a resource identifier to the SCP 106. This response may also include data to be delivered to the resource by the PSTN 100. The SCP 106 receives the routing response from the AIN processor 114 and instructs the carrier switch node 104 to resume the call setup process and complete the call through the PSTN 100 to the designated subscriber resource 108, 109 or 110.

0024Routing queries from different callers for the same dialed number may simultaneously originate in more than one carrier network, based on the needs of the subscriber and the capabilities of each carrier. The subscriber interfaces with the AIN processor 114 through the user terminal 116.

0025With reference to Fig. 2, the components of the AIN processor 114 are illustrated in greater detail. Particularly, the AIN processor 114 consists of a Call Management Language (CAMEL) editor 130, a routing processor 132, an activity monitor 134, an event processor 136 and the end user terminal 116. The CAMEL editor 130 allows the subscriber to create routing rules, stored in a DASD 138, which contain a majority of the routing decisions that affect the delivery of a call to the subscriber's resources. The routing processor 132 uses the completed routing rules to deliver N00 calls to the proper resource. The event processor 136 receives information from the resources, such as ACD's, 108-110, to be used for both route processing and call center operations monitoring. The activity monitor 134 controls the flow of information in the AIN processor 114 and uses that information in conjunction with the routing rule structures to determine the variable information within the routing rules used by the routing processor 132 to select the final destination.

0026The plot process through the advanced intelligent network processor 114 begins when a call routing query exits the PSTN SCP 106 looking for a destination. The routing processor 132 fills in the destination using the routing rules and returns a call routing response based on the routing rule initially created by the subscriber and dynamically updated by the activity monitor 134. The call is routed, via the PSTN, to the desired ACD 108-110 and status information is passed back through the event processor 136 to the activity monitor 134 to assist in the next routing decision and to provide agents statistics information.

0027In accordance with the invention, the CAMEL editor 130, routing processor 132, activity monitor 134 and event processor 136 may comprise separate processing systems connected via a suitable network, or may be part of a single processing system. Advantageously, the functions are implemented in a Customer Information Control System (CICS). A CICS is a program operating in a processing system enabling transactions entered at remote terminals to be processed concurrently by subscriber application programs. The various functions consist of either regions within the system, or tasks forming part of a particular region.

0028The SCP 106 is part of the public switched telephone network 100 and is not itself part of the invention. Nevertheless, the subscriber network 112 must communicate with the SCP 106 to permit routing decisions to be made externally of the PSTN 100. A new node type similar to existing allocation, time and NPA nodes is defined for the N00 system in the SCP 106. This node allows the subscriber to interact with the SCP via a communication interface to make real time routing decisions based on subscriber defined algorithms described in the subscriber network 112. If a valid response is not obtained on the communication interface, then a default decision is made by the routing tree in the SCP 106.

0029Within the SCP 106 the new node type is processed exactly as the existing time, NPA and allocation nodes until it is accessed for call processing. When accessed for call processing in the SCP 106 the line identification field in the node is compared with an SCP configuration database for the selection of the session and physical communication line to be used. If that session is in accessible status for an open with no unresolved errors, then the SCP 106 transmits an inquiry message on that session. If a response to the inquiry is received within a short timeout (nominally 500 milliseconds) which matches a selection on the node, then processing continues with that selection. Otherwise, processing continues with the default selection defined within the SCP 106.

0030In order to assure rapid response, high speed point-to-point lines are used. Inquiries are transmitted as calls are processed without waiting for a response to previous inquiries. Volumes of up to forty inquiries per second are supported in a single session. Thus, up to twenty inquiries may be outstanding at any one time. Application timeout is measured from transmission of the inquiry to receipt of response to that inquiry. Both sides of the interface perform automatic error recovery when possible consistent with minimizing response time and code complexity. In order to minimize the impact on call processing performance, the SCP 106 maintains its status indication on each session and only attempts to send inquiries on a session with verified status.

0031The SCP 106 generates an inquiry message when a communication node is encountered in processing a call. Each inquiry message will contain: <ul id="ul0002" list-style="none"><li>1. Inquiry ID;</li><li>2. NPA-NXX-XXXX (or as many digits as available);</li><li>3. N00 Number;</li><li>4. Node Identifier;</li><li>5. Caller information (such as class of service); and</li><li>6. Caller entered digits.</li></ul>

0032The subscriber network 112 generates a response message in response to an inquiry. The inquiry ID will be from the inquiry message received. Each response will contain: <ul id="ul0003" list-style="none"><li>1. Inquiry ID;</li><li>2. Select Code;</li><li>3. Destination tag; and</li><li>4. Caller information.</li></ul>

0033When the session is established, initialization messages are exchanged identifying the communication node processed by node identifier.

0034"NPA-NXX-XXXX" designates the area code, exchange and station of the caller. "N00 number" identifies the number dialed by the caller. "Mode identifier" specifies a character string identifying a particular usage of the communication feature known to both the SCP 106 and the AIN processor 114. The "Select code" and "Destination tag" identify a particular routing choice. These are also known to both the SCP 106 and the AIN processor 114.

Activity Monitor

0035The activity monitor 134 is the central control point for all subscriber network processing. This processing includes: polling of status information from resources; reporting of agents status information; implementation of user-defined routing rules; use of algorithms to create tables necessary to provide load balancing and other features; facilities to accept manual input from an end user terminal to effect routing of calls; facility to report and respond to an outage in the subscriber network 112; facility to accept routing statistics from the routing processor 132; facility to obtain caller specific information from the caller database 118 and to collect and store information about the caller; and facility to log the routing statistics for trend analysis, billing and agent planning.

0036The subscriber resources can include automatic call distributors (ACD's), audio response units (ARU's) and other intelligent peripheral (IP) devices, agents working out of their own homes, network message announcement devices, or any other PSTN termination. The activity monitor 134 uses status information from these resources, application programs and subscriber input to calculate capacity percentages used to route calls to available resources over the next thirty second period of time. This information is also used in the reporting of statistics or resource availability for the previous thirty second period of time.

0037The ACD resource may comprise any commercially available ACD. In the embodiment illustrated and described herein, the system is used in connection with a Rockwell Galaxy ACD. The polled Rockwell ACD information format, specifically known as Call Side Real Time Gate and Overflow records, is found in a Rockwell International manual titled "Galaxy ACD, External Record Description - Galaxy 7", pages 1-5. Also, information on the RMC foreign processor data link request format can be found in the Rockwell International manual titled "Galaxy ACD RMC Foreign Processor Data Link", pages 1-13. The specific reporting fields provided are agents available, number of abandoned calls, average delay in queue, average speed of answer, primary agent call work time, cumulative calls offered, number of calls held, number of calls offered, number of out calls, percent calls held, primary positions manned, secondary positions manned and service level.

0038While the processing of routing rule creation and processing is initiated by the subscriber through the CAMEL editor 130, the implementation of these routing rules involves messages between the activity monitor 134 and the affected routing processor 132. For new rules, this involves sending a message to the routing processor 132 to a terminal owning region (TOR) to retrieve a routing rule from the file. The TOR forwards the message to the appropriate application owning region (AOR) for processing. The routing processor's AOR region accesses the routing rule from the rule object file and places the rule in the AOR's rule data table. The routing processor notifies the activity monitor 134 through the TOR that the rule has been installed in the AOR's data table. Upon receipt of all notifications from all routing processors 132, the activity monitor 134 instructs the routing processor 132 to update their active rule tables in memory and on any DASD's 138. For scheduled requests, the activity monitor 134 coordinates the sending of the initial retrieve message. Included are messages to include new rules from each routing processor 132. For existing rules, the same procedures as above apply. Existing rules are retrieved from the common rule file at the activity monitor 134.

0039The routing algorithm uses call volume information from the routing processors 132 and resource status information from ACD's, ARU's, or other customer premise equipment and agents at home to create proper call routing destination tables to be in effect for the next thirty second period of time. The activity monitor 134, every thirty seconds, initiates the request for information from the above mentioned resources via the event processor 136 and waits a period of time to receive responses to that request. All information received is time stamped to indicate its age. If any resource does not respond within two cycles, then that resource is flagged as inactive and the resource is placed into manual operation mode. It remains in manual operation mode until there is a successful poll. Under normal processing, the select process creates tables and sends them to the routing processor 132 for use in call distribution.

0040Upon creation of the routing rules, an outcome table is generated from CAMEL SELECT statements, as described below, which describes destinations and preferences to which the calls can be routed. The routing algorithm uses this information to determine the final outcome of the call. This table resides in the activity monitor 134 and can be updated, if necessary, when a new routing rule is implemented. This depends solely upon if the new rule contains a new routing outcome or uses a previously defined routing outcome.

0041The specific format of the routing rules is discussed in greater detail below relative to the CAMEL editor 130. The use of the routing algorithm for generating the outcome table is also described below after sufficient background has been provided so that it is more readily understandable.

Event Processor

0042The principal activity of the event processor 136 is to collect statistical information from the various resources, namely, ACD's, ARU's, other CPE's, application programs and manual input. Once collected, the information is available for reporting on any user terminal 116.

0043With respect to ACD's, the event processor includes a direct link to read information directly into the activity monitor 134. Indeed, in one implementation, the event processor 136 comprises a task in the region used for the activity monitor 134. The event processor 136 uses an appropriate protocol data link and polls the ACD's every processing cycle to receive the above-discussed information.

0044With ACD's other than the Rockwell Galaxy ACD, a mini-computer can be used at the ACD site connected to a supervisory serial interface of the same to request status information as necessary. The mini-computer is configured to emulate the supervisory terminal and isolates the event processor from the ACD. It also functions as a protocol converter for the ACD supervisory serial interface to the Advanced Intelligent Network protocol.

Call Management Language (CAMEL) Editor

0045The CAMEL editor 130 is used principally for the creation of routing rules. This includes listing routing rules in a CAMEL data set, entry and maintenance of CAMEL data set members, compiling translation of routing rules into codes acceptable for use by routing processors 132 and activity monitors 134, and error identification, presentation and correction of CAMEL statements. The CAMEL editor uses a Call Management Language (thus the name CAMEL). Particularly, the subscriber using the CAMEL language creates routing rules to make routing decisions used for routing calls to a selected resource. A specification of the Call Management Language used is provided in the appendix hereto. Reference may be had to this specification for an understanding as to the details of the language for creating routing rules.

0046The format for a completed routing rule is illustrated in Fig. 3. The routing rule contains all of the information and instructions for responding to the routing inquiry initiated by the SCP 106, see Fig. 1. The routing rule contains three types of information, including system variables 200, rule variables 202 and call routing instructions 204.

0047The system variables 200 include constants, variables and system parameters which are created and maintained automatically by the AIN processor 114, see Fig. 1. These include system dates, system time, system day and other various input/output areas for communicating with other network elements.

0048The rule variables 202 comprise values and parameters specified by the subscriber and used as criteria against which to screen call characteristics and caller attributes, and used to allocate or load balance calls among resources. The rule variables 202 include static variables 206 and dynamic variables 208. The static variables 206 are specified at the time the rule is created and cannot be changed unless a new rule is created to replace an existing one. The static variables 206 consist of a name and a pointer to a list of values. Only the name and pointer address are static. The values in the list are the dynamic variables 208. For example, there is a static variable 206 for each resource to which a call can be routed, the dynamic variables 208 corresponding to the performance criteria that apply to each precedence level.

0049There is a static variable 206 for each call screening criteria, the list values being NPA's, NPA-NXX's, days of the week, dates, time ranges, caller entered digits, etc. There is a static variable 206 for each allocation table. The list contains resources and their corresponding call allocation percentages. There is a static variable 206 for each routing outcome. The list values contain the resources that participate at each precedence level. The dynamic variables 208 can be changed by the subscriber in real time without the need for creating, validating, and activating a new rule. These changes take place within seconds of the time they are requested by the subscriber.

0050The call routing instructions 204 specify the sequence in which the call characteristics and caller attributes are compared to the various call screening criteria. Based on one or more successful matches between the values of the call being routed and the values in the call screening criteria, one of three routing functions is performed. The three types of routing functions available are ROUTE, ALLOCATE and SELECT. The ROUTE function directs the call to a single specified destination. The ALLOCATE function routes the call based on the contents of an allocation table comprising a list of destinations with specific percentages assigned. The SELECT function uses load balancing based on the contents of a routing outcome, precedence level, and performance criteria.

0051A routing outcome comprises a hierarchy of resource groups that define preferred routing to the load balancing process. In so doing, the load balancing algorithm attempts to route as many calls to destinations in the most preferred resource group as possible. If sufficient call handling capacity is not available at a given level of precedence, then calls are routed to the resource group at the next level of precedence. An example of a SELECT function might contain in the first precedence level a single destination that is geographically close to the caller to minimize expenses. The second precedence level could define a short list of destinations whose applications are all serviced out of the same data processing center to minimize application inquiry response times. The third precedence level could define all of the possible destinations to enable the largest degree of agent availability.

0052The following is an example of a typical routing rule:<img file="EP0559979A2_D0001.tif" />

0053In this example, the system variables are not specifically illustrated.

0054The rule variables include as static variables a group of ACD's under the designator TELECAT, and a list of IP's under the designator ARUSITES. A table identified TELENDAY includes an allocation list and associated dynamically updatable allocation percentages. Additionally, a dynamically updatable error message number is defined, as well as dynamically updatable hours for a typical workday.

0055The call routing instructions commence with a BEGIN instruction followed by a logic statement determining if the system time falls within the working hours specified under the designator WORKDAY. If so, then a SELECT statement is used to provide load balancing. The SELECT statement uses the TELECAT resource table at the first precedence level and the ARUSITES resource table at the second precedence level. No third precedence level is specified. The BALANCE parameter indicates the technique used for balancing, while the default ERRMSG indicates the recording to be used as the last choice destination if the call is not otherwise properly routed. If the system time is not within the normal work hours, then the logic proceeds to the ALLOCATE function which allocates based on the allocation percentages present under the TELENDAY table.

0056The allocation percentages used in the ALLOCATE statement can be entered by the subscriber. The capacity percentages used with the SELECT statement cannot be entered. Instead, the client specifies desired performance and the algorithm calculates the capacity percentages every processing cycle.

0057The routing rules are created using the business subscriber terminal 116 and its associated preconfigured displays. In so doing, a subscriber can select dynamic rule modification, routing rule activation, refreshal of routing rules and destination management. Each comprises a hierarchy of displays in which the subscriber enters the necessary information according to the desired function. For example, with a dynamic tree modification display, the subscriber can select a routing rule for modification and then the types of modifications and dynamic variables for which changes are to be made. With an allocation table, the allocation percentages can be modified, provided that the percentages add up to one hundred percent. Time decision points in the routing rule can be modified, as well as dates.

0058A routing rule is activated by selecting the rule and assigning an N00 number to a previously created rule to immediately start use of the same. The rule can be used by multiple N00 numbers.

0059The routing rule refresh function resets all rules to the original format prior to any dynamic rule modifications. An individual rule can be reset by selecting routing rule activation.

0060All of the above display functions relate to maintaining routing rule tables.

0061The destination view management function allows the subscriber to configure and display resources for viewing polled statistics.

0062Also provided is a facility to allow the subscriber to enter in by resource, for example an ACD gate, the performance to be maintained in each of the three available precedence levels. In a resource service level management table, the subscriber designates at each precedence level the number of seconds within which 90% of the calls are to be answered for each resource. If no options are selected, then the default performance target is for ninety percent of all calls to be answered in twenty seconds or less. This information is used in the algorithm for generating call capacity tables to specify a maximum queue length acceptable at each precedence level.

0063The call capacity table is created for each ninetieth percentile performance target. The tables have an entry with an acceptable queue length for each potential number of agents at the gate. Standard Erlang's B-tables are the starting point for the initial tables. Erlang's B formula predicts the performance of an end-server device fed by a single queue whose length is limited. Once the maximum queue length is reached, then further arrivals bypass the queue. Under the illustrated embodiments, pure Erlang tables are not utilized since the system violates the equilibrium assumption of the Erlang equations by intervening every thirty seconds to adjust the rate of traffic to the ACD gates. This indirectly resets the queue length. Because random fluctuations are not allowed to overload a gate, it can maintain high performance with longer queues than the pure equilibrium state equation suggests. Therefore, test results are used from simulation modeling to establish anchor points to extrapolate the Erlang curves.

Routing Processor

0064With reference to Fig. 4, a flow diagram illustrates interrelationship of the tasks implemented in the routing processor CICS region.

0065Communications with a carrier, such as Sprint or AT&amp;T, use send/receive pipes using a generic receive task 401 which calls a carrier-specific message inbound decoding routine. The receive task 401 reads an input from a receive pipeline, determines the appropriate send task on which to return the response, prefix the message with the CICS terminal ID and recall the message decode routine. After decoding is performed, the message is either queued back to the send task 402, queued to the AOR for further processing, or logged, based on the action indicator setting. The inbound message decode routine takes the input format used by the carrier and performs any reformatting to internal AIN formats or response formats as necessary. The message decoder returns an action indicator to inform the main line as to the disposition of the message. Three other processing routines include queue to DTP, queue to send, and error logging. The queue to DTP routine passes a pointer to the message to be queued, and the name of the target remote CICS region.

0066The queue to DTP routine passes a pointer to the message to be queued, and the name of the target remote CICS region. This routine builds a parm list containing the address of the data to be queued, the length of the data, the type of queuing to be performed (in this case, queue to connection group). This parm list is passed to a routine which does the actual queuing. A return code is set in the parm list to indicate the result of the queuing attempt. At the bottom of the infinite loop a delay for one minute is issued before returning to the top to begin reading the queue again. If inquiries are added to the queue during this delay time, the task queues the inquiries to the interpreter, issues a cancel for the delay, thus waking up the sleeping rule interpreter.

0067Communications with a carrier using send/receive pipes are done using a generic send task 402 which calls carrier specific outbound message processing and formatting routines. The send task 402 first reads data from its queue using a retrieve/wait facility. The carrier specific message outbound processing is called at this point to determine further processing for the message. After the processing is complete, it is possible to call an exit function to perform additional processing if necessary. The outbound message formatting routine places the message into the format expected by the carrier to which it will be sent. Lastly, the message is sent to the carrier via a SEND command.

0068Multiple rule tasks 403 may be started and run in parallel, primarily for failure recovery purposes. The main processing routine consists of an infinite loop in which it attempts to read and process an inquiry queue until it is empty, and then wait for one minute before returning to the top of the loop. The main routine accesses a pointer to the next item to be processed.

0069The inquiry queue is initially a temporary storage queue. The pointer is enqueued upon, the inquiry read from the queue, the pointer updated to reflect the next item to be read, then a de-queue is issued to release the pointer. Another task is then queued on the pointer, then the rule task waits for it to become available.

0070The inquiry record contains an 800 number which is checked against the active rule table to determine which routing rule is selected for processing. Each N00 number is associated with a particular rule in this initialization time, as discussed above, based on information from the activity monitor 134. Once the routing rule name has been determined, then the rule, residing on a subscriber maintained data table, is retrieved. Pointers to the rule, the inquiry record, the AIM anchor table, and the IO buffer are passed to the interpreter function. The interpreter function executes the routing rule and builds the outcome record in the IO buffer. The main processing then calls the queue to DTP routine to send the outcome record back to the TOR.

0071With reference to Fig. 5, a flow diagram further illustrates operation of the rule task 403 of Fig. 4. The main routine begins at a decision block 500 which determines if a routing inquiry has been received. If none is received, then no action is taken. Particularly, the routing query is received from the carrier SCP 106, see Fig. 1. If one is received, then a routing rule is selected at a block 502. The N00 Number dialed by the caller, contained in the routing query, is used to select the rule as defined when the rule is activated. The routing rule may already be in the memory of the AIN processor 114 or may have to be loaded into the memory from DASD storage 138. At a decision block 504, the caller identity, if available, is checked against the subscriber's customer database 118, see Fig. 1. If a match is found, then caller attributes contained in the customer database 118 are retrieved at a block 506 for use in routing the call. The routing rule is then implemented at a block 508.

0072The routing rule is implemented at the block 508 by executing the call routing instructions 204, see Fig. 3, for the selected rule. Particularly, the caller attributes and call characteristics are compared to the various screening criteria specified by the subscriber in the call routing instructions. The screening criteria consists of lists of possible values of the caller attributes and call characteristics. Logical comparisons are performed between the actual values of the caller attributes and call characteristics for the particular call being processed and the values contained in the system variable list 200 and rule variable list 202. The format of each list depends on the caller attribute or call characteristics being screened. For example, if the originating area code, or NPA, of the caller is being compared, then the list would contain a sequence of possible NPA values. A successful logical comparison would occur if one of the values in the list equalled the originating NPA. In another example, if the time of day is being compared, then the list would contain two time values. A successful logical comparison occurs if the time of day the call is being made is between the two time values in the list.

0073The call characteristics come partly from the carrier and are contained in the call routing inquiry. These include such items as the NPA-NXX-XXXX of where the call is originating, any caller entered digits, and any information known to the PSTN 100 about the line the call is originating on, such as class of service, terminal type, business group, etc. Other call characteristics come from the AIN processor 114 and include such items as the current date, time and day of week, and customer business information. Based on successful matches between the call information and the screening criteria, one of three routing functions is chosen in the routing rule. The three routing functions are a ROUTE function made in decision block 510, an ALLOCATE function in a decision block 512, and a SELECT function in a decision block 514.

0074If a ROUTE function is to be performed, as determined at the block 510, then the outcome record contains an internal destination identifier at a block 516. This is then resolved into the carrier's destination tag.

0075In the case of an ALLOCATE function, as determined at the block 512, the outcome record contains a vector of internal destination identifiers accompanied by the percentage of all calls each destination is to receive. The percentages must total one hundred. A distribution algorithm is implemented at a block 518 for designating one of the destinations using the percentages. The identifier for the destination is resolved to a carrier destination tag at a block 520.

0076SELECT function processing from the block 514 is similar to allocate processing. The outcome record contains an outcome identifier. The outcome identifier points to a table entry containing resource group identifiers for each precedence level. The entry is terminated by a network treatment identifier or other last choice destination. Beginning with the first precedence level, the routing processor 132 uses the resource group identifier to locate an entry in a table of destinations by resource group. The destination identifier's point entry is in a capacity table that contains the output of the output table generated by the activity monitor 134. For each precedence level reached, a distribution algorithm is implemented at blocks 522 and 524 using the capacity percentages within that precedence level. When the destination is selected, then the destination identifier is resolved into a carrier destination tag at a block 526. If no destination passes the test, then the calls will be directed away from the customer destination and presented to the carrier for network treatment, or delivery to a last choice destination.

0077For example, during non-business hours during the week and all day on weekends the subscriber may have only one resource operational. During those times all calls would be routed to a single resource at a designated location. In another example, calls during the business day west of the Mississippi River are to be equally distributed among three West Coast locations. In this case an allocation function is used with the percentages set to one third for each of the locations. In still another example, the subscriber has ten locations throughout the country. The fifty states are divided into ten groups with calls from each group routed to the nearest location as a first choice route. If the resource is predicted to be busy at the first choice location, then the call is routed to a second choice which may include any of the locations in the country. If all of the second choice locations are predicted to be busy, then the call is queued at the nearest location.

0078The final result produced by any of the three routing functions at the block 516, 520 and 526 is a single resource identifier which constitutes the routing response sent at a block 528 into the carrier's SCP 106. There may be additional information about the call or caller obtained during the routing process that is included in the carrier routing response to be delivered along with the call to the subscriber resource. This information may include, for example, the caller attributes obtained from the caller database 118, as discussed above. From the block 528, the routine returns as by looping back to the beginning of the same.

0079With reference to Fig. 6, a flow diagram illustrates operation of the routing algorithm performed once each processing cycle for generating the outcome table describing the destinations and preference to which calls are routed by the routing processor for the SELECT function. Particularly, the load balance algorithm runs continuously and as soon as one process ends a subsequent one begins. The length of time it takes to complete one load balance process is termed a processing cycle. In the illustrated embodiment of the invention, each processing cycle takes approximately thirty seconds to complete.

0080The algorithm begins at a block 600 by collecting resource statistics from the subscriber resources. The subscriber resources include ACD's, ARU's, etc. The information collected from each resource includes the number of devices currently active, the number of devices currently idle, and the number of calls currently queued to the resource. For example, a resource may be a group of agents at an ACD with the device being each individual agent, or a resource might be an audio response unit with the device being each individual port on the system. At a block 602, real time outcome statistics are collected from each subscriber routing processor 132. The number of routing processors 132 is dependent on the volume of routing queries and there is typically a minimum of two for reasons of reliability. The statistical information collected from each routing processor 132 includes the number of calls routed to each resource 108-110 and the number of calls for each routing outcome during the most recent processing cycle. Calls that match a particular set of screening criteria are handled by a routing outcome. Particularly, the routing outcome is a SELECT function included in any of the programmed routing rules. For example, a routing outcome might handle all calls west of the Mississippi River, or all calls from non-English speaking customers, or calls from the southeastern part of the country between 10:00 a.m. and noon on Mondays. The routing outcome is a preferred list of resources that are allowed to handle the call. The list may include all of the resources that the subscriber has or a subset of those resources. A resource may appear in many routing outcomes.

0081The list of resources is organized into groups, first choice, second choice and third choice, with one or more resources in each group. Each level of choice is defined as a precedence level and the same resource may appear in one or more levels.

0082At a block 604, the resource base capacity for each resource is calculated. The base capacity represents the call activity that can be sent to each resource so there are no calls in queue or idle resources at the end of the next processing cycle. This is referred to herein as a "zero queue length number". Particularly, the call carrying capacity of each resource is calculated based on the calls actually routed to the resource during the most recent processing cycle, and the impact that volume of calls had on the number of devices idle and number of calls in queue as reported at the end of the processing cycle. The change in the number of devices active is taken into account which may reflect such circumstances as additional agents reporting for work or beginning a lunch break. The figure calculated for each resource is a base capacity which represents the call volume which can be sent to the resource in the next processing cycle which results in no devices idle and no calls in queue at the end of that processing cycle.

0083At a block 606, a predicted demand for each routing outcome is calculated. This comprises a moving average of the number of calls for each routing outcome for the two most recent processing cycles to predict what the call demand for that routing outcome will be in the next processing cycle.

0084At a block 608, a precedence level value is set equal to one to provide for calculating load balancing for the first precedence level. For each resource, the subscriber defines the performance criteria to be applied for each precedence level. The precedence level one criteria indicates the maximum queue length that the subscriber will allow for the resource as a first choice in any routing outcome. The base capacity plus the maximum queue length at the level is equal to the call carrying capacity of the resource at the first precedence level. This is done by adjusting the base capacity for the level at a block 610. The load balancing process then distributes the call carrying capacity of each resource among the routing outcomes according to demand at the first precedence level at a block 612. If a resource appears in more than one group at the same precedence level, then that resource's call capacity is distributed between the groups in proportion to the call demand being made on each group. For multiple resources in a group, the call capacity of each resource is used to the same degree, that is, for the same percentage of the available capacity.

0085At a decision block 614, a determination is made if all demand has been distributed at the current precedence level. If demand remains after all the possible capacity at this precedence level has been used, then the load balance process advances to a block 616 which advances to the next precedence level and then returns to the block 610. New resources may make themselves available to a routing outcome at the next precedence level as it is adjusted at the block 610. Thereafter, at the block 612, the system attempts to distribute any unused capacity for these resources. Particularly, the call carrying capacity at this precedence level is calculated for each resource. This is done by adding any unused capacity still remaining to the increase in queue length allowed between this precedence level and the prior one. The load balance distribution is then repeated and the decision block 614 again determines if all demand has been distributed. In the described embodiment of the invention, there are three precedence levels and this loop is performed up to three times, once for each precedence level.

0086When all demand has been handled for all routing outcomes, then for each routing outcome a set of demand distribution capacity percentages and maximum resource capacities is passed to the routing processor at a block 618. The call routing process uses these percentages and capacities to route calls for that routing outcome during the next processing cycle. The next load balancing processing cycle then begins.

0087The call routing process applies the capacity percentages for each precedence level to each call routed according to a particular routing outcome. The result is a call routed to a particular resource or the call spilling to the next precedence level. A cumulative count is maintained of the calls routed to each resource and when that count exceeds the maximum resource capacity at a given precedence level, then the call is not routed to that resource but instead spills to the next precedence level.

0088The following is an example of how the load balancing algorithm is implemented. In this example, four routing outcomes are defined as follows: ROUTING OUTCOME 1: SELECT (A,(A,B),(A,B,C,D)), BALANCE, ERRMSG    LEVEL 1 - DESTINATION A    LEVEL 2 - A,B    LEVEL 3 - A,B,C,D ROUTING OUTCOME 2: SELECT (B,(A,B),(A,B,C,D)), BALANCE, ERRMSG    LEVEL 1 - B    LEVEL 2 - A,B    LEVEL 3 - A,B,C,D ROUTING OUTCOME 3: SELECT (C,(C,D),(A,B,C,D)), BALANCE, ERRMSG    LEVEL 1 - C    LEVEL 2 - C,D    LEVEL 3 - A,B,C,D ROUTING OUTCOME 4: SELECT (D,(C,D),(A,B,C,D)), BALANCE, ERRMSG    LEVEL 1 - D    LEVEL 2 - C,D    LEVEL 3 - A,B,C,D    In this example, the following Table 1 lists resource statistics collected as at the block 600 of Fig. 6 during previous processing cycles. <tables id="tabl0001" num="0001"><table frame="all"><title>Table 1</title><tgroup cols="9" colsep="1" rowsep="0"><colspec colnum="1" colname="col1" colwidth="17.50mm" /><colspec colnum="2" colname="col2" colwidth="17.50mm" /><colspec colnum="3" colname="col3" colwidth="17.50mm" /><colspec colnum="4" colname="col4" colwidth="17.50mm" /><colspec colnum="5" colname="col5" colwidth="17.50mm" /><colspec colnum="6" colname="col6" colwidth="17.50mm" /><colspec colnum="7" colname="col7" colwidth="17.50mm" /><colspec colnum="8" colname="col8" colwidth="17.50mm" /><colspec colnum="9" colname="col9" colwidth="17.50mm" /><thead valign="top"><row><entry namest="col1" nameend="col1" align="center">DEST</entry><entry namest="col2" nameend="col5" align="center">MOST RECENT POLL</entry><entry namest="col6" nameend="col9" align="center">2ND MOST RECENT POLL</entry></row><row><entry namest="col1" nameend="col1" /><entry namest="col2" nameend="col2" align="center">PPM</entry><entry namest="col3" nameend="col3" align="center">AAV</entry><entry namest="col4" nameend="col4" align="center">Q</entry><entry namest="col5" nameend="col5" align="center">ACTUAL CALLS HANDLED</entry><entry namest="col6" nameend="col6" align="center">PPM</entry><entry namest="col7" nameend="col7" align="center">AAV</entry><entry namest="col8" nameend="col8" align="center">Q</entry><entry namest="col9" nameend="col9" align="center">ACTUAL CALLS HANDLED</entry></row></thead><tbody valign="top"><row><entry namest="col1" nameend="col1" align="left">A</entry><entry namest="col2" nameend="col2" align="right">78</entry><entry namest="col3" nameend="col3" align="right">2</entry><entry namest="col4" nameend="col4" align="right">0</entry><entry namest="col5" nameend="col5" align="right">5</entry><entry namest="col6" nameend="col6" align="right">75</entry><entry namest="col7" nameend="col7" align="right">0</entry><entry namest="col8" nameend="col8" align="right">4</entry><entry namest="col9" nameend="col9" align="right">10</entry></row><row><entry namest="col1" nameend="col1" align="left">B</entry><entry namest="col2" nameend="col2" align="right">100</entry><entry namest="col3" nameend="col3" align="right">0</entry><entry namest="col4" nameend="col4" align="right">2</entry><entry namest="col5" nameend="col5" align="right">21</entry><entry namest="col6" nameend="col6" align="right">100</entry><entry namest="col7" nameend="col7" align="right">10</entry><entry namest="col8" nameend="col8" align="right">0</entry><entry namest="col9" nameend="col9" align="right">15</entry></row><row><entry namest="col1" nameend="col1" align="left">C</entry><entry namest="col2" nameend="col2" align="right">125</entry><entry namest="col3" nameend="col3" align="right">5</entry><entry namest="col4" nameend="col4" align="right">0</entry><entry namest="col5" nameend="col5" align="right">18</entry><entry namest="col6" nameend="col6" align="right">125</entry><entry namest="col7" nameend="col7" align="right">10</entry><entry namest="col8" nameend="col8" align="right">0</entry><entry namest="col9" nameend="col9" align="right">13</entry></row><row rowsep="1"><entry namest="col1" nameend="col1" align="left">D</entry><entry namest="col2" nameend="col2" align="right">5</entry><entry namest="col3" nameend="col3" align="right">1</entry><entry namest="col4" nameend="col4" align="right">0</entry><entry namest="col5" nameend="col5" align="right">0</entry><entry namest="col6" nameend="col6" align="right">0</entry><entry namest="col7" nameend="col7" align="right">0</entry><entry namest="col8" nameend="col8" align="right">0</entry><entry namest="col9" nameend="col9" align="right">0</entry></row></tbody></tgroup></table></tables>

0089In Table 1, the heading PPM designates primary positions manned, the heading AAV designates agents available, and the heading Q designates the number of calls in queue.

0090The following Table 2 illustrates by way of example outcome statistic information collected at the block 602, see Fig. 6, from the routing processor 132. <tables id="tabl0002" num="0002"><table frame="all"><title>Table 2</title><tgroup cols="3" colsep="1" rowsep="0"><colspec colnum="1" colname="col1" colwidth="52.50mm" /><colspec colnum="2" colname="col2" colwidth="52.50mm" /><colspec colnum="3" colname="col3" colwidth="52.50mm" /><thead valign="top"><row><entry namest="col1" nameend="col1" align="center">ROUTING OUTCOME</entry><entry namest="col2" nameend="col2" align="center">MOST RECENT POLL ACTUAL CALLS ROUTED</entry><entry namest="col3" nameend="col3" align="center">2ND MOST RECENT POLL ACTUAL CALLS ROUTED</entry></row></thead><tbody valign="top"><row><entry namest="col1" nameend="col1" align="left">1</entry><entry namest="col2" nameend="col2" align="right">12</entry><entry namest="col3" nameend="col3" align="right">20</entry></row><row><entry namest="col1" nameend="col1" align="left">2</entry><entry namest="col2" nameend="col2" align="right">14</entry><entry namest="col3" nameend="col3" align="right">6</entry></row><row><entry namest="col1" nameend="col1" align="left">3</entry><entry namest="col2" nameend="col2" align="right">10</entry><entry namest="col3" nameend="col3" align="right">5</entry></row><row><entry namest="col1" nameend="col1" align="left">4</entry><entry namest="col2" nameend="col2" align="right">8</entry><entry namest="col3" nameend="col3" align="right">7</entry></row><row rowsep="1"><entry namest="col1" nameend="col1" align="left">TOTAL</entry><entry namest="col2" nameend="col2" align="right">44</entry><entry namest="col3" nameend="col3" align="right">38</entry></row></tbody></tgroup></table></tables>

0091For this example, it is assumed that the allowable queue length at level one for all destinations is zero, at level two for all destinations it is one, and at level three for all destinations it is four.

0092The algorithm initially calculates base capacity for each resource as follows: <dl id="dl0001"><dt>STEP 1.0:</dt><dd>Calculate base capacity for each resource.</dd><dt>STEP 1.1:</dt><dd>Resource A C = AAV - Q - PPM delta C = current cycle - prior cycle base = prev cycle (calls handled) + AAV - Q + delta C (see Note 1 below) C = 2 - 0 - 78 = -76 Delta C = -76 - (0 - 4 - 75) = 3 base = 5 + 2 - 0 + 3 = 10</dd><dt>STEP 1.2:</dt><dd>Resource B base = 21 + 0 - 2 + (-12) = 7</dd><dt>STEP 1.3:</dt><dd>Resource C base = 18 + 5 - 0 + (-5) = 18</dd><dt>STEP 1.4:</dt><dd>Resource D base = 0 + 1 - 0 + (-4) = -3</dd></dl>    since the result is less than the number of agents currently available then set equal to AAV = 1 Note 1: The formula for base capacity is arrived at as follows: <ul id="ul0004" list-style="none"><li>a) The base capacity is the call load that will result in zero agents available and zero calls in queue at the end of the next processing cycle.</li><li>b) Starting with the calls actually handled in the most recent cycle, add the agents available and subtract the calls in queue as of the end of the most recent cycle. The assumption is that the average call length is much longer than a processing cycle.</li><li>c) Adjust the result based on the change in agents available and change in queue length between the end of the most recent processing cycle and the prior one. To the result in b) add the increase in agents available and subtract the increase in calls in queue.</li><li>d) If the increase in agents available came from an increase in agent positions manned (PPM) subtract the PPM increase from the agents available increase.</li></ul>

0093The next step calculates the predicted demand for each routing outcome as follows: <dl id="dl0002"><dt>STEP 2.0:</dt><dd>Calculate predicted demand for each routing outcome</dd><dt>STEP 2.1:</dt><dd>Outcome 1 demand = (current calls routed + previous calls routed)/2 = (12 + 20)/2 = 16</dd><dt>STEP 2.2:</dt><dd>Outcome 2 demand = (14 + 6)/2 = 10</dd><dt>STEP 2.3:</dt><dd>Outcome 3 demand = (10 + 5)/2 = 7.5</dd><dt>STEP 2.4:</dt><dd>Outcome 4 demand = (8 + 7)/2 = 7.5</dd></dl>    Next, the adjusted base capacity is calculated at precedence level 1, as follows: <dl id="dl0003"><dt>STEP 3.0:</dt><dd>Calculate adjusted base capacity at precedence level 1</dd><dt>STEP 3.1:</dt><dd>Resource A L1cap = base + acceptable Q length at level 1 = 10 + 0 = 10</dd><dt>STEP 3.2:</dt><dd>Resource B L1cap = 7 + 0 = 7</dd><dt>STEP 3.3:</dt><dd>Resource C L1cap = 18 + 0 + 18</dd><dt>STEP 3.4:</dt><dd>Resource D L1cap = 1 + 0 = 1</dd></dl>    At this point, it is possible to distribute the capacity among the routing outcomes according to demand at the first precedence level. This is done as follows: <dl id="dl0004"><dt>STEP 4:</dt><dd>For each resource group calculate the fraction of each outcome's demand handled by that resource group at precedence level 1. If there are multiple resources in the group prorate across each individual resource. Perform both of these steps by prorating based on the adjusted capacity of each resource.</dd><dt>STEP 4.1:</dt><dd>Resource A All capacity is used for Outcome 1 and is less than demand. Fraction of outcome 1 handled = L1cap/demand = 10/16 = .625 6 calls from Outcome 1 spill to Level 2</dd><dt>STEP 4.2:</dt><dd>Resource B All capacity is used for Outcome 2 and is less than demand. fraction of outcome 2 handled = 7/10 = .700 3 calls from Outcome 2 spill to Level 2</dd><dt>STEP 4.3:</dt><dd>Resource C All capacity is used for Outcome 3 and is greater than demand. fraction of outcome 3 handled = 1.0 Resource C has 10.5 calls available for Level 2</dd><dt>STEP 4.4:</dt><dd>Resource D All capacity is used for Outcome 4 and is less than demand. fraction of Outcome 4 handled = 1/7.5 = .1333 6.5 calls from Outcome 4 spill to Level 2</dd></dl>    Once the distribution is completed at Level 1 without all demand being distributed, then new adjusted capacities for each resource at level 2 are calculated as follows: <dl id="dl0005"><dt>STEP 5.0:</dt><dd>Calculate new adjusted capacities for each resource at Level 2 by adding the unused capacity from Level 1 to the increase in acceptable Q length between Levels 1 and 2.</dd><dt>STEP 5.1:</dt><dd>Resource A L2cap = unused capacity from L1cap + increase in Q length = 0 + 1 = 1</dd><dt>STEP 5.2</dt><dd>Resource B L2cap = 0 + 1 = 1</dd><dt>STEP 5.3</dt><dd>Resource C L2cap = 10.5 + 1 = 11.5</dd><dt>STEP 5.4</dt><dd>Resource D L2cap = 0 + 1 + 1</dd></dl>    Next, the remaining capacity is distributed among the routing outcomes according to demand at Level 2 as follows: <dl id="dl0006"><dt>STEP 6:</dt><dd>For each Resource group calculate the fraction of each outcome's demand handled by that group at precedence level 2. If there are multiple resources in the group prorate across each individual resource.</dd><dt>STEP 6.1:</dt><dd>Resource Group 2.1 - Group 1 at Level 2 comprised of Resources A and B resource group capacity = sum of each resource = L2cap(A) + L2cap(B) = 1 + 1 = 2 total demand on RG2.1 = sum of spilled calls from each outcome = 6 (Outcome 1) + 3 (Outcome 2) = 9 fraction of demand handled = capacity/demand = 2/9 = .2222 (this fraction applies to both outcome 1 & outcome 2) fraction of demand spilled = 1 - fraction of demand handled - 1 - .2222 = .7778 calls spilled from each outcome = demand to Level 2 * fraction of demand spilled = 6 (Outcome 1) * .7778 = 4.667 = 3 (Outcome 2) * .7778 = 2.333</dd><dt>STEP 6.1.1:</dt><dd>Resource A fraction of RG2.1 capacity = L2cap(A)/resource group capacity = 1/2 = .5 fraction of demand handled = fraction of capacity * fraction of demand = .5 * .2222 = .1111 (this fraction applies to both outcome 1 and outcome 2) unused capacity available to Level 3 = 0</dd><dt>STEP 6.1.2:</dt><dd>Resource B fraction of RG2.1 capacity = L2cap(B)/resource group capacity = 1/2 = .5 fraction of demand handled = fraction of capacity * fraction of demand = .5 * .2222 = .1111 (this fraction applies to both outcome 1 and outcome 2) unused capacity available to Level 3 = 0</dd><dt>STEP 6.2:</dt><dd>Resource Group 2.2 - Group 2 at Level 2 comprised of Resources C and D resource group capacity = sum of each resource = L2cap(C) + L2cap(D) = 11.5 + 1 = 12.5 total demand on RG2.2 = sum of spilled calls from each outcome = 0 (Outcome 3) + 6.5 (Outcome 4) = 6.5 fraction of demand handled = 1.0 since capacity > demand (this fraction applies to both outcome 3 & outcome 4) fraction of demand spilled = 0</dd><dt>STEP 6.2.1:</dt><dd>Resource C fraction of RG2.2 capacity = L2cap(C)/resource group capacity = 11.5/12.5 = .92 fraction of demand handled = fraction of capacity * fraction of demand = .92 * 1.0 = .92 (this fraction applies to both outcome 3 and outcome 4) unused capacity available to Level 3 = L2cap(C) - (demand * fraction of demand handled) 11.5 - (6.5 * .92) = 5.52</dd><dt>STEP 6.2.2:</dt><dd>Resource D fraction of RG2.2 capacity = L2cap(D)/resource group capacity = 1/12.52 = .08 fraction of demand handled = fraction of capacity * fraction of demand - .08 * 1.0 = .08 (this fraction applies to both outcome 3 and outcome 4) unused capacity available to Level 3 = L2cap(D) - (demand * fraction of demand handled) 1 - (6.5 * .08) = 0.48</dd></dl>    With not all demand yet handled, the new adjusted capacities for each resource at Level 3 are calculated as follows: <dl id="dl0007"><dt>STEP 7.0:</dt><dd>Calculate new adjusted capacities for each resource at Level 3 by adding the unused capacity from Level 2 to the increase in acceptable Q length between Levels 2 and 3.</dd><dt>STEP 7.1:</dt><dd>Resource A L3cap = unused capacity from L2cap + increase in Q length = 0 + 3 = 3</dd><dt>STEP 7.2:</dt><dd>Resource B L3cap = 0 + 3 = 3</dd><dt>STEP 7.3:</dt><dd>Resource C L3cap = 5.52 + 3 = 8.52</dd><dt>STEP 7.4:</dt><dd>Resource D L3cap = 0.48 + 3 = 3.48</dd></dl>    This demand is then distributed in step 8 as follows: <dl id="dl0008"><dt>STEP 8.0:</dt><dd>For each Resource group calculate the fraction of each outcome's demand handled by that group at precedence level 3. If there are multiple resources in the group prorate across each individual resource.</dd><dt>STEP 8.1:</dt><dd>Resource Group 3.1 - Group 1 at Level 3 comprised of Resources A, B, C, and D resource group capacity = sum of each resource = L3cap(A+B+C+D) = 3+3+8.52+3.48 = 18 total demand on RG3.1 = sum of spilled calls from each outcome = 4.667 (Outcome 1) + 2.333 (Outcome 2) = 7 fraction of demand handled = 1.0 (this fraction applies to both outcome 1 & outcome 2) fraction of demand spilled = 0</dd><dt>STEP 8.1.1:</dt><dd>Resource A fraction of RB3.1 capacity = L3cap(A)/resource group capacity = 3/18 = .167 fraction of demand handled = fraction of capacity * fraction of demand = .167 * 1.0 = .167 (this fraction applies to both outcome 1 & outcome 2)</dd><dt>STEP 8.1.2:</dt><dd>Resource B fraction of RG3.1 capacity = L3cap(B)/resource group capacity = 3/18 = .167 fraction of demand handled = fraction of capacity * fraction of demand = .167 * 1.0 = .167 (this fraction applies to both outcome 1 & outcome 2)</dd><dt>STEP 8.1.3:</dt><dd>Resource C fraction of RG3.1 capacity = L3cap(C)/resource group capacity = 8.52/18 = .473 fraction of demand handled = fraction of capacity * fraction of demand = .473 * 1.0 = .473 (this fraction applies to both outcome 1 & outcome 2)</dd><dt>STEP 8.1.4:</dt><dd>Resource D fraction of RG3.1 capacity = L3cap(D)/resource group capacity = 3.48/18 = .193 fraction of demand handled = fraction of capacity * fraction of demand = .193 * 1.0 = .193 (this fraction applies to both outcome 1 and outcome 2)</dd></dl>    Finally, with all demand having been distributed and no demand spilling to the last choice destination, the outcome tables are generated so that the determined fractions can be passed to the routing processor 132 as follows: STEP 9.0: The fractions passed to the Routing Processor 132 are as follows:    Fraction of an Outcome's demand handled by each Resource <tables id="tabl0003" num="0003"><img file="EP0559979A2_D0002.tif" /></tables> Following is the table of fractions converted to the number of calls based on the predicted demand. <tables id="tabl0004" num="0004"><img file="EP0559979A2_D0003.tif" /></tables>

0094During the subsequent processing cycle calls routed based on one of the four routing outcomes use these tables for making the routing decision, as discussed above relative to Fig. 5.

0095In summary, the AIN processor 114 provides call-by-call routing to provide businesses with the maximum routing flexibility possible and allows for the effective management of call centers, call center agents, and other resources, independent of the public switched telephone network 100. Calls can be routed based on the specific identity of the caller. Even more importantly, calls can be routed based on the availability of each resource relative to the desired performance criteria.

0096The number translation and routing decision is based on cooperative processing performed in both the carrier's SCP 106 and the subscriber's AIN processor 114. A query is launched from the carrier's SCP 106 to the subscriber's AIN processor 114 with all information known about the call. The AIN processor 114 makes the actual number translation and routing decision which is returned to the carrier SCP 106 which implements the decision within the carrier network. With cooperative processing, the carrier SCP 106 can route all calls using the current non-dynamic routing process should there be a temporary failure in the AIN processor 114. There may also be periods of time when the subscriber's AIN processor 114 is planned not to be operational, such as when N00 calling volume is low and a non-dynamic routing process is satisfactory.<img file="EP0559979A2_D0004.tif" /><img file="EP0559979A2_D0005.tif" /><img file="EP0559979A2_D0006.tif" /><img file="EP0559979A2_D0007.tif" /><img file="EP0559979A2_D0008.tif" /><img file="EP0559979A2_D0009.tif" /><img file="EP0559979A2_D0010.tif" /><img file="EP0559979A2_D0011.tif" /><img file="EP0559979A2_D0012.tif" /><img file="EP0559979A2_D0013.tif" /><img file="EP0559979A2_D0014.tif" /><img file="EP0559979A2_D0015.tif" /><img file="EP0559979A2_D0016.tif" /><img file="EP0559979A2_D0017.tif" /><img file="EP0559979A2_D0018.tif" /><img file="EP0559979A2_D0019.tif" /><img file="EP0559979A2_D0020.tif" /><img file="EP0559979A2_D0021.tif" /><img file="EP0559979A2_D0022.tif" /><img file="EP0559979A2_D0023.tif" /><img file="EP0559979A2_D0024.tif" /><img file="EP0559979A2_D0025.tif" /><img file="EP0559979A2_D0026.tif" /><img file="EP0559979A2_D0027.tif" /><img file="EP0559979A2_D0028.tif" /><img file="EP0559979A2_D0029.tif" /><img file="EP0559979A2_D0030.tif" /><img file="EP0559979A2_D0031.tif" /><img file="EP0559979A2_D0032.tif" /><img file="EP0559979A2_D0033.tif" /><img file="EP0559979A2_D0034.tif" /><img file="EP0559979A2_D0035.tif" /><img file="EP0559979A2_D0036.tif" />