CA2288459C - System for discounting in a bidding process based on quality of service - Google Patents
System for discounting in a bidding process based on quality of service Download PDFInfo
- Publication number
- CA2288459C CA2288459C CA002288459A CA2288459A CA2288459C CA 2288459 C CA2288459 C CA 2288459C CA 002288459 A CA002288459 A CA 002288459A CA 2288459 A CA2288459 A CA 2288459A CA 2288459 C CA2288459 C CA 2288459C
- Authority
- CA
- Canada
- Prior art keywords
- bid
- bids
- service
- class
- agent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/04—Network management architectures or arrangements
- H04L41/046—Network management architectures or arrangements comprising network management agents or mobile agents therefor
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/30—Decision processes by autonomous network management units using voting and bidding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5029—Service quality level-based billing, e.g. dependent on measured service level customer is charged more or less
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04Q—SELECTING
- H04Q3/00—Selecting arrangements
- H04Q3/0016—Arrangements providing connection between exchanges
- H04Q3/0062—Provisions for network management
- H04Q3/0095—Specification, development or application of network management software, e.g. software re-use
Abstract
A multi-agent system for selecting a service based on competing bids, comprising a Bid Manager agent for issuing a call for bids, receiving the bi ds and selecting a best bid from among the received bids taking into consideration Quality of Service, and a plurality of Bidder agents for issuing the bids in response t o the call for bids, wherein one of the Bidder agents issues the best bid and provides the service upon selection of the best bid by the Bid Manager.
Description
SYSTEM FOR DISCOUNTING IN A BIDDING PROCESS BASED ON QUALITY
OF SERVICE
FIELD OF THE INVENTION
The present invention relates in general to mufti-agent systems and more particularly to a system for taking into account the Quality of Service (QoS) of a bidder agent when quantifying the value of a bid for providing services.
1o BACKGROUND OF THE INVENTION
Telecommunication systems have recently been designed for providing a variety of real-time services and features in an open distributed environment through the collaboration of a set of software components called agents. Such mufti-agent 15 systems are designed in such a way that they may adapt and evolve in the face of changing environments. One such mufti-agent system is known as MANA (Multi-Agent Architecture for Networking Applications) developed by Mitel Corporation.
Through the use of a distributed agent architecture, the system meets high reliability levels and adapts to accommodate technological or service evolution. To achieve 20 these goals, intelligence or learning mechanisms are provided to update service information derived from the operation of the agents. This information is used to redefine the agents and to reallocate resources for correcting failures and to meet the requirements of a defined service more precisely.
25 An application or service in a mufti-agent system is mapped as a series of calls amongst agents to perform the service. Each agent specifies its type, quantity and quality of service (QoS) in order to provide for an overall application. Since multi-agent systems are implemented in an open environment, no agent has prior knowledge of any other agent. The only knowledge that an agent possesses is its requirements 3o and capabilities to provide a specific type of service. Thus, an agent may be required to find other agents to fulfill certain of its service requirements. A calling agent (referred to herein as a Bid Manager) sends out a bid for services to a plurality of ,) called agents (referred to herein as Bidders), each of whom may be capable of providing the necessary resources for the Bid Manager to complete its task.
The Bid Manager receives and evaluates the bids from the various Bidders and selects the agent which has the best chance of success in performing the requested service. This S is referred to herein as serecting the "lowest bid".
One example of a classic bidding mechanism of the foregoing type is disclosed in commonly owned U.S. Patent 5,6?5,636 entitled Adaptive Method of Allocating Calls.
A problem with mufti-agent bidding systems, such as discussed above, is that QoS is not normally taken into consideration in evaluating bids from the bidding agents. Thus, whereas a certain resource (e.g. long distance carrier) may present a low bid in terms of actual cost per call, it may suffer from undesirable bandwidth restrictions or failure rates which are not reflected in the bid. A Bid Manager may be well advised to enter into a contract with another Bidder which presents a higher cost bid but whose QoS is significantly higher.
SUMMARY OF THE INVENTION
According to the present invention, a system is provided for automatically monitoring and learning the QoS of each Bidder and incorporating the QoS into the cost of the bids (i.e. a poor performance leads to a lesser value of the bid).
Therefore, various aspects of the invention are provided as follows:
A mufti-agent system for selecting a service based on competing bids, comprising:
a Bid Manager agent for issuing a call for bids, receiving said bids and selecting a best bid from among said bids taking into consideration Quality of Service; and 2a a plurality of Bidder agents for issuing said bids in response to said call for bids, wherein one of said Bidder agents issues said best bid and provides said service upon selection of said best bid by said Bid Manager.
In a mufti-agent system for selecting a service based on competing bids, said mufti-agent system including a Bid Manager agent and a plurality of Bidder agents, the method comprising the steps of a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids based on a stored Quality of Service and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent selecting a best bid from a predetermined one of said Bidder agents and rejecting all other ones of said Bidder agents;
d) after said service has been provided generating feedback from said Bid Manager agent to said predetermined one of said Bidder agents concerning its Quality of Service; and e) within said predetermined one of said Bidder agents storing said feedback for usage in calculating bids responsive to subsequent calls for bids from said Bid Manager agent.
In a mufti-agent system for selecting a service based on competing bids, said mufti-agent system including a Bid Manager agent and a plurality of Bidder agents, the method comprising the steps of:
a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent modifying said bids based on previously stored Quality of Service for said Bidder agents, selecting a best bid from a predetermined one of said Bidder agents on the basis of said modifying and rejecting all other ones of said Bidder agents; and 2b d) after said service has been provided updating said Quality of Service for usage in subsequent modification of bids received from said Bidder agents responsive to subsequent calls for bids from said Bid Manager agent.
BRIEF DESCRIPTION OF THE DRAWINGS
An embodiment of the invention is described herein below with reference to the accompanying drawings, in which:
Figure 1 is a use case diagram showing use of the system by various external actors;
Figure 2 is a class diagram showing the internal representation of an agent in general terms;
Figure 2A is a class diagram of a Bidder agent according to the representation of Figure 2;
Figure 2B is a class diagram of a Bid Manager agent according to the representation of Figure 2;
to Figure 3A is an interaction diagram showing a system according to the present invention where cost discounting is effected by each Bidder;
Figure 3B is an interaction diagram showing a system according to an alternative embodiment of the present invention where cost discounting is effected by 15 the Bid Manager using Informants to store information about the Bidders;
Figure 4 is a flowchart showing the steps of a method for assessing QoS
according to the present invention; and 2o Figure 5 shows a graph of Success Rate as a function of Success/Failure of an example service provided by a Bidder for the calculation of a bid.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
25 The Figures and description herein of a preferred embodiment of the invention follow the ODP (Open Distributed Processing) principle of viewpoints (ITU-T
Recommendations X901 to X905 ~ ISO/IEC 10746), and use UML (Unified Modeling Language - http://www.rational.com/uml) object-oriented methodology notations.
Turning to Figure 2, which represents an enterprise viewpoint of the system, a plurality of Actors and Use Cases are shown, as follows:
Actors The Bid Manager: The Bid Manager is a software agent that calls for bids and selects the best bid from a plurality of Bidders. An example of a Bid Manager in U.S.
Patent 5,675,636 is the assessment agent.
The Bidder: The Bidder is a software agent that, upon receipt of a call for a bid, sends a bid for a given service that it would provide if it is selected.
An example of a Bidder in U.S. Patent 5,675,636 is the costing agent.
Use cases Monitor and learn QoS: This use case starts when a given selected Bidder has finished offering its service. The Bid Manager assesses the quality of service and either:
1- returns the QoS to the selected Bidder as feedback; or
OF SERVICE
FIELD OF THE INVENTION
The present invention relates in general to mufti-agent systems and more particularly to a system for taking into account the Quality of Service (QoS) of a bidder agent when quantifying the value of a bid for providing services.
1o BACKGROUND OF THE INVENTION
Telecommunication systems have recently been designed for providing a variety of real-time services and features in an open distributed environment through the collaboration of a set of software components called agents. Such mufti-agent 15 systems are designed in such a way that they may adapt and evolve in the face of changing environments. One such mufti-agent system is known as MANA (Multi-Agent Architecture for Networking Applications) developed by Mitel Corporation.
Through the use of a distributed agent architecture, the system meets high reliability levels and adapts to accommodate technological or service evolution. To achieve 20 these goals, intelligence or learning mechanisms are provided to update service information derived from the operation of the agents. This information is used to redefine the agents and to reallocate resources for correcting failures and to meet the requirements of a defined service more precisely.
25 An application or service in a mufti-agent system is mapped as a series of calls amongst agents to perform the service. Each agent specifies its type, quantity and quality of service (QoS) in order to provide for an overall application. Since multi-agent systems are implemented in an open environment, no agent has prior knowledge of any other agent. The only knowledge that an agent possesses is its requirements 3o and capabilities to provide a specific type of service. Thus, an agent may be required to find other agents to fulfill certain of its service requirements. A calling agent (referred to herein as a Bid Manager) sends out a bid for services to a plurality of ,) called agents (referred to herein as Bidders), each of whom may be capable of providing the necessary resources for the Bid Manager to complete its task.
The Bid Manager receives and evaluates the bids from the various Bidders and selects the agent which has the best chance of success in performing the requested service. This S is referred to herein as serecting the "lowest bid".
One example of a classic bidding mechanism of the foregoing type is disclosed in commonly owned U.S. Patent 5,6?5,636 entitled Adaptive Method of Allocating Calls.
A problem with mufti-agent bidding systems, such as discussed above, is that QoS is not normally taken into consideration in evaluating bids from the bidding agents. Thus, whereas a certain resource (e.g. long distance carrier) may present a low bid in terms of actual cost per call, it may suffer from undesirable bandwidth restrictions or failure rates which are not reflected in the bid. A Bid Manager may be well advised to enter into a contract with another Bidder which presents a higher cost bid but whose QoS is significantly higher.
SUMMARY OF THE INVENTION
According to the present invention, a system is provided for automatically monitoring and learning the QoS of each Bidder and incorporating the QoS into the cost of the bids (i.e. a poor performance leads to a lesser value of the bid).
Therefore, various aspects of the invention are provided as follows:
A mufti-agent system for selecting a service based on competing bids, comprising:
a Bid Manager agent for issuing a call for bids, receiving said bids and selecting a best bid from among said bids taking into consideration Quality of Service; and 2a a plurality of Bidder agents for issuing said bids in response to said call for bids, wherein one of said Bidder agents issues said best bid and provides said service upon selection of said best bid by said Bid Manager.
In a mufti-agent system for selecting a service based on competing bids, said mufti-agent system including a Bid Manager agent and a plurality of Bidder agents, the method comprising the steps of a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids based on a stored Quality of Service and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent selecting a best bid from a predetermined one of said Bidder agents and rejecting all other ones of said Bidder agents;
d) after said service has been provided generating feedback from said Bid Manager agent to said predetermined one of said Bidder agents concerning its Quality of Service; and e) within said predetermined one of said Bidder agents storing said feedback for usage in calculating bids responsive to subsequent calls for bids from said Bid Manager agent.
In a mufti-agent system for selecting a service based on competing bids, said mufti-agent system including a Bid Manager agent and a plurality of Bidder agents, the method comprising the steps of:
a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent modifying said bids based on previously stored Quality of Service for said Bidder agents, selecting a best bid from a predetermined one of said Bidder agents on the basis of said modifying and rejecting all other ones of said Bidder agents; and 2b d) after said service has been provided updating said Quality of Service for usage in subsequent modification of bids received from said Bidder agents responsive to subsequent calls for bids from said Bid Manager agent.
BRIEF DESCRIPTION OF THE DRAWINGS
An embodiment of the invention is described herein below with reference to the accompanying drawings, in which:
Figure 1 is a use case diagram showing use of the system by various external actors;
Figure 2 is a class diagram showing the internal representation of an agent in general terms;
Figure 2A is a class diagram of a Bidder agent according to the representation of Figure 2;
Figure 2B is a class diagram of a Bid Manager agent according to the representation of Figure 2;
to Figure 3A is an interaction diagram showing a system according to the present invention where cost discounting is effected by each Bidder;
Figure 3B is an interaction diagram showing a system according to an alternative embodiment of the present invention where cost discounting is effected by 15 the Bid Manager using Informants to store information about the Bidders;
Figure 4 is a flowchart showing the steps of a method for assessing QoS
according to the present invention; and 2o Figure 5 shows a graph of Success Rate as a function of Success/Failure of an example service provided by a Bidder for the calculation of a bid.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
25 The Figures and description herein of a preferred embodiment of the invention follow the ODP (Open Distributed Processing) principle of viewpoints (ITU-T
Recommendations X901 to X905 ~ ISO/IEC 10746), and use UML (Unified Modeling Language - http://www.rational.com/uml) object-oriented methodology notations.
Turning to Figure 2, which represents an enterprise viewpoint of the system, a plurality of Actors and Use Cases are shown, as follows:
Actors The Bid Manager: The Bid Manager is a software agent that calls for bids and selects the best bid from a plurality of Bidders. An example of a Bid Manager in U.S.
Patent 5,675,636 is the assessment agent.
The Bidder: The Bidder is a software agent that, upon receipt of a call for a bid, sends a bid for a given service that it would provide if it is selected.
An example of a Bidder in U.S. Patent 5,675,636 is the costing agent.
Use cases Monitor and learn QoS: This use case starts when a given selected Bidder has finished offering its service. The Bid Manager assesses the quality of service and either:
1- returns the QoS to the selected Bidder as feedback; or
2- stores the QoS in a database for later use in modifying the value of subsequently received bids from the Bidder according to past performance. In this case, the Bid Manager is also able to learn (i.e. extract more general information) out of the database of previous QoS, which can be used to disregard further bids from a 2o particular Bidder if the Bidder's performance is particularly poor.
Calculate bid using QoS: This use case starts when the Bidder receives a call for bid. The Bidder alters the classic calculation of the bid by taking into account the QoS feedback received from the Bid Manager. In the event that the Bid Manager does not send any QoS feedback, the Bidder simply calculates the bid in a classical way, such as described in U.S. Patent 5,675,636, and the Bid Manager is responsible for modifying the bid value by taking into account the previously stored QoS
information for the particular Bidder.
Calculate bid: As discussed above, U.S. Patent 5,675,636 provides a description of how to calculate a bid in the particular context of least cost routing.
Turning to Figure 2, a class diagram is provided comprising a general representation of the information viewpoint of the system, with emphasis on the resource management and selection aspects. The class diagram includes a class for ResourceManager, ResourceAdapter, Feature and Informant, as follows:
Classes The Feature class represents a portion of the logic of an agent. It uses the ResourceManager to select ResourceAdapters to trigger actions on its environment.
The ResourceManager class manages a set of ResourceAdapters, and is responsible for selecting them. In order to do so, the ResourceAdapter may consult an Informant.
The ResourceAdapter is a class that provides a uniform interface to access APIs of resources.
The Informant is an optional class entity for monitoring the use and the 2o performance of a given selected ResourceAdapter. It can thereby help the ResourceManager in its selection process by providing a history of its use, or even a learned pattern of its behavior.
Turning to Figures 2A and 2B, the general class design of Figure 2 is applied to the particular cases of the Bidder and the Bid Manager, respectively.
In Figure 2A, the BidderFeature accesses the success rate of a given ResourceAdapter with QoS properties for use in the calculation of a bid. This class also sends QoS feedback to the ResourceAdapter for modifying its success rate based 3o on the success or failure of the last usage. The QoSResourceAdapter is a particular type of ResourceAdapter that contains notions of Quality of Service. Based on feedback concerning success or failure of a given operation, it calculates a SuccessRate() that is accessible to the feature, which can then be used to alter the calculation of a bid.
In Figure 2B, the BidManagerFeature receives bid values from the Bidders (using a BidderAgentAdapter) in order to select the best service. Once the service has been completed, the BidManagerFeature sends feedback to the selected Bidder (via the BidderAgentAdapter) concerning the quality of provided service. The 1o BidderAgentAdapter is an interface for communicating with the Bidder and using its services. The Quality of Service (QoS) of the BidderAgent can then be monitored by an Informant.
Turning to Figures 3A and 3B, interaction diagrams are provided for two 15 scenarios. Figure 3A illustrates the case where the discounting is done on the Bidder side. Figure 3B illustrates the case when the discounting is done on the Bid Manager side using Informants to store information about Bidders, in which case there is no requirement to send QoS feedback to the Bidder.
20 In the scenario of Figure 3A, the BidManager relies on the Bidders to provide a bid value that takes the QoS into account. Therefore the BidManager needs to send feedback to the selected bidder (step 12) after the service has been provided so that the Bidder can calculate the new value of its bid for the next time (step 13 -setSuccessRate).
In the scenario of Figure 3B, the BidManager monitors the QoS of the Bidders. Therefore it does not need to send feedback to them.
There are two calculations involved in the discounting method of the present 3o invention: assessment of the current Quality of Service of a given Bidder, and the use of that value in the calculation of the bid.
With respect to the first calculation, the simplest assessment of QoS is the success rate of the provided service. The success rate is calculated from received QoS
feedback concerning the most recent usage and information about past usage (such as previous success rates, how many times the service has been requested, etc.) Different calculation methods are possible, such as simple averaging, exponential averaging, etc.
Figure 4 is a flowchart showing a simple method based on a sampling window to of the last n uses of a service. This method is based on the realization that since the QoS can vary in time, it is not necessary to take into account all previous cases, and it is more important to know the current performance of the service. Thus, in the example shown in Figure 4, if the most recent use of the service leads to a failure (i.e.
the most recent usages go from two successes and a failure to two failures and only 15 one success in the last n uses), the new success rate for the Bidder decreases from 2/3 to 1/3. An example of how the success rate can be calculated is shown in the graph of Figure 5.
With respect to the second calculation (i.e. calculation of the bid), if the 20 assessment of the QoS consists simply of a certain success rate; then it is easy to modify the value of the bid so that it takes QoS into account, as follows:
NewBid = Bid / Success Rate where a lower Bid is considered to be better than a higher bid.
25 This simple calculation can either be performed by the Bidder or by the Bid Manager. If performed by the Bidder (which is only possible if the Bidder is part of the overall system, and is not provided by a third party, for reasons of trust), the Bid Manager has to chose the best (i.e. the lowest) bid, but also must provide feedback to the selected Bidder after its use of the service, as discussed above with reference to 3o Figure 3A. If the calculation is performed by the Bid Manager (thereby permitting the use of third-party Bidders), the Bid Manager monitors and stores the observed QoS of the selected Bidder in a database for future reference. Thus, in this scenario there is no requirement for the Bid Manager to provide feedback to the Bidder, as discussed above with reference to Figure 3B.
Taking as an example the automatic route selection application set forth in U.S. Patent 5,675,636, the assessing agent assumes the role of Bid Manager and the costing agents assume the role of Bidders. Consider a situation wherein there are two costing agents, A and B, one of which is to be selected by the assessing agent to route a long distance call. If agent A's bid is 10 cents per minute and agent B's bid is 12 cents per minute, but the assessing agent knows that while A and B behave similarly 1o and in an acceptable manner for voice calls, A has failed 4 times in its last 10 data transmissions (thereby requiring wasteful and costly repeated calls for the same data session), whereas B has only failed once, the assessing agent recalculates the bids in the following way:
A's bid:=10/(1-0.4)=16.67 B's bid:=12/(1-0.1)=13.33 Since the discounted bid from agent B is less expensive, the assessing agent chooses B to provide the long distance service.
Assuming that B fails in this instance to provide the requested service, the 2o assessing agent, having detected the failure as a semantic error, updates the success rate value of B. If, for example, the most recent previous failure of B was 6 calls previously, the assessing agent now knows that B has failed 2 times during the last 10 calls. This information is then used to discount the next bid received from agent B.
According to the alternative embodiment discussed above, the assessing agent relies on the costing agents to calculate their own failure rates, in which case the assessing agent provides each costing agent with feedback on its performance after each call, to assist the costing agent in calculating the new success rate for its next bid.
As discussed above, this alternative is only viable where the costing agents are part of 3o the system, and are not third party agents provided by the long distance carriers themselves.
The foregoing example may be represented using pseudo Java code, as follows:
BidderFeature code:
calculateBid() {
cost = getCostPerMinute(BillingPlan);
cost . cost / getSuccessRate();
return cost;
QoSResourceAdapter code:
//instance variables:
int successRate;
Queue lastncalls;
int N; //size of the queue getSuccessRate() {
return successRate;
feedback(bool success) {
//implementation of the flowchart // storing only the success values of the last N calls lastncalls.push(success);
lastncalls.pop;
setSuccessRate();
}
setSuccessRate() {
// count numbers of successes in the last N calls successes = count(lastncalls, true);
successRate = successes / N;
}
On the Bid Manager's side the pseudo Java code is as follows:
BidderAgentAdapter code:
sendFeedback(Bidder bidder, bool success) {
if (call completed) send(bidder, true);
else send(bidder, false);
getBidValue(bidder) {
send(bidder, callForBids);
msg = recvFrom(bidder); //blocking return msg.bidValue;
In summary, a system is provided for automatically monitoring and Learning the QoS of each Bidder and incorporating the QoS into the cost of the bids (i.e. a poor performance leads to a lesser value of the bid) for use in selecting a Bidder to provide a requested service, thereby extending the practical application of mufti-agent bidding processes.
Alternatives and variations of the invention are possible. For example, to although the example presented herein relates to bidding by costing agents to provide a long distance carrier server in an automatic route selection system, the principles of the invention have broad applications to many mufti-agent systems where Bid Managers require the services of resources which are represented in the system by agents. For example, in a networking environment, a print manager agent may tender bids from competing print server agents, each representing a printer resource of the system. Or, an airline reservation agent may tender bids from several airline ticketing agents and discount the bids based on flight cancellations, departure delays, etc. All such applications are believed to be within the sphere and scope of the claims appended hereto.
Calculate bid using QoS: This use case starts when the Bidder receives a call for bid. The Bidder alters the classic calculation of the bid by taking into account the QoS feedback received from the Bid Manager. In the event that the Bid Manager does not send any QoS feedback, the Bidder simply calculates the bid in a classical way, such as described in U.S. Patent 5,675,636, and the Bid Manager is responsible for modifying the bid value by taking into account the previously stored QoS
information for the particular Bidder.
Calculate bid: As discussed above, U.S. Patent 5,675,636 provides a description of how to calculate a bid in the particular context of least cost routing.
Turning to Figure 2, a class diagram is provided comprising a general representation of the information viewpoint of the system, with emphasis on the resource management and selection aspects. The class diagram includes a class for ResourceManager, ResourceAdapter, Feature and Informant, as follows:
Classes The Feature class represents a portion of the logic of an agent. It uses the ResourceManager to select ResourceAdapters to trigger actions on its environment.
The ResourceManager class manages a set of ResourceAdapters, and is responsible for selecting them. In order to do so, the ResourceAdapter may consult an Informant.
The ResourceAdapter is a class that provides a uniform interface to access APIs of resources.
The Informant is an optional class entity for monitoring the use and the 2o performance of a given selected ResourceAdapter. It can thereby help the ResourceManager in its selection process by providing a history of its use, or even a learned pattern of its behavior.
Turning to Figures 2A and 2B, the general class design of Figure 2 is applied to the particular cases of the Bidder and the Bid Manager, respectively.
In Figure 2A, the BidderFeature accesses the success rate of a given ResourceAdapter with QoS properties for use in the calculation of a bid. This class also sends QoS feedback to the ResourceAdapter for modifying its success rate based 3o on the success or failure of the last usage. The QoSResourceAdapter is a particular type of ResourceAdapter that contains notions of Quality of Service. Based on feedback concerning success or failure of a given operation, it calculates a SuccessRate() that is accessible to the feature, which can then be used to alter the calculation of a bid.
In Figure 2B, the BidManagerFeature receives bid values from the Bidders (using a BidderAgentAdapter) in order to select the best service. Once the service has been completed, the BidManagerFeature sends feedback to the selected Bidder (via the BidderAgentAdapter) concerning the quality of provided service. The 1o BidderAgentAdapter is an interface for communicating with the Bidder and using its services. The Quality of Service (QoS) of the BidderAgent can then be monitored by an Informant.
Turning to Figures 3A and 3B, interaction diagrams are provided for two 15 scenarios. Figure 3A illustrates the case where the discounting is done on the Bidder side. Figure 3B illustrates the case when the discounting is done on the Bid Manager side using Informants to store information about Bidders, in which case there is no requirement to send QoS feedback to the Bidder.
20 In the scenario of Figure 3A, the BidManager relies on the Bidders to provide a bid value that takes the QoS into account. Therefore the BidManager needs to send feedback to the selected bidder (step 12) after the service has been provided so that the Bidder can calculate the new value of its bid for the next time (step 13 -setSuccessRate).
In the scenario of Figure 3B, the BidManager monitors the QoS of the Bidders. Therefore it does not need to send feedback to them.
There are two calculations involved in the discounting method of the present 3o invention: assessment of the current Quality of Service of a given Bidder, and the use of that value in the calculation of the bid.
With respect to the first calculation, the simplest assessment of QoS is the success rate of the provided service. The success rate is calculated from received QoS
feedback concerning the most recent usage and information about past usage (such as previous success rates, how many times the service has been requested, etc.) Different calculation methods are possible, such as simple averaging, exponential averaging, etc.
Figure 4 is a flowchart showing a simple method based on a sampling window to of the last n uses of a service. This method is based on the realization that since the QoS can vary in time, it is not necessary to take into account all previous cases, and it is more important to know the current performance of the service. Thus, in the example shown in Figure 4, if the most recent use of the service leads to a failure (i.e.
the most recent usages go from two successes and a failure to two failures and only 15 one success in the last n uses), the new success rate for the Bidder decreases from 2/3 to 1/3. An example of how the success rate can be calculated is shown in the graph of Figure 5.
With respect to the second calculation (i.e. calculation of the bid), if the 20 assessment of the QoS consists simply of a certain success rate; then it is easy to modify the value of the bid so that it takes QoS into account, as follows:
NewBid = Bid / Success Rate where a lower Bid is considered to be better than a higher bid.
25 This simple calculation can either be performed by the Bidder or by the Bid Manager. If performed by the Bidder (which is only possible if the Bidder is part of the overall system, and is not provided by a third party, for reasons of trust), the Bid Manager has to chose the best (i.e. the lowest) bid, but also must provide feedback to the selected Bidder after its use of the service, as discussed above with reference to 3o Figure 3A. If the calculation is performed by the Bid Manager (thereby permitting the use of third-party Bidders), the Bid Manager monitors and stores the observed QoS of the selected Bidder in a database for future reference. Thus, in this scenario there is no requirement for the Bid Manager to provide feedback to the Bidder, as discussed above with reference to Figure 3B.
Taking as an example the automatic route selection application set forth in U.S. Patent 5,675,636, the assessing agent assumes the role of Bid Manager and the costing agents assume the role of Bidders. Consider a situation wherein there are two costing agents, A and B, one of which is to be selected by the assessing agent to route a long distance call. If agent A's bid is 10 cents per minute and agent B's bid is 12 cents per minute, but the assessing agent knows that while A and B behave similarly 1o and in an acceptable manner for voice calls, A has failed 4 times in its last 10 data transmissions (thereby requiring wasteful and costly repeated calls for the same data session), whereas B has only failed once, the assessing agent recalculates the bids in the following way:
A's bid:=10/(1-0.4)=16.67 B's bid:=12/(1-0.1)=13.33 Since the discounted bid from agent B is less expensive, the assessing agent chooses B to provide the long distance service.
Assuming that B fails in this instance to provide the requested service, the 2o assessing agent, having detected the failure as a semantic error, updates the success rate value of B. If, for example, the most recent previous failure of B was 6 calls previously, the assessing agent now knows that B has failed 2 times during the last 10 calls. This information is then used to discount the next bid received from agent B.
According to the alternative embodiment discussed above, the assessing agent relies on the costing agents to calculate their own failure rates, in which case the assessing agent provides each costing agent with feedback on its performance after each call, to assist the costing agent in calculating the new success rate for its next bid.
As discussed above, this alternative is only viable where the costing agents are part of 3o the system, and are not third party agents provided by the long distance carriers themselves.
The foregoing example may be represented using pseudo Java code, as follows:
BidderFeature code:
calculateBid() {
cost = getCostPerMinute(BillingPlan);
cost . cost / getSuccessRate();
return cost;
QoSResourceAdapter code:
//instance variables:
int successRate;
Queue lastncalls;
int N; //size of the queue getSuccessRate() {
return successRate;
feedback(bool success) {
//implementation of the flowchart // storing only the success values of the last N calls lastncalls.push(success);
lastncalls.pop;
setSuccessRate();
}
setSuccessRate() {
// count numbers of successes in the last N calls successes = count(lastncalls, true);
successRate = successes / N;
}
On the Bid Manager's side the pseudo Java code is as follows:
BidderAgentAdapter code:
sendFeedback(Bidder bidder, bool success) {
if (call completed) send(bidder, true);
else send(bidder, false);
getBidValue(bidder) {
send(bidder, callForBids);
msg = recvFrom(bidder); //blocking return msg.bidValue;
In summary, a system is provided for automatically monitoring and Learning the QoS of each Bidder and incorporating the QoS into the cost of the bids (i.e. a poor performance leads to a lesser value of the bid) for use in selecting a Bidder to provide a requested service, thereby extending the practical application of mufti-agent bidding processes.
Alternatives and variations of the invention are possible. For example, to although the example presented herein relates to bidding by costing agents to provide a long distance carrier server in an automatic route selection system, the principles of the invention have broad applications to many mufti-agent systems where Bid Managers require the services of resources which are represented in the system by agents. For example, in a networking environment, a print manager agent may tender bids from competing print server agents, each representing a printer resource of the system. Or, an airline reservation agent may tender bids from several airline ticketing agents and discount the bids based on flight cancellations, departure delays, etc. All such applications are believed to be within the sphere and scope of the claims appended hereto.
Claims (9)
1. A multi-agent system for selecting a service based on competing bids, comprising:
a Bid Manager agent for issuing a call for bids, receiving said bids and selecting a best bid from among said bids taking into consideration Quality of Service;
and a plurality of Bidder agents for issuing said bids in response to said call for bids, wherein one of said Bidder agents issues said best bid and provides said service upon selection of said best bid by said Bid Manager.
a Bid Manager agent for issuing a call for bids, receiving said bids and selecting a best bid from among said bids taking into consideration Quality of Service;
and a plurality of Bidder agents for issuing said bids in response to said call for bids, wherein one of said Bidder agents issues said best bid and provides said service upon selection of said best bid by said Bid Manager.
2. The multi-agent system of claim 1, wherein said Bid Manager provides feedback to said Bidders concerning said Quality of Service and each of said Bidders discounts its bid based on said Quality of Service feedback received from said Bid Manager.
3. The multi-agent system of claim 1, wherein said Bid Manager assesses said Quality of Service for each of said Bidders for storage in a database, and discounts said bids based on said stored Quality of Service.
4. The multi-agent system of claim 2, wherein each one of said Bidder agents comprises a QoSResourceAdapter class for establishing an interface to access an API
of a resource for providing said service and calculating a success rate for said resource in providing said service based on the feedback from said Bid Manager; a ResourceManager class for managing and selecting said QoSResourceAdapter; and a BidderFeature class for accessing said success rate calculated by said QoSResourceAdapter class and in response calculating and issuing a bid to said Bid Manager and for receiving said feedback from said Bid Manager and forwarding said feedback to said QoSResourceAdapter class for use in calculating said success rate.
of a resource for providing said service and calculating a success rate for said resource in providing said service based on the feedback from said Bid Manager; a ResourceManager class for managing and selecting said QoSResourceAdapter; and a BidderFeature class for accessing said success rate calculated by said QoSResourceAdapter class and in response calculating and issuing a bid to said Bid Manager and for receiving said feedback from said Bid Manager and forwarding said feedback to said QoSResourceAdapter class for use in calculating said success rate.
5. The multi-agent system of claim 4, wherein said Bid Manager further comprises a BidderAgentAdapter class for providing an interface for communicating with each of said Bidder agents; and a BidManagerFeature class for receiving said bids and selecting said best bid from among said bids received from said Bidders via said BidderAgentAdapter class and sending said feedback to said Bidders via said BidderAgentAdapter class.
6. The multi-agent system of claim 3, wherein each one of said Bidder agents comprises a ResourceAdapter class for establishing an interface to access an API of a resource for providing said service; a ResourceManager class for managing and selecting said ResourceAdapter; and a BidderFeature class for calculating and issuing a bid to said Bid Manager.
7. The multi-agent system of claim 6, wherein said Bid Manager further comprises a BidderAgentAdapter class for providing an interface for communicating with each of said Bidder agents; an Informant class for monitoring use and performance of said BidderAgentAdapter class; a ResourceManager class for consulting said Informant class; and a BidManagerFeature class for using said ResourceManager class and said BidderAgentAdapter class to receive said bids and selecting said best bid from among said bids based on said use and performance monitored by said Informant class.
8. In a multi-agent system for selecting a service based on competing bids, said multi-agent system including a Bid Manager agent and a plurality of Bidder agents, the method comprising the steps of:
a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids based on a stored Quality of Service and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent selecting a best bid from a predetermined one of said Bidder agents and rejecting all other ones of said Bidder agents;
d) after said service has been provided generating feedback from said Bid Manager agent to said predetermined one of said Bidder agents concerning its Quality of Service; and e) within said predetermined one of said Bidder agents storing said feedback for usage in calculating bids responsive to subsequent calls for bids from said Bid Manager agent.
a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids based on a stored Quality of Service and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent selecting a best bid from a predetermined one of said Bidder agents and rejecting all other ones of said Bidder agents;
d) after said service has been provided generating feedback from said Bid Manager agent to said predetermined one of said Bidder agents concerning its Quality of Service; and e) within said predetermined one of said Bidder agents storing said feedback for usage in calculating bids responsive to subsequent calls for bids from said Bid Manager agent.
9. In a multi-agent system for selecting a service based on competing bids, said multi-agent system including a Bid Manager agent and a plurality of Bidder agents, the method comprising the steps of:
a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent modifying said bids based on previously stored Quality of Service for said Bidder agents, selecting a best bid from a predetermined one of said Bidder agents on the basis of said modifying and rejecting all other ones of said Bidder agents; and d) after said service has been provided updating said Quality of Service for usage in subsequent modification of bids received from said Bidder agents responsive to subsequent calls for bids from said Bid Manager agent.
a) within said Bid Manager agent issuing a call for said competing bids to said plurality of Bidder agents;
b) within each of said Bidder agents calculating said competing bids and tendering said bids to said Bid Manager agent;
c) within said Bid Manager agent modifying said bids based on previously stored Quality of Service for said Bidder agents, selecting a best bid from a predetermined one of said Bidder agents on the basis of said modifying and rejecting all other ones of said Bidder agents; and d) after said service has been provided updating said Quality of Service for usage in subsequent modification of bids received from said Bidder agents responsive to subsequent calls for bids from said Bid Manager agent.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| GB9824567A GB2343583B (en) | 1998-11-09 | 1998-11-09 | System for discounting in a bidding process based on quality of service |
| GB9824567.3 | 1998-11-09 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CA2288459A1 CA2288459A1 (en) | 2000-05-09 |
| CA2288459C true CA2288459C (en) | 2002-12-31 |
Family
ID=10842138
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002288459A Expired - Lifetime CA2288459C (en) | 1998-11-09 | 1999-11-03 | System for discounting in a bidding process based on quality of service |
Country Status (3)
| Country | Link |
|---|---|
| US (1) | US6687682B1 (en) |
| CA (1) | CA2288459C (en) |
| GB (1) | GB2343583B (en) |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7010511B1 (en) * | 1999-03-31 | 2006-03-07 | Kinney Jr Sam E | Method and system for conducting electronic auctions with net present value bidding |
| CA2384436C (en) | 1999-09-10 | 2010-06-29 | Portogo, Inc. | Systems and method for insuring correct data transmission over the internet |
| US7711646B2 (en) | 1999-09-10 | 2010-05-04 | Transurety, Llc | Methods and apparatus for providing coverage for receiver of transmission data |
| JP3732699B2 (en) * | 1999-12-27 | 2006-01-05 | 富士通株式会社 | Electronic purchasing system and method |
| US20030069986A1 (en) * | 2000-05-23 | 2003-04-10 | Lori Petrone | Electronic marketplace system and method using optimization techniques |
| US7373325B1 (en) * | 2000-10-13 | 2008-05-13 | Nortel Networks Limited | Automated trading for e-markets |
| US6915275B2 (en) * | 2001-06-07 | 2005-07-05 | International Business Machines Corporation | Managing customization of projects prior to manufacture in an electronic commerce system |
| US6965877B2 (en) * | 2001-06-07 | 2005-11-15 | International Business Machines Corporation | Brokering and facilitating consumer projects in an e-commerce system |
| US7305272B2 (en) * | 2002-12-16 | 2007-12-04 | Rockwell Automation Technologies, Inc. | Controller with agent functionality |
| US7146232B2 (en) * | 2002-12-16 | 2006-12-05 | Rockwell Automation Technologies, Inc. | Agent program environment |
| US8380528B2 (en) * | 2004-07-13 | 2013-02-19 | At&T Intellectual Property I, L. P. | Controlling service provided by a packet switched network based on bids from consumer equipment |
| US7827260B2 (en) * | 2004-10-12 | 2010-11-02 | International Business Machines Corporation | Apparatus, system, and method for configuring zone control of a network feature in a heterogeneous network |
| US7991627B2 (en) * | 2005-06-21 | 2011-08-02 | General Electric Company | Injected drug identification and fail-safe system |
| US8214450B2 (en) * | 2005-08-01 | 2012-07-03 | Limelight Networks, Inc. | Dynamic bandwidth allocation |
| US20070250342A1 (en) * | 2006-04-21 | 2007-10-25 | Ravinder Sohal | Systems and methods for automatically generating bids for medical services and goods |
| US20080040141A1 (en) | 2006-07-20 | 2008-02-14 | Torrenegra Alex H | Method, System and Apparatus for Matching Sellers to a Buyer Over a Network and for Managing Related Information |
| US20080133375A1 (en) * | 2006-12-01 | 2008-06-05 | Alex Henriquez Torrenegra | Method, System and Apparatus for Facilitating Selection of Sellers in an Electronic Commerce System |
| US20090287596A1 (en) * | 2008-05-15 | 2009-11-19 | Alex Henriquez Torrenegra | Method, System, and Apparatus for Facilitating Transactions Between Sellers and Buyers for Travel Related Services |
| US11238415B2 (en) * | 2019-10-10 | 2022-02-01 | Nice Ltd. | Systems and methods for intelligent adherence or conformance analysis coaching |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5579467A (en) | 1992-05-27 | 1996-11-26 | Apple Computer, Inc. | Method and apparatus for formatting a communication |
| US5440739A (en) | 1992-06-11 | 1995-08-08 | Beck Systems, Inc. | Method of maintaining updated set-up configurations in workgroups on a network |
| US5802502A (en) * | 1993-05-24 | 1998-09-01 | British Telecommunications Public Limited Company | System for selective communication connection based on transaction pricing signals |
| JPH07302236A (en) * | 1994-05-06 | 1995-11-14 | Hitachi Ltd | Information processing system, method therefor and service providing method in the information processing system |
| CA2123068C (en) * | 1994-05-06 | 1998-04-14 | Thomas A. Gray | Adaptive method for allocating calls |
| WO1997029443A1 (en) * | 1996-02-12 | 1997-08-14 | British Telecommunications Public Limited Company | Service provision system for use in distributed processing environments |
| GB9606708D0 (en) * | 1996-03-29 | 1996-06-05 | Plessey Telecomm | Bandwidth bidding |
-
1998
- 1998-11-09 GB GB9824567A patent/GB2343583B/en not_active Expired - Lifetime
-
1999
- 1999-11-03 CA CA002288459A patent/CA2288459C/en not_active Expired - Lifetime
- 1999-11-08 US US09/435,857 patent/US6687682B1/en not_active Expired - Lifetime
Also Published As
| Publication number | Publication date |
|---|---|
| GB9824567D0 (en) | 1999-01-06 |
| GB2343583B (en) | 2004-05-05 |
| CA2288459A1 (en) | 2000-05-09 |
| GB2343583A (en) | 2000-05-10 |
| US6687682B1 (en) | 2004-02-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2288459C (en) | System for discounting in a bidding process based on quality of service | |
| US9720731B2 (en) | Methods and apparatus for coordinating and selecting protocols for resources acquisition from multiple resource managers | |
| US6445782B1 (en) | Service management system for use in communications | |
| US6154778A (en) | Utility-based multi-category quality-of-service negotiation in distributed systems | |
| US5644715A (en) | System for scheduling multimedia sessions among a plurality of endpoint systems wherein endpoint systems negotiate connection requests with modification parameters | |
| US11356282B2 (en) | Sending cross-chain authenticatable messages | |
| US7636350B1 (en) | System and method for access provisioning in multiple telecommunications technologies | |
| TW200525938A (en) | Remote system administration using command line environment | |
| US20060053039A1 (en) | Method and apparatus for business process analysis and optimization | |
| CN111258565B (en) | Method, system, server and storage medium for generating applet | |
| US6442619B1 (en) | Software architecture for message processing in a distributed architecture computing system | |
| EP0976074B1 (en) | Service provision system for use in distributed processing environments | |
| Wang et al. | Quality of service (QoS) contract specification, establishment, and monitoring for service level management | |
| US20070162540A1 (en) | Inter-agent communication | |
| US8224933B2 (en) | Method and apparatus for case-based service composition | |
| US8752061B2 (en) | Resource allocation within multiple resource providers based on the incoming resource request and the expected state transition of the resource requesting application, and selecting a resource provider based on the sum of the percentage of the resource provider currently used, the requesting load as a percentage of the resource provider's total resource, and the additional load imposed by the expected state of the requesting application as a percentage of the resource provider's total resource | |
| US20010046285A1 (en) | INAP processing method for communication between SSP and TCAP | |
| CA2331977C (en) | A caching mechanism to optimize a bidding process used to select resources and services | |
| US6314462B1 (en) | Sub-entry point interface architecture for change management in a computer network | |
| US11120513B2 (en) | Capital chain information traceability method, system, server and readable storage medium | |
| EP0940047B1 (en) | A service management system for use in communications | |
| WO1998023098A9 (en) | A service management system for use in communications | |
| CN112395849B (en) | Method and system for generating communication information | |
| US20200387953A1 (en) | Service coordination device and service coordination method | |
| CN114666332A (en) | Data transmission method, edge box, edge cloud and data transmission system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| MKEX | Expiry |
Effective date: 20191104 |