WO2001028219A2

WO2001028219A2 - Quality-based billing system

Info

Publication number: WO2001028219A2
Application number: PCT/US2000/024941
Authority: WO
Inventors: Thomas Everett Weston; Mary Maureen Madine
Original assignee: Motorola, Inc.
Priority date: 1999-10-14
Filing date: 2000-09-12
Publication date: 2001-04-19
Also published as: AU7576900A; WO2001028219A3

Abstract

By inserting timestamped and/or packet sequence indicia into the communication stream of a communication session (104), one or more quality of service metrics (106) are measured and reported on a per session basis to the billing system. Preferably the quality of service metrics are reported along with usage data that are used to calculate the bill for that session. The billing system (108) maintains records of each communication session, associating the quality of service data with the usage data for that session. In this way the system can be controlled and the customer billed based on the quality of service actually delivered for a given session.

Description

QUALITY-BASED BILLING SYSTEM

BACKGROUND AND SUMMARY OF THE INVENTION

The present invention relates generally to telecommunications systems and networking. More particularly, the invention relates to a system and method of collecting and utilizing quality of service metrics on a per session basis. The metrics may be used, for example, in billing systems and to implement quality-based billing.

Quality of Service Measurement

In current large terrestrial networks, data traffic is typically sent through circuit- switched or connection-oriented channels. The circuit-switched paradigm employed in current technology allows separate quality of service (QoS) measurement packets to be sent over the circuit-switched connections, at times when data payload traffic is not being sent. This is how phone lines are typically tested, for example.

In other words, a telephone company will test its lines, to measure the quality of service being provided, by sending known test signals or messages from point A to point B on a periodic basis. With circuit-switched lines, such signals or messages can provide meaningful QoS measurement because the quality of the connection during testing time is likely to be the same as during use for payload delivery.

Unfortunately, the assumption that the connection remains essentially the same between testing and use falls apart when packet-based communication is employed. In packet-based communication a communication session is broken into a plurality of data packets that are each individually addressed and sequentially identified. Typically the packets are sent separately (possibly over different communication links) and without any guarantee of an ordered sequential delivery. Received packets are ordered and reassembled at the receiver. Conventional circuit-switched QoS measurement techniques do not work well for packet-based communications, because the precise route the data payload packets will take may not be known in advance and may not be capable of being duplicated to perform a subsequent QoS measurement.

Consider, for example, a packet-based communication system employing orbiting satellites. A first sequence of packets may be sent through satellite A from source to destination. A subsequent series of packets comprising part of the same communication session may be sent through satellite B — which has moved in orbit to replace satellite A. Because the spatial arrangement of the orbiting satellites changes constantly, the specific route through which packets are sent changes constantly.

After the communication session has ended, a third series of packets might be sent to measure the quality of service. However, this time a third satellite may have moved into position to replace satellites A and B. Thus, the quality of service measurement at testing time does not reflect the actual quality of service experienced by the user when the data payload packets were sent. Therein lies one of the difficulties in accessing quality of service in packet based systems. Similar issues are also presented in cellular telephone systems, in which the mobile user can be handed off from cell to cell during a communication session, or even in terrestrial based packet systems (e.g. Internet). Packet paths may change based on ever changing traffic loads and equipment failures.

Usage Based Billing

Concurrent with the shift from circuit switched communication systems to packet switched communication systems, there is a growing interest to shift from time based billing to usage based billing. Under the conventional time based billing paradigm, a telecommunication system user is billed according to the time or duration of the communication session. Thus most telephone users today expect to pay by the minute for a long distance call.

Under the usage based billing paradigm, the user is charged according to the quantity of data sent. The quantity of data sent is not necessarily tied to the duration of the communication session, because data throughput can increase or decrease during a communication session.

While there is interest in usage based billing, limitations in conventional technologies have prevented usage based billing from being fully exploited. One problem lies directly in quality of service. In a usage based billing arrangement, quality of service becomes an important factor, because quality of service can affect how quickly packets are delivered. When quality drops, the data throughput rate drops; when quality improves, data throughput increases. Thus for a telecommunications system to work efficiently and cost effectively, the system needs to be well tuned to provide high quality of service at maximum data throughput rates. In addition, users recognize the value of high quality service, and the high throughput rates associated therewith, and are willing to pay more for better quality.

The present invention addresses the foregoing quality of service-related issues and thereby opens new possibilities for quality of service-based billing systems. By adding a small amount of overhead, which can be configured to match system requirements, the quality of service can be measured and reported on a per session basis, at the same time as the data traffic is being sent. Thus the quality of service metric gives a true assessment of the quality of service for that communication session. The session-based quality of service metric makes possible a new billing paradigm whereby billing rates or billing charges can be set or adjusted based on the quality of service actually being delivered for any given session. While the quality of service billing paradigm is particularly well suited to packet switched communication systems, which have traditionally had no effective way of assessing true quality of service, the paradigm can also be used in more conventional circuit switched telecommunications systems. In one aspect, the invention provides a method for billing in a packet-based communication system in which indicia are inserted in at least a portion of the packets transmitted during a communication session between a sender and a receiver. In many implementations, these indicia can be timestamp indicia or packet-ordering identification indicia. The overhead for inserting these indicia is typically quite small, particularly where the telecommunications system may already use such indicia to ensure that packets are timely received and reassembled in proper order at the receiver.

According to the inventive method, at least one quality of service metric is computed based on the inserted indicia. While there are a number of different quality of service metrics that may be computed, an example is the average packet-delay metric (APD) that may be calculated from the timestamp indicia to indicate, on average, how long it took for a packet to be delivered from the sender to the receiver. Another QoS metric example is the packet misinsertion ratio (PMR) that is calculated using the packet- identifying indicia to calculate the number of packets discarded as being too far out of order during the communication session. The foregoing are merely two examples of possible QoS metrics in accordance with the invention. Several others are described below and other equivalent metrics may be developed as well.

Further, in accordance with the inventive method, the quality of service metric or metrics are submitted to a billing system that monitors usage data for the communication session. The quality of service metrics are associated with the usage data, as in a suitable billing system database, for example, for use in billing.

For a more complete understanding of the invention, its objects and advantages, refer to the following specification and to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a system block diagram illustrating a circuit-switched telecommunications network corresponding to one environment in which the invention may be deployed;

Figure 2 is a similar block diagram illustrating a packet-switched telecommunications network, representing another environment in which the invention may be deployed;

Figure 3 is a data structure diagram illustrating one example of a packet header field, useful in understanding how indicia are inserted for subsequent quality of service metric computation;

Figure 4 is a system block diagram of a packet-based system illustrating the data collection process according to the invention;

Figure 5 is a system block diagram illustrating the packet-based system of Figure 4, showing the quality of service reporting and collecting mechanism according to the invention;

Figure 6 is a system block diagram illustrating a circuit-switched system and showing how the quality of service billing system may be deployed in that environment;

Figure 7a-7f illustrate different possible quality of service metrics that may be used in the present invention;

Figure 8 is a data flow diagram showing the relationship among the major components of the quality-based billing system.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

The present provides a system and method for collecting and measuring quality of service information on a per session basis. The system can be used in both circuits- switched and packet-switched telecommunication systems. By way of background illustration, Figure 1 shows a circuit-switched network over which data payload traffic may be sent from sender A to receiver B. In a typical circuit- switched configuration the data payload traffic is sent between one or more nodes through switched connections that are established at the time the session is initiated. Thus in Figure 1, the data payload traffic corresponding to a single session is routed from sender A, through nodes 10, 12, 14, and 16 to the receiver B. It will be appreciated that the network illustrated in Figure 1 could be switched to route traffic via a different route between sender A and receiver B. Typically, once the necessary nodes have been switched to connect sender with receiver, the entire communication session is routed through these connections. This is illustrated diagrammatically in Figure 1 where the entire communication session is illustrated between 18a and 18b. Both sender A and receiver B participate in the entire communication session while connected through the circuit switch connections as illustrated.

In contrast, a packet-switched network configuration is illustrated in Figure 2. The communication session differs in that the session is broken up into a plurality of packets as illustrated diagrammatically at 20. Sender A routes all packets through node 10, however the packets are not required to follow the same circuit- switched connections. Rather, the packets can follow any available route, depending on which nodes are available to receive the packet and have sufficient bandwidth to handle the packet in a timely fashion.

Thus in Figure 2, the packets designated '1' and '2' are routed to node 12; whereas packet '3' is routed to node 13. Thereafter, packet '1' and '2' may also diverge, following different routes as illustrated. Ultimately, however, all packets arriving at receiver B are reassembled into the original data payload message. It is, of course, possible that some packets will arrive out of order. In Figure 2 packet '3' arrives before packet '2'. It is also possible that some packets may not arrive at all, or they may arrive corrupted, or too late to be used. Most communications systems have mechanisms for reliable data delivery that causes missing packets, late packets, or errored packets to be resent. The present invention captures a small amount of network traffic overhead, or piggybacks upon existing overhead, to inject quality of service-related information into the communication session. The system uses mechanisms at the sender or receiver or both to inject indicia that the system uses to extract quality of service information for that communication session.

Although there are numerous opportunities to inject suitable indicia into the communication session, the presently preferred embodiment uses indicia stored in packet payload, or indicia stored in the packet header associated with a data payload packet or both. Figure 3 illustrates such a packet header at 22. The packet header may precede the packet data payload 24, as diagrammatically illustrated. The packet header may comprise several pieces of information that are otherwise pertinent to the packet delivery mechanism. For example, the packet header typically will include a source address identifier 26 and a destination address identifier 28. These are used so the system knows the identity of the sender (source address identifier 26) and the identity of the intended receiver (destination address 28). In addition, the packet header may include a packet sequence number 30 that is used in packet switched systems to allow packets to be reassembled in proper order at the receiving end. Also, the packet header may include a timestamp indicia 32 indicative of the time at which the packet was sent by the sender. The packet data payload 24 may also or alternatively include a timestamp indicia.

In a packet-switched system the packet header, as exemplified by the one shown in Figure 3, is typically associated with or attached to each data payload packet. In a circuit-switched system other techniques may be used for periodically injecting suitable indicia into the data stream. For example, a suitable timestamp indicia may be periodically inserted in the payload data through any of a variety of mechanisms including signal injection in the time domain or in the frequency domain.

Figures 4 and 5 illustrate a packet-based communication system in which a quality of service-based billing system has been implemented. A communication session is established between the source user equipment 40 and the destination user equipment 42 through a communications data network 44. Thus, packet data is sent from source 40 to destination 42 through the communications network. In the preferred embodiment the user equipment at source 40 sends a plurality of timestamped packets, which the user equipment at the destination 42 may receive. Some of the packets may not be received (they may be lost or garbled). Source 40 may also maintain a running count of the number of packets sent during the communication session, and this information may also be encoded in the data packets sent or other transmitted through the network 44. In the preferred embodiment the system may also include a centralized billing system 46 attached to network 44 to receive data from the source and destination user equipment.

The user equipment at destination 42 assembles a collection of data used to assess quality of service for the communication session. In the preferred embodiment destination 42 maintains a record of the number of received packets, as well as a collection of other quality of service metrics that may be calculated by the user equipment at the destination 42. These quality of service metrics will be more fully described below. They include, the average packet delay (APD), the packet delay variation (PDV), the numbered of errored packets (i.e., the number of packets that were garbled during transmission), the errored packet blocks (i.e., the blocks containing more than a predetermined number of errored packets) and the number of misinserted packets (i.e., packets that were received so far out of order that they needed to be discarded and requested to be resent). The user equipment at source 40 and the user equipment at destination 42 thus maintain respective data stores containing the pertinent information needed for calculating the quality of service metrics used by the billing system 46. Referring to Figure 5, these stored data are transmitted along with the session data to the data network and are then communicated to the centralized billing system 46. The centralized billing system matches the session data between sender and receiver and calculates the quality of service for the session. Depending on the billing system configuration, the quality of service metrics can be attached to or compared with the billing service level agreement to determine what rate should be charged for that session. If the session is being billed on a timed basis, the quality of service metric can be used to adjust the billing rate to charge the customer the appropriate modified rate, based on the quality of service actually delivered. Alternatively, if the communication is being billed on a usage basis, the quantity of data sent during the communication session can be tallied up and the appropriate rate applied based on the quality of service actually delivered during that session. The centralized billing system 46 associates the usage data (or time data) with the quality of service metrics so that the appropriate billing formula can be applied. Thus the centralized billing system maintains a record of the number of packets sent and number of packets received (or alternatively the time duration of the session) and one or more quality of service metrics. The presently preferred embodiment may employ one or more of the following quality metrics:

• average packet delay (APD)

• packet delay variation (PDV) • packet misinsertion ratio (PMR)

• packet loss ratio (PLR)

• packet error ratio (PER)

• severely erred block ratio (SEBR).

While the above metrics are presently preferred, others may be fashioned within the scope of the invention.

The technique for implementing the quality-based billing system is similar for circuit-switched networks. As illustrated in Figure 6 the user equipment at source 40 communicates with the user equipment at destination 42 through the circuit-switched data communication network 44. The quality of service metrics pertinent to the circuit- switched communication are generated, by injecting suitable test messages into the communication data stream. The injected test messages can be inserted into short (essentially inaudible) time slices during the communication session. Alternatively, the injected test messages can be inserted in the frequency domain, outside of or at the edge of the bandwidth used to carry the payload traffic. The raw QoS data can be accumulated at either source 40 or destination 42 or both, and so that the user equipment at such location may calculate the QoS metrics. These QoS metrics are then communicated to the centralized billing system 46 along with the usage data. In this way, the QoS metrics for the communication session are captured and associated with the usage data for use in quality-based billing. Figure 7 comprises a series of sub figures 7a-7f illustrating how the presently preferred quality of service metrics are calculated. Referring to Figure 7a, the APD metric is calculated by initiating a timestamped packet at a periodic rate, such as every 100 data packets (configurable) at the sender. These timestamped packets allow the receiving user equipment to be able to determine the transport delay or time it took for the packet to traverse the distance from sender to receiver. As illustrated in Figure 7a the sender A transmits a plurality of packets (n packets, where n > 0) to receiver B. The sender includes its source address identifier and destination address identifier along with a timestamp indicating the time at which at least one of the n packets was sent. If desired, the sender can timestamp each of the packets. In Figure 7a, sender A timestamps the n - 1th packet with timestamp t_s. The packet is sent through network 44, carrying the t_s timestamp as illustrated at 45.

Receiver B has its own clock or access to a known time source with which it determines the receive time t_r of the packet. The sent time t_s and the received time t_r are then extracted and processed by the system to determine the time difference (tr-t_s) at processing module 60. This time difference is then accumulated with similarly computed time differences for other packets in the session and the average packet delay is thus calculated by the system at process 62. This yields the average packet delay metric 64.

The computational processes illustrated at 60 and 62 may be performed at any suitable node on the network. For example, these calculations may be performed at the receiver B. Alternatively, the raw time stamp data (t_s and t_r) can be routed through network 44 to the centralized billing system 46 (Figs. 4 and 5) for the billing system to perform the calculations.

Figure 7b illustrates how the packet delay variation (PDV) metric is calculated. The process is similar to that illustrated in Figure 7a, except that the time stamped data t_s and t_r are used to determine the time delay variation over the communication session. Thus module 66 stores the time delay values (t_rt_s) for all labeled packets in the session and then selects the maximum and minimum packets to compute a difference. This difference indicates the amount of deviation or variation between the shortest delay time and the longest delay time. Processing module 66 in this way calculates the PDV metric 68. As with the APD metric 64, the raw timestamp data may be processed at any node in the system including at the receiver and at the centralized billing system.

Figure 7c illustrates how the packet misinsertion ratio (PMR) is computed. Sender A includes in its packet header a packet sequence number (see Fig. 3) which receiver B uses to reorder the packets when received. Receiver B includes an out-of- sequence rejection module 70 that will automatically screen out or reject packets that arrive so far out of sequence that the incoming message can no longer be reliably assembled. Counter 72 maintains a running count of the number of packets rejected by module 70 during a given communication session. The counter may be reset at the end of each session so that the count at the end of the session may serve as the packet misinsertion ratio metric 74. If desired, the running count may be divided by the number of packets sent to yield a ratio.

Figure 7d illustrates how the packet loss ratio (PLR) is computed. Sender A maintains a data store of the number of packets sent during the communication session. Receiver B maintains a similar data store of the number of packets received during the communication session. Process 76 calculates the difference between the number of packets sent and the number of packets received divided by the number of packets sent and this ratio is then used as the packet loss ratio metric 78. As before, the PLR calculation 76 can be performed at any node on the system, such as at the receiver B or at the centralized billing system.

Figure 7e illustrates how the packet error ratio (PER) metric is computed. The technique is similar to the one illustrated in Figure 7c for calculating the packet misinsertion ratio. Sender A sends packets of information to receiver B and receiver B includes a process 80 for determining if any of these packets are errored. A packet can be deemed errored for a variety of reasons, such as failure to match the data- dependent checksum attached by the sender A. The system maintains a counter 82 that accumulates the total number of erred packets for a given communication session. This total is then used to define the packet error ratio metric 84 by dividing this raw count by the total number of packets sent. Figure 7f illustrates how the severely errored block ratio (SEBR) metric is calculated. The process is similar to that described in Figure 7e. It is assumed that sender A transmits packets of information in groups called blocks of a predetermined size typically dictated by the sender A or by the system architecture. Receiver B has a block discard mechanism 86 that will discard an entire block if it contains too many errored packets. Counter 88 maintains a running total of the number of blocks discarded for a given session and the value stored in counter 88 at the end of the communication session is used to define the severely errored block ratio metric 90 by dividing the raw count by the total number of packets sent.

Any one of more of the above described quality of service parameters (as well as other similar or functionally equivalent parameters) are logged and reported along with the raw usage data to the centralized billing system. The raw usage data may be reflected in a variety of ways, including the total number of packets sent versus the total number of packets received for the session, the time duration of the session, or the like. The billing system then associates these quality of service metrics along with the raw usage data to determine quality of service ratios and ultimately to determine how the customer should be charged for the communication session.

There are a number of different possibilities for charging the customer based on the quality of service metrics so gathered. In one embodiment the system automatically adjusts the billing rate up or down for the communication session, based on the actual quality of service delivered. An alternate embodiment is envisioned whereby the customer subscribes to a certain level of service quality in advance, and the quality of service metrics are used to ensure that the customer is receiving the quality of service as defined in the applicable service level agreement between service provider and customer. If desired, the quality of service metrics can be reflected on the customer's invoice as a way of enhancing customer satisfaction.

The quality of service metrics can also be used in a dynamic fashion to provide real time feedback for a telecommunication system that prioritizes the delivery of certain packets over others, depending on the quality of service contracted. In such a system, premium subscribers would receive first priority for high quality. So long as the premium customers are actually receiving the high quality service contracted (as indicated by the QoS metrics), the additional high quality bandwidth may be used for non-premium traffic. By way of summary, Figure 8 presents a data flow diagram illustrating some of the more fundamental concepts of the quality based billing system of the invention. The quality based billing system involves a sender 100 and receiver 102 which communicate with one another during a communication session 104. The communication session can be of any duration and may comprise any quantity of information (e.g., any number of packets).

In a communication session that supports voice communication, for example, the communication session would typically begin when the receiver answers the call and ends when either sender or receiver terminates the call. In a packet-based data communication, the session could comprise one or more packets or blocks of data. Thus, for example, when a user clicks on a hyperlink within an Internet web page, the resulting sequence of packets sent to the URL associated with that hyperlink may be treated as a single communication session. The data returned back from the web site in response to the user's click might be viewed as a second communication session. The invention is flexible enough, however, to allow even a single packet sent to constitute a communication session. This is possible because the quality of service metrics can be embedded in the single packet sent, thus the quality of service actually experienced by the packet can be recorded by the billing system. With continued reference to Figure 8, the communication session 104 generates at least one quality of service metric 106. This metric is either directly used or combined with other data to formulate QoS data that is sent to billing system 108. The actual usage data from communication 104 is also sent to billing system 108, as illustrated. The billing system maintains a data store 110 which records associated usage data 112 and QoS data 114 for each communication session 118. Typically the usage data is also associated with a given subscriber 116 to allow the billing system to bill that subscriber for the communication session.

As the foregoing has illustrated, the quality based billing system of the invention extracts real time QoS measurements made during the communication session (and thus during usage data collection). The QoS metrics and the usage data may thus be combined to determine session quality on a per session basis. This makes it possible to implement quality based billing systems in both circuit- switched and packet- switched telecommunications networks.

While the invention has been described in its presently preferred embodiments, it will be understood that the invention is capable of modification without departing from the spirit of the invention as set forth in the appended claims.

Claims

CLAIMSWe claim:

1. A method for billing in a packet-based communication system, comprising the steps of: inserting indicia in at least a portion of the packets transmitted during a communication session between a sender and a receiver; computing at least one quality of service metric based on said indicia inserted during said communication session;

2. The method of claim 1 wherein said indicia is a timestamp indicia.

3. The method of claim 1 wherein said indicia is a timestamp indicia and said step of inserting said indicia is performed by the sender.

4. The method of claim 1 wherein said indicia is a packet ordering identification indicia.

5. The method of claim 1 wherein said indicia is a packet identification indicia and said step of inserting said indicia is performed by the sender.

6. The method of claim 1 wherein said indicia is a timestamp indicia and said step of computing at least one quality of service metric is performed by using said timestamp indicia to calculate an indication of the packet delay time from sender to receiver for said communication session.

7. The method of claim 1 wherein said indicia is a timestamp indicia and said step of computing at least one quality of service metric is performed by using said timestamp indicia to calculate an average packet delay (APD) for said communication session.

8. The method of claim 1 wherein said indicia is a timestamp indicia and said step of computing at least one quality of service metric is performed by using said timestamp indicia to calculate a packet delay variation (PDV) for said communication session.

9. The method of claim 1 wherein said indicia is a packet identifying indicia and said step of computing at least one quality of service metric is performed by using said packet identifying indicia to calculate a packet misinsertion ratio (PMR) corresponding to the number of packets discarded as being too far out of order during said communication session.

10. The method of claim 1 wherein said indicia is a packet identifying indicia and said step of computing at least one quality of service metric is performed by using said packet identifying indicia to calculate a packet loss ratio (PLR) corresponding to the number of packets lost during said communication session.

11. The method of claim 1 wherein said indicia is a packet identifying indicia and said step of computing at least one quality of service metric is performed by using said packet identifying indicia to calculate a packet error ratio (PER) corresponding to the number of errored packets received during said communication session.

12. The method of claim 1 wherein said indicia is a packet identifying indicia and said step of computing at least one quality of service metric is performed by using said packet identifying indicia to calculate a severely errored block ratio (SEBR) corresponding to the number of packet blocks that are discarded as having a predetermined number of errored packets during said communication session.

13. The method of claim 1 wherein said step of submitting said quality of service metric is performed by said receiver sending information to said billing system.

14. The method of claim 1 wherein said step of submitting said quality of service metric is performed by said sender sending information to said billing system.

15. The method of claim 1 wherein said step of submitting said quality of service metric is performed by both said sender and said receiver sending information to said billing system.

16. The method of claim 1 wherein said step of computing at least one quality of service metric is performed by said billing system based on information received from at least one of said receiver and said sender.

17. The method of claim 1 wherein said step of computing at least one quality of service metric is performed by said receiver and then transmitted to said billing system.

18. The method of claim 1 further comprising calculating a billing value for said communication session based on said associated quality of service metric and said usage data.

19. A method for billing in a circuit-switched communication system, comprising the steps of: inserting indicia in at least a portion of the message transmitted during a communication session between a sender and a receiver; computing at least one quality of service metric based on said indicia inserted during said communication session.

20. A system for billing communications, comprising: a communication system for defining a communication session between at least one sender and at least one receiver; a measurement mechanism associated with said communication system that generates a quality of service metric associated with said communication session; a usage measurement system associated with said communication system that generates usage data associated with said communication session; a billing system receptive of said quality of service metric and said usage data and that associates said quality of service metric with said usage data for use in billing based on quality of service.