Communications of the ACM

Communications of the ACM

Seven Comments on Charging and Billing

By Philip Ginzboorg
Communications of the ACM, November 2000, Vol. 43 No. 11, Pages 89-92
10.1145/353360.353369
Comments

Profit is the usual motive for commercial service provision. For that reason charging and billing are at the core of the telecommunications business. Charging and billing for telephony service is based on the contract between the customers (buyers) and the telephone company (seller). The essence of this agreement is that the buyer gets access to telephone network services,¹ that is, the user can both place and receive calls, and will pay for those services according to the bills produced by the telephone company. The bills may contain both fixed and usage-dependent charges. To compute usage-dependent charges, the telephone company has to monitor the calls placed and the calls received. This type of monitoring is considered call-based and is done by the switches in the network. For each placed call a switch in the network will produce a Call Detailed Record (CDR). Millions of CDRs are stored on magnetic tape or on hard disk; this is the raw data for the billing system. Thus, the implementation of billing for telephone services is based on an offline billing system and on having CDRs generated in the network. The network records the use and the release of connections.

The telecommunications industry is undergoing three major changes. The first change is regulatory: national, highly regulated industries are evolving into a competitive, single market. Second, wireless telephony is displacing fixed as the main source of revenue (more than half of the new telephones in the world are wireless). Third, by volume, Internet traffic (that is, data services) is becoming the main constituent of fixed network's traffic. One effect of those changes is a new composition of revenues. Witness, for example, the shrinking of the profit margins on long-distance calls caused by deregulation. As another example, the ascent of mobile telephony into a mass market means there is a higher proportion of mobile subscribers who fail to pay their bills. This has triggered the introduction of prepaid charging schemes modeled after a parking meter. Yet another example is the opportunity for arbitrage² caused by the situation where voice is expensive and data cheap. The majority of the investments in Internet telephony and Internet fax are motivated by this imbalance [10].

---

To be useful, payment protocols must be complemented by methods to choose the services that are to be delivered and by ways to charge for those services.

---

Back to Top

The Challenge of Data

There are technical and business challenges in the charging and billing of data services. The delivery of content over the Internet differs from telephony service in several ways. First, the number of services and service providers on the Internet is much larger. Also, in a fixed telephone network, the identity of a user is tied to the user's telephone number and that number serves as the user's network address. On the Internet, identification of a buyer³ cannot be solely based on the buyer's network access point. (Dynamic IP addresses are common; in addition, it is too easy to replace the sender's address in an IP packet.) Second, in a fixed network, telephony is a very reliable service, whereas the quality of service delivered over the Internet is often unpredictable. Third, in a telephone network, the metering of the service usage and the generation of accounting records (CDRs) is done within the network by the switches. The network can know about delivery of data from one terminal to another, but it cannot know about the content in that data. The information about the content, such as the name of a particular movie and its price, is beyond the knowledge of the network. (These difficulties can be dealt with in a variety of ways—see Comment 6 later in this article.)

In general, people are concerned about the security of transactions over the Internet. As a response to those concerns we see more and more integration of cryptographic modules into protocol stacks, operating systems, and application programs.

Comment 1. In a cash-like transaction over the Internet the payment may be done by a variety of mechanisms, some of which include electronic money such as eCash [1], micropayments, like those of IBM [4] and electronic purses, such as that of Mondex [8]. Many people associate the electronic economy with cash-like transactions. However, in the nonelectronic economy, people pay for some things with coins and bank notes and for some things they don't. For example, you do not insert a coin every time you turn the lights on, or open a water tap, or make a telephone call from your residence. Similarly, in the electronic economy, there will be services that are better paid with charging and billing mechanisms.

Comment 2. Today, most of the payments by residential users over the Internet are made through credit card companies. The buyer's Web browser and the seller's Web server establish an encrypted communication link over which the buyer sends his or her credit card number. To secure the link, the browser and the server use the Secure Socket Layer (SSL) protocol [7]. The SSL protocol, introduced by Netscape in 1994, is a general-purpose protocol that provides privacy of communication. In 1996 the credit card companies specified a suite of protocols, called Secure Electronic Transaction (SET), for performing network payments [9]. An advantage of SET over the current practice is that SET reduces the risk of fraud. For example, during a SET transaction the credit card number of the buyer is hidden from the seller.⁴

Note that the payment—by credit card or any other method—is only a part of the activity we call "shopping." In the future, SET or similar protocols may become the preferred way to pay for information services, but they must be complemented by methods to choose the services that are to be delivered and by ways to charge for those services.

Comment 3. In order to guarantee quality of service, one needs to reserve resources in the network. A reservation of resources in a public network should require payment. If the reservation is free, then a few greedy users will reserve all the available resources—such as bandwidth or memory—for themselves. Thus, for any resource reservation scheme that operates in a public network, there should be a way to charge for the reserved resources. The charging method shouldn't be much more intricate to utilize than the service itself.

Comment 4. A give-away commodity is a product that either brings little profit or is sold at loss (that is, it is given away). For example, some mobile network operators give away cellular phones in order to sell calls. Wired access to the Internet is becoming a commodity. But in wireless networks there is a tradeoff between mobility and inexpensive access.

Low-cost wireless access is possible today with Wireless Local Area Network (WLAN) technology [5] and unlicensed radio frequencies—for example, the 2.4GHz band. But the geographical coverage and the speed at which people can move while being connected in such network are quite limited. Also, there is a finite amount of licensed radio frequencies available for mobile communication and the rights to use those frequencies are expensive.

Comment 5. We are likely to witness charging and billing for digital content by a third party (such as by an organization that is different from the seller). In addition to credit card companies, Internet service providers and telephone companies are the natural providers of the service. Third-party billing is to the advantage of both buyers and sellers. To a buyer, the advantage is that the billing service provider combines services of several sellers into one subscription, with all charges presented in a single bill. To a seller, it is easier to enter the market if the jobs of managing subscriptions, monitoring the service usage, and collecting the money from buyers can be outsourced.

Comment 6. Here is how we have dealt with the challenges in mapping from telephony to Internet charging in the BillNeat system developed at the Nokia Research Center [2, 3]. First, services are labeled: each service for which we charge and bill is assigned a unique identifier that is known to both the seller and the billing service provider. It will be part of the records that account for each service usage. Also, each subscriber receives a digital identity. (The identities can be issued by the billing service provider or by a trusted third party.)

Regarding the second challenge, we have to distinguish between a single delivery of a service, such as that of a Web page, and a continuous delivery, such as that of an audiovisual stream. To deal with the possibility of content loss in the case of a single delivery, we allow retransmission for a certain time after receiving a payment (this is often called "pay once, click many times"). To deal with the possibility of data loss in the case of continuous delivery, the buyer pays periodically for small increments of received service. If the content is faulty or if the delivery is interrupted, the buyer's losses are small; they are restricted to the amount of money that was prepaid.

As for the third challenge, we've moved the point at which accounting records are generated to the buyer's terminal. To prevent fraudulent use, the origin and the integrity of each record is checked in real time by a dedicated network server, called the Watchdog. The records are stored as receipts for the performed transaction; a buyer is accountable for all received and correctly signed records. The charging is uncontestable because of buyers' digital signatures. The money amount in the accounting record is payable only to the billing service provider that is indicated within it.

Comment 7. In addition to today's services, future networks will carry a variety of services that are considered advanced from a charging perspective. A service is advanced from a charging point of view if its price during delivery is not constant, or if it consists of several components that are priced separately, or both. Today, we see advanced services emerging on both the application and network levels. Two examples of services that may have varying prices are variable bit-rate transmission (network level) and the delivery of a video interleaved with advertisements (application level). Two examples of a compound service that may have differently priced components are the transmission of data requiring several resource allocations from the network and the delivery of an audiovisual presentation with components that are owned by several content providers.

Back to Top

Conclusion

The payment is only a part of the shopping activity. To be useful, payment protocols must be complemented by methods to choose the services that are to be delivered and by ways to charge for those services.

The range of the possible charging methods lies between two extremes. The first, flat rate, means that the buyer's charges do not depend on usage of the offered services. The advantage of flat rate is its simplicity; the disadvantage is that heavy users are charged the same as the occasional ones. The second method, which is analogous to bean counting, is to produce an accounting record for each delivered data packet. The advantage of bean counting is its accuracy; its disadvantage is that the charging may become more costly to implement than the service that is offered.

There is a business opportunity in providing charging and billing as a service for companies that sell information. Two kinds of companies may benefit from such service. An example of the first is Interactive Investor [6], which provides financial data for a high monthly fee and wants to move down-market. An example of the second is Uproar [11], which provides interactive games free of charge—it is financed through advertisements—and wants to move up-market.

Back to Top

Back to Top

Back to Top

The Digital Library is published by the Association for Computing Machinery. Copyright © 2000 ACM, Inc.

---

No entries found