In theory, the Internet is a continuous-rate network. Clients that want to gain access to the internet can request for specific rates such as 98.99 Mbps, or 99.01 Mbps. This is possible, but it causes a lot of complications to the company providing the internet service. For one, the managements and operations of the company will definitely be very difficult to handle. This is the summary of a behavioral targeting by Qian Lv and George Rouskas. You can get the pdf of the behavioral targeting here: Internet Service Tiering.
The provider which attempts to maximize allocation of bandwidth will have to come up with a design to differentiate such very specific rates. It will be very difficult to track the robust traffic in this scenario. Traffic engineering will be a daunting task. These are just some of the problems for such a design.
Tiered Broadband Service
That’s why most, if not all providers, follow a tiered service. Consumers have to choose between different levels of bandwidth offerings. This simplifies the functions of providers, including equipment configuration, network management, traffic engineering, etc. With this simplicity, providers can have the capacity to cater to millions of consumers.
There are two kinds of broad tiering structures followed by major Internet Service Providers (ISP). The first class involves hierarchy rates with each level based on the infrastructure that supports the bandwidth it can offer. The second class involves tiers where the former is twice the latter. It is not certain whether the former still matches well with the data applications of today, and the second class is still debatable in terms of usage patterns among the consumer population.
Tier Structures used in this Study
The paper from which this article is based studies Internet access service and its behavior as being elastic. Elastic means the behavior of consumers with respect to pricing may vary widely. This is opposite to an inelastic service such as essential goods, for at least the short term. As such, the study introduces a model implementing user diversity. The study also develops an algorithm for determining the level of offered service and the price.
Four kinds of tier structures are studied in this study: Optimal, Optimal-rounded, Uniform, and Exponential. The simple solutions are the ones for Uniform and Exponential, which are patterned after the structures followed by major providers. A dynamic programming algorithm was used to obtain the Optimal structure, while Optimal-rounded are the rounded off versions of the optimal tiers, to the nearest 256 Kbps.
There are several important conclusions from the results drawn from this study. First, tiers that are equally spaced are found to be better for the Internet service provider in comparison to exponential Tiers. Second, for a given broadband domain that a provider offers, simply adding more tiers in between is not an effective marketing strategy. More effective is increasing the size and the price of tiers.