Source-linked AI summary
A Survey of Smart Data Pricing: Past Proposals, Current Plans, and Future Trends
Soumya Sen, Carlee Joe-Wong, Sangtae Ha, Mung Chiang
TL;DR
The growth of bandwidth-intensive applications and cloud services is renewing congestion concerns and exposing limits of simple data plans. This survey reviews static and dynamic pricing proposals, their adoption, and the technological, socioeconomic, and regulatory challenges surrounding future pricing innovation.
Problem
Growing mobile-device, application, video, and cloud usage is renewing congestion concerns, while users’ bill sensitivity can reduce data usage even under current plans.
Method
The survey synthesizes known static and dynamic data-network pricing proposals, current ISP practices, network-economics foundations, and related research challenges.
Results
The survey finds that pricing can help alleviate congestion and balance data-network functionality goals, while innovative dynamic schemes still require new architectures, field trials, and interdisciplinary research.
Takeaways & Limitations
Future pricing innovation includes day-ahead time- and usage-dependent pricing and app-based pricing, alongside attention to technological, socioeconomic, and regulatory challenges.
Takeaways & Limitations
Responsive pricing has a slight feedback-loop delay because prices are based on network conditions from the previous time period.
Abstract
from arXiv · showhide
Traditionally, network operators have used simple flat-rate broadband data plans for both wired and wireless network access. But today, with the popularity of mobile devices and exponential growth of apps, videos, and clouds, service providers are gradually moving towards more sophisticated pricing schemes. This decade will therefore likely witness a major change in the ways in which network resources are managed, and the role of economics in allocating these resources. This survey reviews some of the well-known past broadband pricing proposals (both static and dynamic), including their current realizations in various consumer data plans around the world, and discusses several research problems and open questions. By exploring the benefits and challenges of pricing data, this paper attempts to facilitate both the industrial and the academic communities' efforts in understanding the existing literature, recognizing new trends, and shaping an appropriate and timely research agenda.
1. INTRODUCTION
Rapid growth in demand from mobile devices, apps, videos, and clouds is renewing interest in data pricing as a congestion-management tool. The survey connects past static and dynamic proposals with current ISP practices and future research challenges.
- Motivation: Pricing innovation is increasingly relevant because network congestion has worsened and operators are pursuing new schemes.Regulators have also recognized providers’ need for flexibility to manage congestion and experiment with business models.
- Motivation: Rapid capacity demand from apps and clouds has made pricing policy a central question for network-resource management.The paper asks how pricing policies will change as capacity demand rises during the decade.
- Scope: The survey reviews well-known pricing proposals from the last two decades, including static and dynamic approaches, and reports pricing schemes currently used by ISPs.It also identifies research challenges and emerging trends in access pricing.
- Contributions: The paper classifies past and recent data-pricing plans and illustrates implementations by ISPs in different parts of the world.It places broadband pricing alongside earlier congestion-pricing ideas from road and electricity networks.
- Contributions: The survey analyzes threats to the Internet ecosystem from ISP, consumer, and content-provider perspectives while outlining open problems for pricing innovation.Its organization covers congestion threats, network economics, pricing proposals, current realizations, emerging trends, and future directions.
- Research agenda: Comparing theoretical foundations with real-world deployments is presented as important for predicting future trends and shaping a research agenda.The paper emphasizes that researchers can overlook innovative practices already deployed by network operators.
2. THREATS TO THE INTERNET ECOSYSTEM
Rising mobile and wired traffic threatens Internet sustainability and economic viability, while data caps, overage fees, throttling, and quality reductions shift costs and constrain usage. The section examines these pressures across ISPs, consumers, and content providers.
- ISPs’ Traffic Growth: Mobile data traffic was predicted to grow at a 78% CAGR from 2011 to 2016, reaching 10.8 exabytes per month in 2016.Average smartphone consumption was projected to rise from 150 MB/month in 2011 to 2.6 GB/month by 2016.
- ISPs’ Traffic Growth: By 2016, Wi-Fi and mobile data were predicted to comprise 61% of Internet traffic, compared with 39% for wired traffic.Wired networks also face congestion from growing edge and middle-mile demand.
- ISPs’ Traffic Growth: ISPs have responded to traffic growth with measures including monthly caps and throttling, such as Comcast’s 300 GB wired-user cap.Cisco projected 40.5 petabytes of fixed Internet video per month by 2016.
- Consumers’ Cost Increase: Consumers face higher subscription costs and overage charges, prompting use of tracking and compression tools to remain within data caps.The passage describes these concerns in both U.S. and South African pricing contexts.
- Pricing and ecosystem risks: Tiered plans, usage-based overage fees, and bundled offerings have renewed debate about net neutrality, openness, price discrimination, and paid prioritization.The concerns include possible anticompetitive behavior in access-plus-content bundles.
- Content Providers’ Worries: Content providers are responding to users’ bill concerns by offering lower-quality streaming or Wi-Fi-only access, which can reduce data usage and quality of experience.The paper characterizes this response as penalizing demand and lessening consumption through content-quality degradation.
- Research responses: Research responses include opportunistic delivery using unused bandwidth and budget-aware video adaptation to sustain quality within users’ quotas or budgets.QAVA predicts usage and leverages video compressibility to keep users within available data allowances.
3. PRICING DATA NETWORKS
Access pricing can pursue competing network goals, requiring trade-offs among efficiency, fairness, reliability, manageability, complexity, and implementability. The survey reviews economic models and static or dynamic pricing strategies, then identifies practical research challenges in deploying them.
- Pricing goals and trade-offs: Access pricing can balance functionality goals including efficiency, fairness, reliability, manageability, complexity, and implementability, which may conflict.Pricing mechanisms are presented as tools for managing trade-offs among these requirements.
- Network economics fundamentals: The survey uses a single-link marketplace model in which users demand bandwidth and the ISP supplies capacity at a chosen price.Users maximize utility to determine individual and aggregate demand, while the ISP maximizes profit from revenue minus link-building cost.
- Static pricing: Static pricing keeps rates fixed over relatively long periods, providing predictable bills but potentially concentrating traffic in fixed peak and discounted off-peak periods.The cited discussion identifies tiered and usage-based plans as examples and notes that fixed peak hours can create separate traffic peaks.
- Dynamic pricing: Dynamic pricing adjusts rates at finer timescales according to congestion, including packet auctions, congestion-dependent prices, and day-ahead prices.Day-ahead pricing provides advance notice to mitigate the inconvenience of prices that fluctuate with current network load.
- Research challenges: Broadband pricing raises open questions about real-time versus offline computation, scalability, and consumer convenience when prices change rapidly.Dynamic systems may require automated client reactions, scalable ISP monitoring, and price-computation infrastructure.
4. RESEARCH CHALLENGES IN ACCESS PRICING
Access-pricing research must address implementability alongside economic efficiency, including scalability, privacy, platform support, user communication, and consumer acceptance. The survey highlights feedback-control designs, aggregate measurements, secure data handling, and user-oriented interfaces as important directions.
- Research on smart data pricing increasingly evaluates whether proposed models can be implemented and work effectively through consumer surveys and field trials.
- Price Computation: Optimal prices require models of ISP provisioning and congestion costs, user delay tolerance, deferral behavior, and price elasticity across traffic classes.
- Communication with Users: Pricing systems must communicate prices and usage, support shared quotas across devices, and accommodate automation or application interfaces according to pricing dynamism.
- Scalability and Functionality Separation: Scalable dynamic pricing uses feedback between ISP servers and users while computing future prices from aggregate congestion rather than monitoring every customer.
- Consumer Privacy and Security: Dynamic pricing must secure recorded, stored, and transferred user information; aggregate-load prices can reduce both privacy exposure and scalability burdens.
- Mobile Platform and Network Support: Mobile-platform restrictions can block application-level usage measurement and background pricing functions, whereas Android and Windows permit such features.
- User Convenience: Consumers generally prefer simple flat rates and static prices, while education, usable interfaces, and precedents from electricity and roads may support more dynamic plans.
- User Budgets: Psychological traits correlated with incentive acceptance or non-compliance were more significant than application type in one study, whose network-efficiency effects remained unexplored.
5. STATIC PRICING
Static pricing includes unlimited, capped, metered, tiered, and time-based plans that trade simplicity and predictable billing against inefficient allocation and congestion risks. Real-world experiments illustrate both the appeal and limits of flat-rate designs.
- Flat-rate broadband access historically charged a fixed monthly fee regardless of time spent online or data consumed.
- Flat-rate plans vary from unlimited service to capped plans, with metered overage charges or different flat prices for different usage caps.
- A TelstraClear free-weekend experiment removed usage metering and data caps, exposing the practical congestion risks of unpenalized flat-rate access.
- Time-based mobile plans specify caps in hours rather than data volume, as illustrated by Mobinil packages of 30 hours for EGP 80/month or 60 hours for EGP 125/month.
- Flat-rate billing is inexpensive, predictable, and demand-stimulating, but can make low-usage customers subsidize heavy users and allocate resources inefficiently.
5.2. Usage-Based Pricing
Usage-based pricing charges users according to traffic volume, commonly through a flat allowance followed by metered overage. Related differentiated-service proposals use self-selection or priority to connect payment with congestion and quality.
- Usage-Based Pricing: New Zealand’s 1989 university experiment charged traffic volume through the NZGate gateway, reflecting expensive trans-Pacific links and limited government subsidies.
- Usage-Based Pricing: Metered pricing charges users in proportion to actual data volume, while cap-then-metered plans apply a flat charge up to a threshold and per-volume charges thereafter.
- Paris Metro Pricing: Paris Metro Pricing partitions resources into logically separate classes with identical packet treatment but different prices, allowing higher-paying users to select less congested classes.
- Paris Metro Pricing: PMP implementation requires setting prices and capacities, maintaining predictable class performance, and improving interfaces for changing preferences and assigning sessions.
- Paris Metro Pricing: The same self-selection principle appears in less congested HOV lanes, which are restricted to vehicles carrying multiple passengers.
5.4. Token Pricing
Token and priority pricing let users reserve better service for urgent traffic through tokens or higher per-byte charges. The surveyed analyses show efficiency or equilibrium properties, but also identify important assumptions and user-control limits.
- Token Pricing: Token pricing combines a fixed monthly fee with two service classes: a congestible basic class and a less congested class requiring redeemed tokens.
- Token Pricing: Users self-prioritize urgent peak-congestion sessions by spending tokens for higher quality, creating a PMP-like differentiation effect.
- Priority Pricing: Priority classes charge higher per-byte fees for better service, and quality-sensitive pricing was shown more Pareto-efficient than flat pricing under a reservation-less assumption.
- Priority Pricing: Static priority pricing admits a Wardrop equilibrium bandwidth allocation, but the equilibrium need not be unique and does not depend on class prices.
- Priority Pricing: Priority pricing limits users’ ability to express desired delay and bandwidth shares and may leave lower-priority classes with little or no usage.
5.6. Reservation-Based Pricing
Reservation-based pricing charges users according to reserved service characteristics or resource capacity, with schemes addressing utilization, revenue, blocking, and service differentiation. These approaches also create fairness, affordability, and efficiency tradeoffs.
- Reservation pricing evaluates network utilization, ISP revenue, and blocking probability under per-packet, setup, and peak-load charges.
- Flat-rate setup costs disadvantage short conversations and may make connections unaffordable for poorer users.
- Setup costs can lower average network utilization, creating a tradeoff between network efficiency and revenue maximization.
- Users may be charged for service class, bandwidth, buffer space, CPU time, reserved resources, channel duration, and time of day.
- Resource-capacity pricing separates reservation and transport costs for real-time and non-real-time traffic and models capacity needs from buffer, processing, and schedulability requirements.
5.7. Time-of-Day Pricing
Time-of-day pricing varies rates between peak and off-peak periods to spread demand over time. Proposed and practical schemes range from reservation-based models to unlimited access during selected hours, while expected-capacity pricing lets users specify service expectations.
- Time-of-day pricing charges different rates during peak and off-peak hours to disperse demand more uniformly over time.
- Peak-load pricing reduces peak utilization and blocking probability across traffic classes while increasing revenue.
- Practical two-period plans use different daytime and nighttime rates, including BSNL’s unlimited 2–8 am downloads on qualifying monthly plans.
- Clark’s expected capacity pricing allows users to specify service expectations such as file-transfer time while accounting for data volume and delay tolerance.
- Traffic flagging marks packets inside or outside a purchased profile, with out-of-profile packets preferentially dropped during congestion.
5.9. Cumulus Pricing
Cumulus pricing combines an initial contract, usage monitoring, and later negotiation. Feedback on actual usage helps users reassess whether their selected plan matches their resource requirements.
- Cumulus pricing has specification, monitoring, and negotiation stages based on estimated and observed resource requirements.
- The provider initially offers a flat-rate contract, monitors actual usage, and reports cumulus points indicating whether requirements were exceeded.
- Vodafone’s U.K. data test drive provides unlimited data access for three months before feeding usage reports into plan renegotiation.
- After the trial, users can retain their plan with possible overage charges or switch to Vodafone’s suggested alternative.
5.10. Application- and Content-Based Pricing
Application- and content-based pricing differentiates data treatment by application type or sponsored content. Current plans often bundle access to selected services, while zero-rating subsidizes particular applications.
- Mobile providers are experimenting with pricing that varies by application or content type, analogous to vehicle-based toll differences.
- App-based plans commonly bundle music or movie streaming with data access and are offered by operators in the U.K., Canada, and Denmark.
- Sponsored content subsidizes selected data, such as Mobistar’s zero-rated access to Facebook and Twitter.
- Deep Packet Inspection enables ISPs to differentiate traffic from some applications and offer specialized data plans.
6. DYNAMIC PRICING
Dynamic pricing varies offered prices over time, often in response to congestion, to improve network management. The survey reviews several dynamic mechanisms and highlights adoption barriers and implementation trade-offs.
- Dynamic pricing adjusts prices at a finer timescale in response to network congestion, unlike time-of-day pricing’s fixed peak and off-peak periods.
- Raffle-based pricing uses probabilistic rewards to encourage users to shift demand from peak to off-peak periods.The winning probability can be proportional to each user’s contribution to reducing peak demand.
- An 83% acceptance rate was observed in a cellular trial offering probabilistic rewards for terminating usage during a 10-minute interval.Participants were largely university students using loaned devices, limiting generalizability.
- Effective bandwidth pricing charges users using self-reported peak and mean rates together with observed mean rate and connection duration.The tariff is based on the tangent to an effective-bandwidth formula at the observed mean rate, multiplied by connection duration.
- Users minimize expected cost by accurately reporting connection mean and peak rates, producing duration- and volume-proportional charges.Users may renegotiate the tariff for a flat fee when traffic is highly variable.
- Responsive pricing can use closed-loop feedback or Smart Markets, while game-theoretic pricing can maximize welfare and provider revenue when prices vary across users and times.Game-theoretic models have found little traction among operators because of stylized assumptions and difficulties estimating user utilities and system parameters.
7. EMERGING TECHNOLOGY TRENDS
Emerging pricing trends address growing demand through new architectures, heterogeneous-network choices, shared data plans, and differentiated or sponsored access. The survey also emphasizes architectural design and user-facing controls as implementation concerns.
- Exponential data-demand growth is catalyzing new directions in broadband pricing research and practice.
- Satellite broadband pricing remains fairly simple because satellite networks have not yet experienced serious congestion, though future demand may enable new schemes.
- Heterogeneous-network pricing can steer users among available technologies to improve resource allocation, while newer models also consider provider profit and revenue.
- Pricing architecture must address user acceptability, scalability, privacy, and security through separated client- and server-side functionality.Client-side modules can show consumption and spending, while server-side components support monitoring and price computation.
- Usage-monitoring applications increase awareness and can let users control application bandwidth, but their interfaces must be designed to sustain engagement.
- Shared data plans allow multiple devices to use one data cap; Orange reported that 38% of its iPad owners subscribed to such a plan.The survey identifies field trials as an open need for understanding quota-sharing trade-offs.
- Differentiated, zero-rated, bundled, and sponsored-content plans are emerging ways to attract customers or subsidize access through content providers and advertisers.
8. CONCLUSIONS
The survey frames pricing as an increasingly necessary response to demand growth and network congestion, then synthesizes existing proposals, current ISP adoption, and requirements for future innovation.
- Projected mobile and video demand may exceed the capacity of advances such as 4G/LTE and WiFi offloading because of costly backhauling and wired congestion.
- The survey reviews static and dynamic data-pricing proposals while examining their adoption by Internet service providers.
- Pricing can help alleviate network congestion and balance the functionality goals of data networks.
- Innovative schemes such as day-ahead time- and usage-dependent pricing and app-based pricing require new architecture, field trials, and interdisciplinary research.
- The paper aims to inform researchers about existing access-pricing work, ongoing pricing-plan developments, and challenges for a smart-data-pricing research agenda.