Debates over paid peering and usage fees have expanded from the United States to Europe and South Korea. A key part of the debate concerns whether the ISP sets the peering price based on costs incurred by the large video service providers or if it instead reflects the ISP's market power and monopolistic control over end-users. This question gains significance in light of the recent Federal Communications Commission (FCC) decision to reconsider the reinstatement of net neutrality rules. In this paper, we determine the peering price that maximizes an ISP's profit using a two-sided market model in which a profit-maximizing ISP determines broadband prices and the peering price, and in which content providers determine their service prices based on the peering price. Our findings reveal that ISPs, when driven by profit-maximization motives, tend to set a peering fee that is sub-stantially above cost, often reaching the upper limit of what content providers are willing to pay. Additionally, our model sheds light on the influence of various factors, such as the number of interconnection points and the level of traffic localization, on the determination of an ISP's proflt-maximizing peering price.

Children's and adolescents' online data privacy are regulated by laws such as the Children's Online Privacy Protection Act (COPPA) and the California Consumer Privacy Act (CCPA). Online services that are directed towards general audiences (i.e., including children, adolescents, and adults) must comply with these laws. In this paper, first, we present DiffAudit, a platform-agnostic privacy auditing methodology for general audience services. DiffAudit performs differential analysis of network traffic data flows to compare data processing practices (i) between child, adolescent, and adult users and (ii) before and after consent is given and user age is disclosed. We also present a data type classification method that utilizes GPT-4 and our data type ontology based on COPPA and CCPA, allowing us to identify considerably more data types than prior work. Second, we apply DiffAudit to a set of popular general audience mobile and web services and observe a rich set of behaviors extracted from over 440K outgoing requests, containing 3,968 unique data types we extracted and classified. We reveal problematic data processing practices prior to consent and age disclosure, lack of differentiation between age-specific data flows, inconsistent privacy policy disclosures, and sharing of linkable data with third parties, including advertising and tracking services.

Disagreements over peering fees have risen to the level of potential government regulation. ISPs assert that content providers should pay them based on the volume of downstream traffic. Transit providers and content providers assert that consumers have already paid ISPs to transmit the content they request and that peering agreements should be settlement-free. Our goal is to determine the fair payment between an ISP and an interconnecting network. We consider fair cost sharing between two Tier-1 ISPs, and derive the peering fee that equalizes their net backbone transportation costs. We then consider fair cost sharing between an ISP and a transit provider. We derive the peering fee that equalizes their net backbone transportation costs, and illustrate how it depends on the traffic ratio and the amount of localization of that content. Finally, we consider the fair peering fee between an ISP and a content provider. We derive the peering fee that results in the same net cost to the ISP, and illustrate how the peering fee depends on the number of interconnection points and the amount of localization of that content. We dispense with the ISP argument that it should be paid regardless of the amount of localization of content.

Large Internet Service Providers (ISPs) often require that peers meet certain requirements to be eligible for free-settlement peering. The conventional wisdom is that these requirements are related to the perception of roughly equal value from the peering arrangement, but the academic literature has not yet established such a relationship. The focus of this paper is to relate the settlement-free peering requirements between two large ISPs and understand the degree to which the settlement-free peering requirements between them should apply to the peering between large ISPs and content providers. We analyze settlement-free peering requirements about the number and location of interconnection points (IXPs). Large ISPs often require interconnection at a minimum of 6 to 8 interconnection points. We find that the ISP's traffic-sensitive cost is decreasing and convex with the number of interconnection points. We also observe that there may be little value in requiring interconnection at more than 8 IXPs. We then analyze the interconnection between a large content provider and an ISP. We show that it is rational for an ISP to agree to settlement-free peering if the content provider agrees to interconnect at a specified minimum number of interconnection points and to deliver a specified minimum proportion of traffic locally.

Disagreements between Internet Service Providers (ISPs) and content providers over peering fees have risen to the level of potential government regulation. ISPs assert that content providers should pay peering fees based on the volume of downstream traffic. Content providers assert that consumers pay ISPs to transmit the content they request, and thus peering agreements should be settlement-free. We determine the fair peering fee between an ISP and a transit provider or content provider. We first consider cost sharing between an ISP and a transit provider. We derive the peering fee that equalizes their net backbone transportation costs. We illustrate how the peering fee depends on the traffic ratio and the amount of localization of that content. We then derive the peering fee between an ISP and a content provider that results in the same net cost to the ISP, and illustrate how the peering fee depends on the number of interconnection points and the amount of localization. We use these results to dispense with the ISP argument that they should be paid regardless of the amount of localization of content.

Peering between two networks may be either settlement-free or paid. In order to qualify for settlement-free peering, large Internet Service Providers (ISPs) require that peers meet certain requirements. However, the academic literature has not yet shown the relationship between these settlement-free peering requirements and the value to each interconnecting network. We develop two models to analyze the value to each network from the most common and important requirements in the United States. Large ISPs in the U.S. often require potential settlement-free peers to interconnect at a minimum of 6–8 locations. We find that there is a substantial benefit from this requirement to the ISP, but little incremental benefit from a larger number of interconnection points. Large ISPs often require that the ratio of incoming traffic to outgoing traffic remain below approximately 2:1. In the case of two interconnecting ISPs, we find that this requirement ensures a roughly equal exchange of value. We also show that it is rational for an ISP to agree to settlement-free peering if the content provider agrees to interconnect at a specified minimum number of interconnection points and to deliver a specified minimum proportion of traffic locally, but a limit on the traffic ratio is irrational.

Peering between two networks may be either settlement-free or paid. In order to qualify for settlement-free peering, large Internet Service Providers (ISPs) require that peers meet certain requirements. It is widely perceived that these requirements represent the conditions under which the two peering networks perceive a roughly equal exchange of value. However, the academic literature has not yet shown the relationship between these settlement-free peering requirements and the value to each interconnecting network. We analyze the value to each network from the most common and important requirements. Large ISPs often require potential settlement-free peers to interconnect at a minimum of 6–8 locations from a predetermined list. We find that there is a substantial benefit from this requirement to the ISP, but little incremental benefit from a larger number of interconnection points. Large ISPs often require that the ratio of incoming traffic to outgoing traffic remain below approximately 2:1. We find that this requirement ensures a roughly equal exchange of value.

Internet Service Provider design of service tiers are modeled and analyzed, based on demand for web browsing and video streaming. A basic model that considers user willingness to pay, network capacity, and application performance is formulated to determine when multiple tiers maximize profit. An extended model that also considers the time that users devote to each application is formulated to determine the optimal network capacity, tier rates, and tier prices. We show that an Internet Service Provider may simplify tier and capacity design by allowing its engineering department to set network capacity, its marketing department to set tier prices, and both to jointly set tier rates. Numerical results are presented to illustrate the magnitude of the decrease in profit compared to the optimal profit resulting from such a simplified design.

We are concerned with whether a vertically integrated broadband and content provider can unreasonably advantage itself over competing content providers, either by selling quality-of-service (QoS) to content providers at unreasonably high prices, or by refusing to provide access to QoS to competing content. We address this question by modeling the competition between one such vertically integrated provider and one over-the-top (OTT) content provider. We analytically determine when the broadband provider will sell QoS and when the OTT content provider or users will purchase QoS. We characterize the optimal QoS and video service prices. The ISP's market share increases with the difference in the value of the two video services and decreases with the difference in the corresponding costs. Numerical results illustrate the effect of QoS price on content price, and the variation of market share and profit with QoS price. The ISP may sell QoS to users at a lower price than when QoS is sold to the OTT provider.

We model the design and impact of Internet pricing plans with data caps. We consider a monopoly ISP that maximizes its profit by setting tier prices, tier rates, network capacity, data caps, and overage charges. We show that when data caps are used to maximize profit, a monopoly ISP will keep the basic tier price the same, increase the premium tier rate, and decrease the premium tier price and the basic tier rate. We give analytical and numerical results to illustrate the increase in ISP profit, and the corresponding changes in user tier choices, user surplus, and social welfare.

In the academic literature on cellular network design, resource allocation algorithms often attempt to maximize total utility or throughput over a short time period, and connection access control often admits arrivals if and only if there are sufficient resources. In this paper, we investigate how connection access control and resource allocation can be coordinated to jointly achieve maximum total utility. We propose a decomposition in which resource allocation maximizes long-term average utility for each system state, and connection access control maximizes long-term average utility over all system states. We discuss the resulting interface and give examples of algorithms that satisfy this decomposition and interface. Simulation results illustrate that the optimal connection access control policy may block applications with relatively low average utility per unit rate even when capacity is available, and that coordinated connection access control and resource allocation can outperform uncoordinated approaches.

We propose models to evaluate the impact of data caps upon subscribers and Internet Service Providers (ISPs). These models incorporate the time users devote to Internet applications and the opportunity cost of a user's free time. We consider a monopoly ISP that maximizes profit by setting tier prices, tier rates, network capacity, data caps, and overage charges. We illustrate which users are affected by data caps, and the resulting impact upon both users and ISPs. We show that an ISP will increase the tier rate and decrease the tier price when data caps are used to maximize profit. We give numerical results to illustrate the increase in ISP profit when data caps are used, and the corresponding changes in user and social welfare.

We investigate how connection access control and resource allocation in cellular systems can be coordinated to jointly achieve maximum total utility. We propose a decomposition in which resource allocation maximizes long-term average utility for each system state and connection access control maximizes long-term average utility over all system states. We discuss the resulting interface and gives examples of algorithms that satisfy this decomposition and interface. Performance is illustrated by numerical results.

Internet research often assumes users may connect devices without consent by their service providers. However, in many networks the service provider only allows use of devices obtained directly from the provider. We review how United States communications law addresses the rights of users to connect devices of their choice. We explicate a set of user and service provider rights. We propose legal requirements for attachment and management of devices. We illustrate how these proposed regulations would affect the services currently offered on telephone, cable, satellite, video networks, and cellular networks, as well as on the Internet.

Much of the research on an Internet of Things assumes that users will be able to connect devices without consent by or interference from their service providers. However, in cable and satellite television networks, cellular networks, and some broadband Internet networks, the service provider often only allows use of set-top boxes, smart phones, and residential gateways obtained directly from the provider. The ability of a provider to implement such restrictions is limited by communications law. We review how communications law in the United States currently addresses the right of a user to connect devices of their choice. We propose a set of user and service provider rights that should guide development of new legal principles. We identify the pertinent network architectural principles, and propose concepts that can be used to create a new legal framework for device attachment to telephone networks, video networks, cellular networks, and the Internet. We illustrate how this legal framework, combined with standardized interfaces and protocols, can ensure an open network that supports innovation in devices.

Resource allocation for elastic applications is often accomplished using concave utility functions. We consider the more challenging case of resource allocation for semi-elastic applications modeled using sigmoid utility functions. Whereas elastic applications only require rate scheduling, we find that semi-elastic applications require both rate scheduling and connection access control (CAC). We pose a joint rate scheduling and CAC optimization problem, and find that direct solution is computationally intractable. We formulate the dual problem, and use this to propose a near-optimal algorithm that iteratively finds optimal shadow prices for power and rates. We then further reduce computational complexity by proposing a heuristic search algorithm that decomposes the subcarrier allocation and rate scheduling problems. The heuristic has a computational complexity linear in the number of users, and is shown by simulation to produce near-optimal results.

We examine whether wireless ISPs should be able to legally limit the applications used on wireless devices. The analysis is based on wireless network architecture and communications law. We review how wired and wireless networks differ in traffic management, and conclude that wireless networks require stronger traffic management than wired networks at and below the network layer. We review communications law, and conclude that ISPs should be prohibited from giving themselves an unfair competitive edge. We examine three scenarios of how applications may be restricted, and find that none guarantees a level playing field between ISPs and application providers while allowing ISPs to reasonably manage network resources. We propose regulations that limit an ISP's ability to restrict applications by requiring an open interface between network and transport layers. We illustrate how ISPs may deploy QoS within such a regulatory framework, and how this proposed policy can achieve the goals.

Many wireless ISPs limit the applications that may be used on wireless devices. In the United States, Congress is debating whether wireless network subscribers should have the right to use applications of their choice. We examine whether wireless ISPs should be able to limit applications. We address how wired and wireless networks differ with respect to traffic management, and conclude that wireless networks often require stronger traffic management than wired networks at and below the network layer. We propose dual goals of providing a level playing field between applications offered by ISPs and those offered by competing application providers and guaranteeing wireless ISPs the ability to reasonably manage wireless network resources. We consider three scenarios for how applications may be restricted on wireless networks, and find that none achieves both goals. We review United States communications law, and conclude that ISPs should be prohibited from giving themselves an unfair competitive edge by blocking applications or by denying QoS to competing application providers. We propose a set of regulations based on network architecture and communication law that limits an ISP's ability to restrict applications by requiring an open interface between network and transport layers. We illustrate how ISPs may deploy QoS within such a regulatory framework, and how this proposed policy can achieve our goals.

We consider complete partitioning of bandwidth among multiple services. When class bandwidth is an integer multiple of the next lower class and total bandwidth is an integer multiple of the largest class bandwidth, we develop a recursive algorithm that determines the optimal complete partitioning policy with a significantly lower complexity than that of known dynamic programming or mixed integer programming approaches.

Traffic management practices of ISPs are an issue of public concern. We propose a framework for classification of traffic management practices as reasonable or unreasonable. We present a survey of traffic management techniques and examples of how these techniques are used by ISPs. We suggest that whether a traffic management practice is reasonable rests on the answers to four questions regarding the techniques and practices used. We propose a framework that classifies techniques as unreasonable if they are unreasonably anticompetitive, cause undue harm to consumers, or unreasonably impair free speech. We propose alternatives to unreasonable or borderline congestion management practices.

As part of the wider debate over net neutrality, traffic management practices of Internet Service Providers have become an issue of public concern. The Federal Communications Commission has asked for public input on whether deep packet inspection and other traffic management practices are reasonable forms of network management. Little attention has been paid to this issue within the academic networking community, and most Internet policy researchers have recommended a case-by-case analysis. This paper proposes a framework for the classification of traffic management practices as reasonable or unreasonable. I suggest that whether a traffic management practice is reasonable largely rests on the answers to four questions regarding the techniques and practices used. Based on these questions, I propose a framework that classifies techniques as unreasonable if they are unreasonably anti-competitive, cause undue harm to consumers, or unreasonably impair free speech. The paper concludes with a brief discussion of how some unreasonable traffic management practices can be more directly and transparently accomplished through alternative practices that would be classified as reasonable.

As part of the wider debate over net neutrality, traffic management practices of Internet Service Providers have become an issue of public concern. The Federal Communications Commission has asked for public input on whether deep packet inspection and other traffic management practices are reasonable forms of network management. Little attention has been paid to this issue within the academic networking community, and most Internet policy researchers have recommended a case-by-case analysis. This paper proposes four questions that can be used to determine whether a traffic management practice is reasonable or unreasonable.

Net neutrality represents the idea that Internet users are entitled to service that does not discriminate on the basis of source, destination, or ownership of Internet traffic. The United States Congress is considering legislation on net neutrality, and debate over the issue has generated intense lobbying. Congressional action will substantially affect the evolution of the Internet and of future Internet research. In this article, we argue that neither the pro nor anti net neutrality positions are consistent with the philosophy of Internet architecture. We develop a net neutrality policy founded on a segmentation of Internet services into infrastructure services and application services, based on the Internet's layered architecture. Our net neutrality policy restricts an Internet service Provider's ability to engage in anticompetitive behavior while simultaneously ensuring that it can use desirable forms of network management. We illustrate the effect of this policy by discussing acceptable and unacceptable uses of network management.

We consider the problem of scheduling multiple transmissions on the downlink of a wireless network with performance guarantees in the form of the probabilities that short term throughputs exceed user specified thresholds. Many interactive data applications have some degree of a latency requirement, and measure performance by throughput over a relatively short time interval. We refer to the fraction of time such user throughput reaches a predefined rate threshold or higher as tail probability. The problem is formulated as maximizing the minimum ratio of tail probability to the user specified probability threshold. We present necessary and sufficient optimality conditions for the case in which the time interval of interest is consistent with the time scale of channel variation. An online algorithm is proposed which can achieve the optimality. For the case in which the time interval of interest is large compared to the time scale of channel variation, we develop an online algorithm which attempts to maximize the minimum normalized tail probability by taking the advantage of channel variation over users and over time. Simulation results demonstrate that the proposed algorithm can achieve better performance than other algorithms such as the proportional fair algorithm and the Max C/I algorithm.

We consider the problem of scheduling multiple transmissions on the downlink of a wireless network with probabilistic guarantees on users' short-term average throughputs. Many interactive data applications have some degree of a latency requirement, and measure performance by average throughput over a relatively short time interval. We refer to the fraction of time that the transmission rate of a user reaches a predefined rate threshold or higher as tail probability. The problem is formulated as minimizing the total resource cost while meeting users' respective requirements on the tail probability. We present the necessary and sufficient optimality conditions, and propose an online algorithm which can achieve the optimal resource allocation. Simulation results illustrate the convergence properties of the algorithm, the effect of rate thresholds and tail probability thresholds, and the feasible region of system resources for given performance requirements.

We consider resource allocation for elastic wireless applications that measure utility by target connection average throughput and achieved throughput. We construct a framework for connection access control and rate scheduling that supports this class of applications by maximizing long term average utility. We present a decomposition of the problem into connection access control and rate scheduling layers. The connection access control layer considers current commitments and decides whether to admit new sessions. For admitted users, the layer sets a target throughput to be achieved of the lifetime of the session. The rate scheduling layer adjusts the instantaneous rates of each connection on the basis of how well it is achieving each performance goal and on the relative strength of each connectionpsilas current channel. We illustrate how commonly used connection access control and rate scheduling techniques can be applied to design these two layers using an exchange of information regarding future target throughputs and achieved throughputs. Through numerical analysis, we show how wireless channel time diversity and multi-user diversity can be exploited to construct a rate scheduling algorithm that is superior to proportional fairness. Finally, this utility-based framework is compared to a method that does not use utility.

We consider the problem of providing guarantees on throughput and rate variation for interactive users in the downlink of CDMA systems, while minimizing the peak total transmit power, by using joint power and rate control. Under the assumption of independent and identically distributed channels, we derive that the optimal power allocation policy follows an increasing and leveling off pattern. We also develop an efficient method from the active set method, which is used in quadratic programming, to solve our problem. The computational complexity depends on the number of users and the number of joint channel states. Numerical results illustrate the sensitivity of the peak power to the required mean rate, the maximum allowed rate variation, and the number of users.

We consider the allocation of power across forward-link packets in a wireless data network. The packets arrive according to a random (Poisson) process, and have fixed length so that the data rate for a given packet is determined by the assigned power and the channel gain to the designated user. Each user's service preferences are specified by a utility function that depends on the received data rate. The objective is to determine a power assignment policy that maximizes the time-averaged utility rate, subject to a constraint on the probability that the total power exceeds a limit (corresponding to an outage). For a large, heavily loaded network, we introduce a Gaussian approximation for the total transmitted power, which is used to decompose the power constraint into three more tractable constraints. We present a solution to the modified optimization problem that is a combination of admission control and pricing. The optimal trade-off between these approaches is characterized. Numerical examples illustrate the achievable utility rate and power allocation as a function of the packet arrival rate.

Emerging WCDMA data services will likely require resource allocation to ensure that throughput targets are met. Scheduling of power and rate among competing data streams is a key component to any such resource allocation strategy. We believe a scheduler should be both efficient and fair. We propose here a scheduler that takes advantage of variations in the wireless channel both channel fluctuations in time for each user, and channel variations among multiple users at a particular time. By mixing a max-min policy with a policy of serving users with relatively good channels, the scheduler can achieve individual average throughput targets in a manner that encourages system efficiency. We analyze the proposed scheduler using numerical examples, which demonstrate the superiority in both efficiency and fairness over proportional fair scheduling. We also analyze the limiting effects of system power and rate supply

Differentiated services can ensure that traffic on some codepoints receives higher quality of service (QoS) than traffic on other codepoints, but without additional mechanisms it cannot target any particular QoS. Congestion-based pricing has been suggested as a method to target QoS in other network architectures. Here, we investigate whether congestion-based pricing can be used to control aggregate traffic into each codepoint by motivating users to choose the codepoints appropriate for each application. We focus on what information needs to be exchanged in a distributed fashion and assert that both price and QoS information must be available for users to make decisions.

We consider pricing of network resources in a reservation-based quality-of-service architecture. The pricing policy implements a distributed resource allocation to provide guaranteed bounds on packet loss and end-to-end delay for real-time applications. Distributed pricing roles are assigned to each user, each network node, and an arbitrager in between the user and the network. When delay constraints are not binding, we investigate two dynamic pricing algorithms using gradient projection and Newton's method to update prices, and prove their convergence. We analyze the performance of the dynamic pricing policies and show that the gradient algorithm using Newton's method converges more quickly and displays only a few small fluctuations. When delay constraints are binding, we investigate subgradient methods which can provide convergence to some range of the optimal allocation.

Power allocation across users in two adjacent cells is studied for a code-division multiple access (CDMA) data service. The forward link is considered and cells are modeled as one-dimensional with uniformly distributed users and orthogonal signatures within each cell. Each user is assumed to have a utility function that describes the user's received utility, or willingness to pay, for a received signal-to-interference-plus-noise ratio (SINR). The objective is to allocate the transmitted power to maximize the total utility summed over all users subject to power constraints in each cell. It is first shown that this optimization can be achieved by a pricing scheme in which each base station announces a price per unit transmitted power to the users, and each user requests power to maximize individual surplus (utility minus cost). Setting prices to maximize total revenue over both cells is also considered, and it is shown that, in general, the solution is different from the one obtained by maximizing total utility. Conditions are given for which independent optimization in each cell, which leads to a Nash equilibrium (NE), is globally optimal. It is shown that, in general, coordination between the two cells is needed to achieve the maximum utility or revenue.

We consider a single node which multiplexes a large number of traffic sources. We ask a simple question: how do the optimal allocations of bandwidth and buffer vary with the number of sources? We investigate this issue using previous results on the probability of overflow for an aggregate of i.i.d. flows, e.g., overflow resulting from effective bandwidth models. We wish to determine the variation of the minimum cost allocations of bandwidth and buffer with the number of sources, given a cost per unit of each resource. We first consider a class of on/off fluid flows. We find that the optimal bandwidth allocation above the mean rate and the optimal buffer allocation are both proportional to the square root of the number of sources. Correspondingly, we find that the excess cost incurred by a fixed buffer allocation or by linear buffer allocations is proportional to the square of the percentage difference between the assumed number of sources and the actual number of sources and to the square root of the number of sources. We next consider a class of general i.i.d. sources for which the aggregate effective bandwidth is a decreasing convex function of buffer and linearly proportional to the number of sources. We find that the optimal buffer allocation is strictly increasing with the number of sources. Correspondingly, we find that the excess cost incurred by a fixed buffer allocation is an increasing convex function of the difference between the assumed number of sources and the actual number of sources.

We consider forward link power allocation for voice users in a code-division multiple-access wireless network. Admission control policies are investigated, which base a new call admission decision not only upon available capacity, but also upon the required forward link transmit power and upon the user's willingness to pay. We assume that each voice user has a utility function that describes the user's willingness to pay as a function of forward link signal-to-interference plus noise ratio. The network objective is to maximize either total utility summed over all users or total revenue generated from all users. Properties of the optimal power and code allocations are presented. Our key results show how these optimal allocations can be achieved using pricing. The analysis is complemented with a numerical study, which shows how the optimal prices and corresponding utility or revenue vary with load.

Power and code allocation across two adjacent cells is studied for the downlink of a code-division multiple-access voice network. Each user has a utility function that measures the user's willingness to pay, or utility, as a function of the received signal-to-interference-plus-noise-ratio. The objective is to maximize the total utility over the two cells subject to code and power constraints. When all active users receive the same utility, the optimal allocation is characterized by a pair of threshold radii for the two cells, where each radius specifies the set of active users in that cell. The behavior of the optimal radii are characterized as a function of load and available resources (power and codes). The corresponding optimal power allocation can be achieved through a pricing scheme, in which each base station announces a price for each resource, and each user responds by requesting the amount of resources that maximizes the user's surplus (utility minus cost). We show that, depending on the load and resource constraints, the two cells may have to coordinate, or exchange information, in order to maximize the total utility.

Differentiated services and integrated services architectures rely on mechanisms to dynamically allocate network resources. Here, we investigate the feasibility of dynamically adjusting prices based on congestion. The pricing policy implements a distributed resource allocation to provide guaranteed bounds on packet loss and end-to-end delay for real-time applications. Distributed pricing roles are assigned to each user, each network node, and an arbitrager in between the user and the network. We find that such dynamic pricing can be accomplished using simple gradient algorithms, but that this approach requires a small amount of steady communication about demand and supply along each route.

We consider resource allocation policies to guarantee bounds on packet loss and end-to-end delay for real-time applications in a network with a reservation-based QoS architecture. We consider using pricing to distribute the allocation of buffer and bandwidth at each node along a set of paths, with the goal of maximizing the total utility of all users in the network. Each user is modelled as an aggregate of flows with similar traffic characterizations and similar utility functions, and utility is assumed to be a function of loss probability, which in turn depends on the reserved buffer and bandwidth at each node. We show that this optimization problem has a unique solution, and that the corresponding shadow costs associated with each resource and delay constraint can be related to user's marginal utilities. We propose distributed pricing implementations to achieve the optimal resource reservation policy, by assigning individual roles to the users and to the network, and perhaps to an intervening arbitrager layer. We prove that an allocation is optimal if all users, arbitragers, and the network are in equilibrium.

Resource allocation is studied for a forward-link two-cell code division multiple access (CDMA) voice network with multiple service classes. System resources are transmitted power and codes. The service classes are specified by different user utility functions that relate utility to received signal-to-interference-plus-noise-ratio (SINR). The objective of the resource allocation is to maximize total utility over the two cells. The optimal power allocation is characterized by a set of distances, or radii, from the desired base station. Each radius corresponds to the set of active users in a particular service class, and can be enforced through a pricing scheme. We also consider setting prices to maximize revenue. In general, the prizes and power allocation that maximize revenue differ from those that maximize utility.

We consider forward-link power allocation in a wireless network with stochastically varying data requests. We assume a user's service preferences are specified via a utility function that depends on the received data rate. The allocation of power across users is studied, where this allocation may depend on both a user's channel and utility. The objective is to maximize the time-averaged utility rate subject to a stochastic total power constraint at the transmitter. For a large, heavily loaded network, we introduce a Gaussian approximation for the total transmitted power, which is used to decompose the power constraint into three more tractable constraints. We present a solution to this problem that is a combination of admission control and pricing of power. The optimal trade-off between these approaches is characterized. Numerical examples are given to illustrate these ideas.

Power allocation across users in two adjacent cells is studied for a wireless code division multiple access (CDMA) network with mixed voice and data services. We assume that each user has a utility function that measures the user's satisfaction, or utility, as a function of the received signal-to-interference-plus-noise ratio (SINR). Each particular service (voice or data) is associated with a different utility function. We consider the forward link. Our objective is to allocate transmitted power to maximize the total utility summed over all active users subject to rate and power constraints. We show that the maximum utility can be achieved with a pricing scheme. We characterize the solution to a one-cell utility maximization problem with fixed interference from the other cell. For two-cell utility maximization, the two cells must cooperate to achieve the maximum utility.

We examine a method to improve the service of incoming requests at a World Wide Web server. The motivating factor is the existence of an impatient user pool: a user aborts a pending Web request if a response is not received within a random timeout period. We use a queueing theory approach to derive an optimal service ordering for this server, assuming Poisson arrivals and exponential service times. We find that the optimal policy is greedy, in that at any time the server processes the request with the highest perceived payoff. We verify these results both analytically and via simulation.

We consider dynamic resource allocation (DRA) for the reverse-link of a multiuser packet DS-CDMA system with voice and data traffic. Our objective is to minimize the total received power for voice users, and allocate all remaining power to data users so as to maximize throughput. The performance of a DRA scheme depends on how frequently resources are reassigned. We compare the performance of an "aggressive" scheme, in which the processing gain (PG) is reassigned to all users each time a packet arrives or departs, with "timid" schemes, in which the PG cannot be changed in the middle of a packet ("packet-timid") or a session ("flow-timid"). For the parameters considered, our results show that aggressive and packet-timid DRA perform similarly. Flow-timid DRA performs significantly worse than the other two schemes, and the difference in performance increases with traffic intensity.

Resource allocation is studied for the forward link of a two-cell wireless direct sequence (DS)-code division multiple access (CDMA) network. A voice service is assumed, for which system resources are transmitted power and codes. Each user has a utility function that measures the user's willingness to pay as a function of the received quality of service (QoS). Our objective is to maximize the total utility over all active users. We characterize the optimal power distribution across the two cells, and show that in general, the cells must coordinate, or exchange information, to achieve the maximum utility. The optimal allocation can be achieved by a pricing scheme, in which each base station announces a price for each resource, and each user responds by buying an amount of resources that maximizes the user's surplus (utility minus cost).

This paper studies pricing as a means for resource allocation in a wireless direct-sequence (DS) code division multiple access (CDMA) system. We consider the forward link of a single cell with orthogonal codes and voice traffic. The base station announces a price per unit transmitted power and a price per code, and the users respond according to their individual utilities. The objective is to set prices to maximize either total user utility or total revenue. The solution to the former problem (maximize utility) is presented. To study the latter problem we derive the large system revenue as the number of users and codes tend to infinity with fixed ratio. The large system revenue depends on the distribution of utilities and path loss across the user population, and may not be a unimodal function of the prices. Numerical results based on a simple model for user utility show how the optimal prices and revenue vary with the offered load.

In this paper, we discuss the role of prices in combining user characterization, network resource allocation, and contract negotiation to form a complete connection establishment process. We suggest that such a process should encourage network efficiency through distributed resource allocation among virtual circuits, circuit bundles, and virtual paths. We adopt effective bandwidth as our user traffic characterization and our pricing base, and we measure network efficiency by total user benefit. We allow a limited degree of statistical multiplexing by incorporating multiplexing gain into the prices. Finally, we propose a hierarchical and distributed negotiation structure under which only hierarchically adjacent and geographically local network entities communicate with each other.

The evolving view of connection establishment for connection-oriented services in high-speed networks such as ATM involves a contract negotiation process between a user agent and a network agent. The first stage consists of separate roles for the user and the network. The user agent must characterize the information streams that will be transmitted and the performance parameters that define the desired quality of service for that user. Similarly, the network agent must determine the network's resources and its capabilities to accommodate various mixes of service types. The second stage involves negotiations between multiple network and user agents, in which the parties agree to set up connections to transmit the agreed information streams in a manner to guarantee the agreed qualities of service, and at agreed prices. We focus on these two stages that together form the connection establishment process. After this process, during the connection, the network must police the user to determine compliance with the information stream characteristics, and must implement flow control, service priority mechanisms and packet multiplexing disciplines as necessary to guarantee the quality of service.>

The merging of telephone and computer networks is introducing multiple resources into networks, and information is becoming increasingly distributed across the network. Related services are being integrated onto a single network rather than being offered on separate uncoordinated networks. We focus upon communication networks that integrate multiple services using multiple resources. In previous work, such networks have been modeled by multidimensional Markov chains with product form distributions. We approximate the distribution on the original discrete state space by a similar product form distribution on a continuous state space. We consider access control of such a system and prove that the resulting optimal coordinate-convex control policy is convex. Based on this result, we suggest an algorithm for ending a near-optimal policy for the discrete problem that has much less complexity than existing methods for finding optimal or near-optimal policies.>

We are concerned with characterizing the variation of multiplexing gains with source type and burstiness in integrated service systems such as ATM. We model a fixed capacity high speed bit pipe that multiplexes a moderate number of bit streams with minimal buffer under a low loss constraint. Each service type is defined by its instantaneous bitrate distribution, but the bitrate distribution of multiplexed streams is approximated as Gaussian. The Gaussian approximation is not as accurate as Chernoff bounds, but it allows for stronger characterization of multiplexing gains. We consider three schemes for allocating bandwidth to services: by individual user, by path and service type, and by path only. We find explicit formulae for sensitivities of required capacity to source rate mean and variance and to loss rate. We characterize multiplexing gains and costs to identify the benefits of each allocation policy. We find that the capacity savings resulting from sharing resources is proportional to the square root of the ratio of source rate variance to source rate mean. This suggests that although bursty sources require more bandwidth, multiplexing gains are increasing with burstiness. We also find that the extra capacity required to multiplex dissimilar source types is increasing with the difference between their burstinesses. This suggests that when bit streams are partially grouped, it is most important first, to group similar source types.

The merging of telephone and computer networks is introducing multiple resources into networks, and information is becoming increasingly distributed across the network. Related services are being integrated onto a single network rather than being offered on separate uncoordinated networks. The authors focus upon communication networks that integrate multiple services using multiple resources. In particular, they look at the decision of whether to accept or deny service requests in such a system. They prove a conjecture for the optimal policy for a related system introduced in Foschini and Gopinath (1983) and characterize the optimal coordinate convex policy for the present multiple service, multiple resource system.>

The authors focus upon communication networks that integrate multiple services using multiple resources. In particular, the authors address the decision of whether to accept or deny service requests in such a system. A conjecture for the optimal policy for a related system introduced by G.J. Foschini and B. Gopinath (1983) is proved, and the optimal coordinate convex policy for a multiple service, multiple resource system is characterized.>

Communication networks that integrate multiple services using multiple resources are considered. In particular, the authors pose resource allocation problems, present a sensitivity analysis, and provide a glimpse of the possible behavior of such networks. The simplest discipline is assumed: a service request is accepted if the necessary resources are available; otherwise it is rejected. Two results are obtained. The first gives the sensitivity of throughput of service requests of type i with respect to offered traffic and service rates of type j. The second result is that the set of vectors of achievable throughput rates is a convex polyhedron given by an explicit set of linear inequalities.>