Search

In order to achieve the dual goals of privacy and learning across distributed data, Federated Learning (FL) systems rely on frequent exchanges of large files (model updates) between a set of clients and the server. As such FL systems are exposed to, or indeed the cause of, congestion across a wide set of network resources. Lossy compression can be used to reduce the size of exchanged files and associated delays, at the cost of adding noise to model updates. By judiciously adapting clients' compression to varying network congestion, an FL application can reduce wall clock training time. To that end, we propose a Network Adaptive Compression (NAC-FL) policy, which dynamically varies the client's lossy compression choices to network congestion variations. We prove, under appropriate assumptions, that NAC-FL is asymptotically optimal in terms of directly minimizing the expected wall clock training time. Further, we show via simulation that NAC-FL achieves robust performance improvements with higher gains in settings with positively correlated delays across time.

Federated learning systems facilitate training of global models in settings where potentially heterogeneous data is distributed across a large number of clients. Such systems operate in settings with intermittent client availability and/or time-varying communication constraints. As a result, the global models trained by federated learning systems may be biased towards clients with higher availability. We propose F3AST, an unbiased algorithm that dynamically learns an availability-dependent client selection strategy which asymptotically minimizes the impact of client-sampling variance on the global model convergence, enhancing performance of federated learning. The proposed algorithm is tested in a variety of settings for intermittently available clients under communication constraints, and its efficacy demonstrated on synthetic data and realistically federated benchmarking experiments using CIFAR100 and Shakespeare datasets. We show up to 186% and 8% accuracy improvements over FedAvg, and 8% and 7% over FedAdam on CIFAR100 and Shakespeare, respectively.

Ridesourcing platforms recently introduced the ``schedule a ride'' service where passengers may reserve (book-ahead) a ride in advance of their trip. Reservations give platforms precise information that describes the start time and location of anticipated future trips; in turn, platforms can use this information to adjust the availability and spatial distribution of the driver supply. In this article, we propose a framework for modeling/analyzing reservations in time-varying stochastic ridesourcing systems. We consider that the driver supply is distributed over a network of geographic regions and that book-ahead rides have reach time priority over non-reserved rides. First, we propose a state-dependent admission control policy that assigns drivers to passengers; this policy ensures that the reach time service requirement would be attained for book-ahead rides. Second, given the admission control policy and reservations information in each region, we predict the ``target" number of drivers that is required (in the future) to probabilistically guarantee the reach time service requirement for stochastic non-reserved rides. Third, we propose a reactive dispatching/rebalancing mechanism that determines the adjustments to the driver supply that are needed to maintain the targets across regions. For a specific reach time quality of service, simulation results using data from Lyft rides in Manhattan exhibit how the number of idle drivers decreases with the fraction of book-ahead rides. We also observe that the non-stationary demand (ride request) rate varies significantly across time; this rapid variation further illustrates that time-dependent models are needed for operational analysis of ridesourcing systems.

Motivated by the potentially high downlink traffic demands of commuters in future autonomous vehicles, we study a network architecture where vehicles use Vehicle-to-Vehicle (V2V) links to form relay network clusters, which in turn use Vehicle-to-Infrastructure (V2I) links to connect to one or more Road Side Units (RSUs). Such cluster-based multihoming offers improved performance, e.g., in coverage and per user shared rate, but depends on the penetration of V2V-V2I capable vehicles and possible blockage, by legacy vehicles, of line of sight based V2V links, such as those based on millimeter-wave and visible light technologies. This paper provides a performance analysis of a typical vehicle's connectivity and throughput on a highway in the free-flow regime, exploring its dependence on vehicle density, sensitivity to blockages, number of lanes and heterogeneity across lanes. The results show that, even with moderate vehicle densities and penetration of V2V-V2I capable vehicles, such architectures can achieve substantial improvements in connectivity and reduction in per-user rate variability as compared to V2I based networks. The typical vehicle's performance is also shown to improve considerably in the multilane highway setting as compared to a single lane road. This paper also sheds light on how the network performance is affected when vehicles can control their relative positions, by characterizing the connectivity-throughput tradeoff faced by the clusters of vehicles.

Network slicing is a key capability for next generation mobile networks. It enables one to cost effectively customize logical networks over a shared infrastructure. A critical component of network slicing is resource allocation, which needs to ensure that slices receive the resources needed to support their mobiles/services while optimizing network efficiency. In this paper, we propose a novel approach to slice-based resource allocation named Guaranteed seRvice Efficient nETwork slicing (GREET). The underlying concept is to set up a constrained resource allocation game, where (i) slices unilaterally optimize their allocations to best meet their (dynamic) customer loads, while (ii) constraints are imposed to guarantee that, if they wish so, slices receive a pre-agreed share of the network resources. The resulting game is a variation of the well-known Fisher market, where slices are provided a budget to contend for network resources (as in a traditional Fisher market), but (unlike a Fisher market) prices are constrained for some resources to provide the desired guarantees. In this way, GREET combines the advantages of a share-based approach (high efficiency by flexible sharing) and reservation-based ones (which provide guarantees by assigning a fixed amount of resources). We characterize the Nash equilibrium, best response dynamics, and propose a practical slice strategy with provable convergence properties. Extensive simulations exhibit substantial improvements over network slicing state-of-the-art benchmarks.

Many problems in signal processing and machine learning can be formalized as weak submodular optimization tasks. For such problems, a simple greedy algorithm (\textsc{Greedy}) is guaranteed to find a solution achieving the objective with a value no worse than $1-e^{-1/c}$ of the optimal, where $c$ is the multiplicative weak-submodularity constant. Due to the high cost of querying large-scale systems, the complexity of \textsc{Greedy} becomes prohibitive in contemporary applications. In this work, we study the tradeoff between performance and complexity when one resorts to random sampling strategies to reduce the query complexity of \textsc{Greedy}. Specifically, we quantify the effect of uniform sampling strategies on \textsc{Greedy}'s performance through two metrics: (i) probability of identifying an optimal subset, and (ii) suboptimality with respect to the optimal solution. The latter implies that uniform sampling strategies with a fixed sampling size achieve a non-trivial approximation factor; however, we show that with overwhelming probability, these methods fail to find the optimal subset. Our analysis shows that the failure of uniform sampling strategies with fixed sample size can be circumvented by successively increasing the size of the search space. Building upon this insight, we propose a simple progressive stochastic greedy algorithm and study its approximation guarantees. Moreover, we demonstrate effectiveness of the proposed method in dimensionality reduction applications and feature selection tasks for clustering and object tracking.

We propose a data-driven framework to enable the modeling and optimization of human-machine interaction processes, e.g., systems aimed at assisting humans in decision-making or learning, work-load allocation, and interactive advertising. This is a challenging problem for several reasons. First, humans' behavior is hard to model or infer, as it may reflect biases, long term memory, and sensitivity to sequencing, i.e., transience and exponential complexity in the length of the interaction. Second, due to the interactive nature of such processes, the machine policy used to engage with a human may bias possible data-driven inferences. Finally, in choosing machine policies that optimize interaction rewards, one must, on the one hand, avoid being overly sensitive to error/variability in the estimated human model, and on the other, being overly deterministic/predictable which may result in poor human 'engagement' in the interaction. To meet these challenges, we propose a robust approach, based on the maximum entropy principle, which iteratively estimates human behavior and optimizes the machine policy--Alternating Entropy-Reward Ascent (AREA) algorithm. We characterize AREA, in terms of its space and time complexity and convergence. We also provide an initial validation based on synthetic data generated by an established noisy nonlinear model for human decision-making.

This paper explores an emerging wireless Internet-of-things (IoT) architecture based on unmanned aerial vehicles (UAVs). We consider a network where a fleet of UAVs at a fixed altitude flies on planned trajectories and IoT devices on the ground are scheduled to transmit their data to the UAVs when the latter are nearby. In such a system, the UAVs' motion triggers the uplink transmissions of the IoT devices. As a result, network performance is determined by the geometric and dynamic characteristics of the system. We propose a joint stationary model for UAVs and IoT devices and then evaluate the interference, the coverage probability, and the data rate of the typical UAV. To assess the harvesting capability of the proposed architecture, we derive a formula for the amount of data uploaded from each IoT device to a UAV. We also establish a linear relationship between the UAV coverage and the harvesting capability of the network, which provides insights into the design of the proposed harvesting scheme. In addition, we use our analytical results to numerically show that there exists a trade-off between the uploaded data and the size of the IoT scheduling window. Specifically, for a given UAV and IoT geometry, there exists an optimal scheduling window that maximizes the harvesting capability of the proposed network., Comment: IEEE Trans. Wireless Commun. (short version in Proc. IEEE ISIT)

In order to meet the performance/privacy requirements of future data-intensive mobile applications, e.g., self-driving cars, mobile data analytics, and AR/VR, service providers are expected to draw on shared storage/computation/connectivity resources at the network "edge". To be cost-effective, a key functional requirement for such infrastructure is enabling the sharing of heterogeneous resources amongst tenants/service providers supporting spatially varying and dynamic user demands. This paper proposes a resource allocation criterion, namely, Share Constrained Slicing (SCS), for slices allocated predefined shares of the network's resources, which extends the traditional alpha-fairness criterion, by striking a balance among inter- and intra-slice fairness vs. overall efficiency. We show that SCS has several desirable properties including slice-level protection, envyfreeness, and load driven elasticity. In practice, mobile users' dynamics could make the cost of implementing SCS high, so we discuss the feasibility of using a simpler (dynamically) weighted max-min as a surrogate resource allocation scheme. For a setting with stochastic loads and elastic user requirements, we establish a sufficient condition for the stability of the associated coupled network system. Finally, and perhaps surprisingly, we show via extensive simulations that while SCS (and/or the surrogate weighted max-min allocation) provides inter-slice protection, they can achieve improved job delay and/or perceived throughput, as compared to other weighted max-min based allocation schemes whose intra-slice weight allocation is not share-constrained, e.g., traditional max-min or discriminatory processor sharing.

A critical requirement for automated driving systems is enabling situational awareness in dynamically changing environments. To that end vehicles will be equipped with diverse sensors, e.g., LIDAR, cameras, mmWave radar, etc. Unfortunately the sensing `coverage' is limited by environmental obstructions, e.g., other vehicles, buildings, people, objects etc. A possible solution is to adopt collaborative sensing amongst vehicles possibly assisted by infrastructure. This paper introduces new models and performance analysis for vehicular collaborative sensing and networking. In particular, coverage gains are quantified, as are their dependence on the penetration of vehicles participating in collaborative sensing. We also evaluate the associated communication loads in terms of the Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) capacity requirements and how these depend on penetration. We further explore how collaboration with sensing capable infrastructure improves sensing performance, as well as the benefits in utilizing spatio-temporal dynamics, e.g., collaborating with vehicles moving in the opposite direction. Collaborative sensing is shown to greatly improve sensing performance, e.g., improves coverage from 20% to 80% with a 20% penetration. In scenarios with limited penetration and high coverage requirements, infrastructure can be used to both help sense the environment and relay data. Once penetration is high enough, sensing vehicles provide good coverage and data traffic can be effectively `offloaded' to V2V connectivity, making V2I resources available to support other in-car services.

5G wireless networks are expected to support Ultra Reliable Low Latency Communications (URLLC) traffic which requires very low packet delays ( < 1 msec.) and extremely high reliability ($\sim$99.999\%). In this paper we focus on the design of a wireless system supporting downlink URLLC traffic. Using a queuing network based model for the wireless system we characterize the effect of various design choices on the maximum URLLC load it can support, including: 1) system parameters such as the bandwidth, link SINR , and QoS requirements; 2) resource allocation schemes in Orthogonal Frequency Division Multiple Access (OFDMA) based systems; and 3) Hybrid Automatic Repeat Request (HARQ) schemes. Key contributions of this paper which are of practical interest are: 1) study of how the the minimum required system bandwidth to support a given URLLC load scales with associated QoS constraints; 2) characterization of optimal OFDMA resource allocation schemes which maximize the admissible URLLC load; and 3) optimization of a repetition code based HARQ scheme which approximates Chase HARQ combining.

Emerging 5G systems will need to efficiently support both enhanced mobile broadband traffic (eMBB) and ultra-low-latency communications (URLLC) traffic. In these systems, time is divided into slots which are further sub-divided into minislots. From a scheduling perspective, eMBB resource allocations occur at slot boundaries, whereas to reduce latency URLLC traffic is pre-emptively overlapped at the minislot timescale, resulting in selective superposition/puncturing of eMBB allocations. This approach enables minimal URLLC latency at a potential rate loss to eMBB traffic. We study joint eMBB and URLLC schedulers for such systems, with the dual objectives of maximizing utility for eMBB traffic while immediately satisfying URLLC demands. For a linear rate loss model (loss to eMBB is linear in the amount of URLLC superposition/puncturing), we derive an optimal joint scheduler. Somewhat counter-intuitively, our results show that our dual objectives can be met by an iterative gradient scheduler for eMBB traffic that anticipates the expected loss from URLLC traffic, along with an URLLC demand scheduler that is oblivious to eMBB channel states, utility functions and allocation decisions of the eMBB scheduler. Next we consider a more general class of (convex/threshold) loss models and study optimal online joint eMBB/URLLC schedulers within the broad class of channel state dependent but minislot-homogeneous policies. A key observation is that unlike the linear rate loss model, for the convex and threshold rate loss models, optimal eMBB and URLLC scheduling decisions do not de-couple and joint optimization is necessary to satisfy the dual objectives. We validate the characteristics and benefits of our schedulers via simulation.

In a large-scale computing cluster, the job completions can be substantially delayed due to two sources of variability, namely, variability in the job size and that in the machine service capacity. To tackle this issue, existing works have proposed various scheduling algorithms which exploit redundancy wherein a job runs on multiple servers until the first completes. In this paper, we explore the impact of variability in the machine service capacity and adopt a rigorous analytical approach to design scheduling algorithms using redundancy and checkpointing. We design several online scheduling algorithms which can dynamically vary the number of redundant copies for jobs. We also provide new theoretical performance bounds for these algorithms in terms of the overall job flowtime by introducing the notion of a speedup function, based on which a novel potential function can be defined to enable the corresponding competitive ratio analysis. In particular, by adopting the online primal-dual fitting approach, we prove that our SRPT+R Algorithm in a non-multitasking cluster is $(1+\epsilon)$-speed, $\ O(\frac{1}{\epsilon})$-competitive. We also show that our proposed Fair+R and LAPS+R($\beta$) Algorithms for a multitasking cluster are $(4+\epsilon)$-speed, $\ O(\frac{1}{\epsilon})$-competitive and {($2 + 2\beta + 2\epsilon)$-speed $O(\frac{1}{\beta \epsilon})$-competitive} respectively. We demonstrate via extensive simulations that our proposed algorithms can significantly reduce job flowtime under both the non-multitasking and multitasking modes.

Next generation wireless architectures are expected to enable slices of shared wireless infrastructure which are customized to specific mobile operators/services. Given infrastructure costs and the stochastic nature of mobile services' spatial loads, it is highly desirable to achieve efficient statistical multiplexing amongst such slices. We study a simple dynamic resource sharing policy which allocates a 'share' of a pool of (distributed) resources to each slice-Share Constrained Proportionally Fair (SCPF). We give a characterization of SCPF's performance gains over static slicing and general processor sharing. We show that higher gains are obtained when a slice's spatial load is more 'imbalanced' than, and/or 'orthogonal' to, the aggregate network load, and that the overall gain across slices is positive. We then address the associated dimensioning problem. Under SCPF, traditional network dimensioning translates to a coupled share dimensioning problem, which characterizes the existence of a feasible share allocation given slices' expected loads and performance requirements. We provide a solution to robust share dimensioning for SCPF-based network slicing. Slices may wish to unilaterally manage their users' performance via admission control which maximizes their carried loads subject to performance requirements. We show this can be modeled as a 'traffic shaping' game with an achievable Nash equilibrium. Under high loads, the equilibrium is explicitly characterized, as are the gains in the carried load under SCPF vs. static slicing. Detailed simulations of a wireless infrastructure supporting multiple slices with heterogeneous mobile loads show the fidelity of our models and range of validity of our high load equilibrium analysis.

Network slicing to enable resource sharing among multiple tenants --network operators and/or services-- is considered a key functionality for next generation mobile networks. This paper provides an analysis of a well-known model for resource sharing, the 'share-constrained proportional allocation' mechanism, to realize network slicing. This mechanism enables tenants to reap the performance benefits of sharing, while retaining the ability to customize their own users' allocation. This results in a network slicing game in which each tenant reacts to the user allocations of the other tenants so as to maximize its own utility. We show that, under appropriate conditions, the game associated with such strategic behavior converges to a Nash equilibrium. At the Nash equilibrium, a tenant always achieves the same, or better, performance than under a static partitioning of resources, hence providing the same level of protection as such static partitioning. We further analyze the efficiency and fairness of the resulting allocations, providing tight bounds for the price of anarchy and envy-freeness. Our analysis and extensive simulation results confirm that the mechanism provides a comprehensive practical solution to realize network slicing. Our theoretical results also fill a gap in the literature regarding the analysis of this resource allocation model under strategic players., Comment: Extended version of the 2017 INFOCOM publication

We consider a system of processor-sharing queues with state-dependent service rates. These are allocated according to balanced fairness within a polymatroid capacity set. Balanced fairness is known to be both insensitive and Pareto-efficient in such systems, which ensures that the performance metrics, when computable, will provide robust insights into the real performance of the system considered. We first show that these performance metrics can be evaluated with a complexity that is polynomial in the system size if the system is partitioned into a finite number of parts, so that queues are exchangeable within each part and asymmetric across different parts. This in turn allows us to derive stochastic bounds for a larger class of systems which satisfy less restrictive symmetry assumptions. These results are applied to practical examples of tree data networks, such as backhaul networks of Internet service providers, and computer clusters.

This paper focuses on modeling and analysis of the temporal performance variations experienced by a mobile user in a wireless network and its impact on system level design. We consider a simple stochastic geometry model: the infrastructure nodes are Poisson distributed while the user's motion is the simplest possible i.e., constant velocity on a straight line. We first characterize variations in the SNR process, and associated downlink Shannon rate, resulting from variations in the infrastructure geometry seen by the mobile. Specifically, by making a connection between stochastic geometry and queueing theory the level crossings of the SNR process are shown to form an alternating renewal process whose distribution can be completely characterized. For large or small SNR levels, and associated rare events, we further derive simple distributional (exponential) models. We then characterize the second major contributor to variation, associated with changes in the number of other users sharing infrastructure. Combining these two effects, we study what are the dominant factors (infrastructure geometry or sharing number) given mobile experiences a very high or low shared rate. These results are then used to evaluate and optimize the system-level Quality of Service (QoS) and system-level capacity to support mobile users sharing wireless infrastructure, including mobile devices streaming video which proactively buffer content to prevent rebuffering and mobiles which are downloading large files. Finally, we use simulation to assess the fidelity of this model and its robustness to factors which are presently taken into account.

An increasing number of real-time applications with compute and/or communication deadlines are being supported on shared infrastructure. Such applications can often tolerate occasional deadline violations without substantially impacting their Quality of Service (QoS). A fundamental problem in such systems is deciding how to allocate shared resources so as to meet applications' QoS requirements. A simple framework to address this problem is to, (1) dynamically prioritize users as a possibly complex function of their deficits (difference of achieved vs required QoS), and (2) allocate resources so to expedite users with higher priority. This paper focuses on a general class of systems using such priority-based resource allocation. We first characterize the set of feasible QoS requirements and show the optimality of max weight-like prioritization. We then consider simple weighted Largest Deficit First (w-LDF) prioritization policies, where users with higher weighted QoS deficits are given higher priority. The paper gives an inner bound for the feasible set under w-LDF policies, and, under an additional monotonicity assumption, characterizes its geometry leading to a sufficient condition for optimality. Additional insights on the efficiency ratio of w-LDF policies, the optimality of hierarchical-LDF and characterization of clustering of failures are also discussed., Comment: This is an extended version of this paper

Cloud-based computing infrastructure provides an efficient means to support real-time processing workloads, e.g., virtualized base station processing, and collaborative video conferencing. This paper addresses resource allocation for a computing system with multiple resources supporting heterogeneous soft real-time applications subject to Quality of Service (QoS) constraints on failures to meet processing deadlines. We develop a general outer bound on the feasible QoS region for non-clairvoyant resource allocation policies, and an inner bound for a natural class of policies based on dynamically prioritizing applications' tasks by favoring those with the largest (QoS) deficits. This provides an avenue to study the efficiency of two natural resource allocation policies: (1) priority-based greedy task scheduling for applications with variable workloads, and (2) priority-based task selection and optimal scheduling for applications with deterministic workloads. The near-optimality of these simple policies emerges when task processing deadlines are relatively large and/or when the number of compute resources is large. Analysis and simulations show substantial resource savings for such policies over reservation-based designs., Comment: This is an extended version of this paper

In this paper, we present an approach for joint rate allocation and quality selection for a novel video streaming scheme called streamloading. Streamloading is a recently developed method for delivering high quality video without violating copyright enforced restrictions on content access for video streaming. In regular streaming services, content providers restrict the amount of viewable video that users can download prior to playback. This approach can cause inferior user experience due to bandwidth variations, especially in mobile networks with varying capacity. In streamloading, the video is encoded using Scalable Video Coding, and users are allowed to pre-fetch enhancement layers and store them on the device, while base layers are streamed in a near real-time fashion ensuring that buffering constraints on viewable content are met. We begin by formulating the offline problem of jointly optimizing rate allocation and quality selection for streamloading in a wireless network. This motivates our proposed online algorithms for joint scheduling at the base station and segment quality selection at receivers. The results indicate that streamloading outperforms state-of-the-art streaming schemes in terms of the number of additional streams we can admit for a given video quality. Furthermore, the quality adaptation mechanism of our proposed algorithm achieves a higher performance than baseline algorithms with no (or limited) video-centric optimization of the base station's allocation of resources, e.g., proportional fairness.

At the core of high throughput DNA sequencing platforms lies a bio-physical surface process that results in a random geometry of clusters of homogenous short DNA fragments typically hundreds of base pairs long - bridge amplification. The statistical properties of this random process and length of the fragments are critical as they affect the information that can be subsequently extracted, i.e., density of successfully inferred DNA fragment reads. The ensemble of overlapping DNA fragment reads are then used to computationally reconstruct the much longer target genome sequence, e.g, ranging from hundreds of thousands to billions of base pairs. The success of the reconstruction in turn depends on having a sufficiently large ensemble of DNA fragments that are sufficiently long. In this paper using stochastic geometry we model and optimize the end-to-end process linking and partially controlling the statistics of the physical processes to the success of the computational step. This provides, for the first time, a framework capturing salient features of such sequencing platforms that can be used to study cost, performance or sensitivity of the sequencing process.

Newly developed HTTP-based video streaming technologies enable flexible rate-adaptation under varying channel conditions. Accurately predicting the users' Quality of Experience (QoE) for rate-adaptive HTTP video streams is thus critical to achieve efficiency. An important aspect of understanding and modeling QoE is predicting the up-to-the-moment subjective quality of a video as it is played, which is difficult due to hysteresis effects and nonlinearities in human behavioral responses. This paper presents a Hammerstein-Wiener model for predicting the time-varying subjective quality (TVSQ) of rate-adaptive videos. To collect data for model parameterization and validation, a database of longer-duration videos with time-varying distortions was built and the TVSQs of the videos were measured in a large-scale subjective study. The proposed method is able to reliably predict the TVSQ of rate adaptive videos. Since the Hammerstein-Wiener model has a very simple structure, the proposed method is suitable for on-line TVSQ prediction in HTTP based streaming.

Adapting video data rate during streaming can effectively reduce the risk of playback interruptions caused by channel throughput fluctuations. The variations in rate, however, also introduce video quality fluctuations and thus potentially affects viewers' Quality of Experience (QoE). We show how the QoE of video users can be improved by rate adaptation and admission control. We conducted a subjective study wherein we found that viewers' QoE was strongly correlated with the empirical cumulative distribution function (eCDF) of the predicted video quality. Based on this observation, we propose a rate-adaptation algorithm that can incorporate QoE constraints on the empirical cumulative quality distribution per user. We then propose a threshold-based admission control policy to block users whose empirical cumulative quality distribution is not likely to satisfy their QoE constraint. We further devise an online adaptation algorithm to automatically optimize the threshold. Extensive simulation results show that the proposed scheme can reduce network resource consumption by $40\%$ over conventional average-quality maximized rate-adaptation algorithms.

We consider the problem of optimizing video delivery for a network supporting video clients streaming stored video. Specifically, we consider the problem of jointly optimizing network resource allocation and video quality adaptation. Our objective is to fairly maximize video clients' Quality of Experience (QoE) realizing tradeoffs among the mean quality, temporal variability in quality, and fairness, incorporating user preferences on rebuffering and cost of video delivery. We present a simple asymptotically optimal online algorithm, NOVA, to solve the problem. NOVA is asynchronous, and using minimal communication, distributes the tasks of resource allocation to network controller, and quality adaptation to respective video clients. Video quality adaptation in NOVA is also optimal for standalone video clients, and is well suited for use with DASH framework. Further, we extend NOVA for use with more general QoE models, networks shared with other traffic loads and networks using fixed/legacy resource allocation., Comment: This is an extended version of (9-page) conference version available at http://wncg.org/publications/dl.php?file=VJGdV_conf_13.pdf

We propose two scheduling algorithms that seek to optimize the quality of scalably coded videos that have been stored at a video server before transmission.} The first scheduling algorithm is derived from a Markov Decision Process (MDP) formulation developed here. We model the dynamics of the channel as a Markov chain and reduce the problem of dynamic video scheduling to a tractable Markov decision problem over a finite state space. Based on the MDP formulation, a near-optimal scheduling policy is computed that minimize the mean square error. Using insights taken from the development of the optimal MDP-based scheduling policy, the second proposed scheduling algorithm is an online scheduling method that only requires easily measurable knowledge of the channel dynamics, and is thus viable in practice. Simulation results show that the performance of both scheduling algorithms is close to a performance upper bound also derived in this paper., Comment: 14 pages

Network Utility Maximization (NUM) provides a key conceptual framework to study reward allocation amongst a collection of users/entities across disciplines as diverse as economics, law and engineering. In network engineering, this framework has been particularly insightful towards understanding how Internet protocols allocate bandwidth, and motivated diverse research efforts on distributed mechanisms to maximize network utility while incorporating new relevant constraints, on energy, power, storage, stability, etc., e.g., for systems ranging from communication networks to the smart-grid. However when the available resources and/or users' utilities vary over time, reward allocations will tend to vary, which in turn may have a detrimental impact on the users' overall satisfaction or quality of experience. This paper introduces a generalization of NUM framework which explicitly incorporates the detrimental impact of temporal variability in a user's allocated rewards. It explicitly incorporates tradeoffs amongst the mean and variability in users' reward allocations, as well as fairness. We propose a simple online algorithm to realize these tradeoffs, which, under stationary ergodic assumptions, is shown to be asymptotically optimal, i.e., achieves a long term performance equal to that of an offline algorithm with knowledge of the future variability in the system. This substantially extends work on NUM to an interesting class of relevant problems where users/entities are sensitive to temporal variability in their service or allocated rewards., Comment: arXiv admin note: substantial text overlap with arXiv:1111.3728

Network Utility Maximization (NUM) provides the key conceptual framework to study resource allocation amongst a collection of users/entities across disciplines as diverse as economics, law and engineering. In network engineering, this framework has been particularly insightful towards understanding how Internet protocols allocate bandwidth, and motivated diverse research on distributed mechanisms to maximize network utility while incorporating new relevant constraints, on energy/power, storage, stability, etc., for systems ranging from communication networks to the smart-grid. However when the available resources and/or users' utilities vary over time, a user's allocations will tend to vary, which in turn may have a detrimental impact on the users' utility or quality of experience. This paper introduces a generalized NUM framework which explicitly incorporates the detrimental impact of temporal variability in a user's allocated rewards. It explicitly incorporates tradeoffs amongst the mean and variability in users' allocations. We propose an online algorithm to realize variance-sensitive NUM, which, under stationary ergodic assumptions, is shown to be asymptotically optimal, i.e., achieves a time-average equal to that of an offline algorithm with knowledge of the future variability in the system. This substantially extends work on NUM to an interesting class of relevant problems where users/entities are sensitive to temporal variability in their service or allocated rewards.

We propose a scheme to reduce the overhead associated with channel state information (CSI) feedback required for opportunistic scheduling in multicarrier access networks. We study the case where CSI is partially overheard by mobiles and one can suppress transmitting CSI reports for time varying channel of inferior quality. As a means to assess channel quality and exploit multiuser diversity we adopt maximum quantile (MQ) scheduling. We show that the problem of minimizing the average feedback overhead can be formulated as a Bayesian network problem. A greedy heuristic using probabilistic inference is proposed to deal with the NP-hardness of the problem. Leveraging properties of MQ scheduling we first show that networks having tree-like overhearing graphs admit simple inference. We then present a class of more general network structures for which exact inference is computationally tractable. Simulation results are provided to demonstrate the improvements offered by the proposed heuristic., Comment: The paper is withdrawn since the material will be reconstructed in an entirely different context

A centralized wireless system is considered that is serving a fixed set of users with time varying channel capacities. An opportunistic scheduling rule in this context selects a user (or users) to serve based on the current channel state and user queues. Unless the user traffic is symmetric and/or the underlying capacity region a polymatroid, little is known concerning how performance optimal schedulers should tradeoff "maximizing current service rate" (being opportunistic) versus "balancing unequal queues" (enhancing user-diversity to enable future high service rate opportunities). By contrast with currently proposed opportunistic schedulers, e.g., MaxWeight and Exp Rule, a radial sum-rate monotone (RSM) scheduler de-emphasizes queue-balancing in favor of greedily maximizing the system service rate as the queue-lengths are scaled up linearly. In this paper it is shown that an RSM opportunistic scheduler, p-Log Rule, is not only throughput-optimal, but also maximizes the asymptotic exponential decay rate of the sum-queue distribution for a two-queue system. The result complements existing optimality results for opportunistic scheduling and point to RSM schedulers as a good design choice given the need for robustness in wireless systems with both heterogeneity and high degree of uncertainty., Comment: Revised version. Major changes include addition of details/intermediate steps in various proofs, a summary of technical steps in Table 1, and correction of typos.