arXiv subjects

Ramesh K. Sitaraman

Publications and source records attributed to Ramesh K. Sitaraman.

At least 19 recordsLinked to original sources

arXiv2026

CATRF: Codec-Adaptive TriPlane Radiance Fields for Volumetric Content Delivery

Volumetric media promises next-generation content delivery applications, but its bandwidth demand remains a key bottleneck. Implicit and hybrid volumetric representations reduce model sizes, yet still require careful coding to reach 2D video-like bitrates. We present CATRF, a standard-codec-in-the-loop compression framework for plane-factorized radiance fields. During training, we quantize and pack 2D feature planes into codec-friendly canvases, run a standard codec roundtrip (JPEG/VP9/HEVC/AV1), then unpack and dequantize the decoded features before volume rendering. We use a straight-through estimator (STE) to insert the non-differentiable, standard codec pipeline into the training loop, allowing radiance-field features to adapt directly to the real, client-side codec distortions without introducing any learned codec parameters. On both static and dynamic benchmarks, CATRF consistently achieves a better rate-distortion trade-off over codec-agnostic and learned-codec-in-the-loop baselines, and also outperforms recent compressed 3DGS methods in both compression efficiency and decoding speed. These results highlight a practical path toward low-bitrate, compression-resilient volumetric representations for free-viewpoint video streaming.

eess.IV↗

arXiv2026

Low-Bitrate Video Compression through Semantic-Conditioned Diffusion

Traditional video codecs optimized for pixel fidelity collapse at ultra-low bitrates and produce severe artifacts. This failure arises from a fundamental misalignment between pixel accuracy and human perception. We propose a semantic video compression framework named DiSCo that transmits only the most meaningful information while relying on generative priors for detail synthesis. The source video is decomposed into three compact modalities: a textual description, a spatiotemporally degraded video, and optional sketches or poses that respectively capture semantic, appearance, and motion cues. A conditional video diffusion model then reconstructs high-quality, temporally coherent videos from these compact representations. Temporal forward filling, token interleaving, and modality-specific codecs are proposed to improve multimodal generation and modality compactness. Experiments show that our method outperforms baseline semantic and traditional codecs by 2-10X on perceptual metrics at low bitrates.

cs.CV↗

arXiv2026

HADIS: Hybrid Adaptive Diffusion Model Serving for Efficient Text-to-Image Generation

Text-to-image diffusion models have achieved remarkable visual quality but incur high computational costs, making latency-aware, scalable deployment challenging. To address this, we advocate a hybrid architecture that achieves query awareness when serving diffusion models. Unlike existing query-aware serving systems that cascade lightweight and heavyweight models with a fixed configuration, our hybrid architecture first routes each query directly to a suitable model variant, then reroutes it to a cascaded heavyweight model only if necessary. We theoretically analyze conditions for the hybrid architecture to outperform non-hybrid alternatives in latency and response quality. Building on this architecture, we design HADIS, a hybrid serving system for latency-aware diffusion models that jointly optimizes cascade model selection, query routing, and resource allocation. To reduce the complexity of resource management, HADIS uses an offline profiling phase to produce a Pareto-optimal cascade configuration table. At runtime, HADIS selects the best cascade configuration and GPU allocation given latency and workload constraints. Empirical evaluations on real-world traces demonstrate that HADIS improves response quality by up to 35% while reducing latency violation rates by 2.7-45$\times$ compared to state-of-the-art model serving systems.

cs.DC↗

arXiv2025

Near-Optimal Emission-Aware Online Ride Assignment Algorithm for Peak Demand Hours

Ridesharing has experienced significant global growth over the past decade and is becoming an integral component of modern transportation systems. However, despite their benefits, ridesharing platforms face fundamental inefficiencies that contribute to negative environmental impacts. A prominent source of such inefficiency is the deadhead miles. This issue becomes especially severe during high-demand periods, when the volume of ride requests exceeds the available driver supply, leading to suboptimal rider-to-driver assignments, longer deadhead trips, and increased emissions. Although limiting these unproductive miles can reduce emissions, doing so may increase passenger wait times due to limited driver availability, thereby degrading the overall service experience. In this paper, we introduce LARA, an online rider-to-driver assignment algorithm that dynamically adjusts the maximum allowable distance between rider and drivers and assigns ride requests accordingly. While LARA is applicable in general settings, it is particularly effective during peak demand periods, achieving reductions in both emissions and wait times. We provide theoretical guarantees showing that LARA achieves near-optimal performance in online environments, with respect to an optimal offline benchmark. Beside our theoretical analysis, our empirical evaluations on both synthetic and real-world datasets show that LARA achieves up to a 34% reduction in carbon emissions and up to a 50% decrease in rider wait times, compared to state-of-the-art baselines. While prior work has explored emission-aware ride assignment, LARA is, to our knowledge, the first algorithm to offer both rigorous theoretical guarantees and strong empirical performance.

eess.SY↗

arXiv2025

Smoothed Online Optimization for Target Tracking: Robust and Learning-Augmented Algorithms

We introduce the Smoothed Online Optimization for Target Tracking (SOOTT) problem, a new framework that integrates three key objectives in online decision-making under uncertainty: (1) tracking cost for following a dynamically moving target, (2) adversarial perturbation cost for withstanding unpredictable disturbances, and (3) switching cost for penalizing abrupt changes in decisions. This formulation captures real-world scenarios such as elastic and inelastic workload scheduling in AI clusters, where operators must balance long-term service-level agreements (e.g., LLM training) against sudden demand spikes (e.g., real-time inference). We first present BEST, a robust algorithm with provable competitive guarantees for SOOTT. To enhance practical performance, we introduce CoRT, a learning-augmented variant that incorporates untrusted black-box predictions (e.g., from ML models) into its decision process. Our theoretical analysis shows that CoRT strictly improves over BEST when predictions are accurate, while maintaining robustness under arbitrary prediction errors. We validate our approach through a case study on workload scheduling, demonstrating that both algorithms effectively balance trajectory tracking, decision smoothness, and resilience to external disturbances.

cs.LG↗

arXiv2025

DiffServe: Efficiently Serving Text-to-Image Diffusion Models with Query-Aware Model Scaling

Text-to-image generation using diffusion models has gained increasing popularity due to their ability to produce high-quality, realistic images based on text prompts. However, efficiently serving these models is challenging due to their computation-intensive nature and the variation in query demands. In this paper, we aim to address both problems simultaneously through query-aware model scaling. The core idea is to construct model cascades so that easy queries can be processed by more lightweight diffusion models without compromising image generation quality. Based on this concept, we develop an end-to-end text-to-image diffusion model serving system, DiffServe, which automatically constructs model cascades from available diffusion model variants and allocates resources dynamically in response to demand fluctuations. Our empirical evaluations demonstrate that DiffServe achieves up to 24% improvement in response quality while maintaining 19-70% lower latency violation rates compared to state-of-the-art model serving systems.

cs.DC↗

arXiv2024

Towards Environmentally Equitable AI

The skyrocketing demand for artificial intelligence (AI) has created an enormous appetite for globally deployed power-hungry servers. As a result, the environmental footprint of AI systems has come under increasing scrutiny. More crucially, the current way that we exploit AI workloads' flexibility and manage AI systems can lead to wildly different environmental impacts across locations, increasingly raising environmental inequity concerns and creating unintended sociotechnical consequences. In this paper, we advocate environmental equity as a priority for the management of future AI systems, advancing the boundaries of existing resource management for sustainable AI and also adding a unique dimension to AI fairness. Concretely, we uncover the potential of equity-aware geographical load balancing to fairly re-distribute the environmental cost across different regions, followed by algorithmic challenges. We conclude by discussing a few future directions to exploit the full potential of system management approaches to mitigate AI's environmental inequity.

cs.LG↗

arXiv2024

BOLA360: Near-optimal View and Bitrate Adaptation for 360-degree Video Streaming

Recent advances in omnidirectional cameras and AR/VR headsets have spurred the adoption of 360-degree videos that are widely believed to be the future of online video streaming. 360-degree videos allow users to wear a head-mounted display (HMD) and experience the video as if they are physically present in the scene. Streaming high-quality 360-degree videos at scale is an unsolved problem that is more challenging than traditional (2D) video delivery. The data rate required to stream 360-degree videos is an order of magnitude more than traditional videos. Further, the penalty for rebuffering events where the video freezes or displays a blank screen is more severe as it may cause cybersickness. We propose an online adaptive bitrate (ABR) algorithm for 360-degree videos called BOLA360 that runs inside the client's video player and orchestrates the download of video segments from the server so as to maximize the quality-of-experience (QoE) of the user. BOLA360 conserves bandwidth by downloading only those video segments that are likely to fall within the field-of-view (FOV) of the user. In addition, BOLA360 continually adapts the bitrate of the downloaded video segments so as to enable a smooth playback without rebuffering. We prove that BOLA360 is near-optimal with respect to an optimal offline algorithm that maximizes QoE. Further, we evaluate BOLA360 on a wide range of network and user head movement profiles and show that it provides $13.6\%$ to $372.5\%$ more QoE than state-of-the-art algorithms. While ABR algorithms for traditional (2D) videos have been well-studied over the last decade, our work is the first ABR algorithm for 360-degree videos with both theoretical and empirical guarantees on its performance.

cs.MM↗

arXiv2024

Loki: A System for Serving ML Inference Pipelines with Hardware and Accuracy Scaling

The rapid adoption of machine learning (ML) has underscored the importance of serving ML models with high throughput and resource efficiency. Traditional approaches to managing increasing query demands have predominantly focused on hardware scaling, which involves increasing server count or computing power. However, this strategy can often be impractical due to limitations in the available budget or compute resources. As an alternative, accuracy scaling offers a promising solution by adjusting the accuracy of ML models to accommodate fluctuating query demands. Yet, existing accuracy scaling techniques target independent ML models and tend to underperform while managing inference pipelines. Furthermore, they lack integration with hardware scaling, leading to potential resource inefficiencies during low-demand periods. To address the limitations, this paper introduces Loki, a system designed for serving inference pipelines effectively with both hardware and accuracy scaling. Loki incorporates an innovative theoretical framework for optimal resource allocation and an effective query routing algorithm, aimed at improving system accuracy and minimizing latency deadline violations. Our empirical evaluation demonstrates that through accuracy scaling, the effective capacity of a fixed-size cluster can be enhanced by more than $2.7\times$ compared to relying solely on hardware scaling. When compared with state-of-the-art inference-serving systems, Loki achieves up to a $10\times$ reduction in Service Level Objective (SLO) violations, with minimal compromises on accuracy and while fulfilling throughput demands.

cs.DC↗

arXiv2024

BONES: Near-Optimal Neural-Enhanced Video Streaming

Accessing high-quality video content can be challenging due to insufficient and unstable network bandwidth. Recent advances in neural enhancement have shown promising results in improving the quality of degraded videos through deep learning. Neural-Enhanced Streaming (NES) incorporates this new approach into video streaming, allowing users to download low-quality video segments and then enhance them to obtain high-quality content without violating the playback of the video stream. We introduce BONES, an NES control algorithm that jointly manages the network and computational resources to maximize the quality of experience (QoE) of the user. BONES formulates NES as a Lyapunov optimization problem and solves it in an online manner with near-optimal performance, making it the first NES algorithm to provide a theoretical performance guarantee. Comprehensive experimental results indicate that BONES increases QoE by 5\% to 20\% over state-of-the-art algorithms with minimal overhead. Our code is available at https://github.com/UMass-LIDS/bones.

eess.SY↗

arXiv2024

Untangling Carbon-free Energy Attribution and Carbon Intensity Estimation for Carbon-aware Computing

Many organizations, including governments, utilities, and businesses, have set ambitious targets to reduce carbon emissions for their Environmental, Social, and Governance (ESG) goals. To achieve these targets, these organizations increasingly use power purchase agreements (PPAs) to obtain renewable energy credits, which they use to compensate for the ``brown'' energy consumed from the grid. However, the details of these PPAs are often private and not shared with important stakeholders, such as grid operators and carbon information services, who monitor and report the grid's carbon emissions. This often results in incorrect carbon accounting, where the same renewable energy production could be factored into grid carbon emission reports and separately claimed by organizations that own PPAs. Such ``double counting'' of renewable energy production could lead organizations with PPAs to understate their carbon emissions and overstate their progress toward sustainability goals, and also provide significant challenges to consumers using common carbon reduction measures to decrease their carbon footprint. Unfortunately, there is no consensus on accurately computing the grid's carbon intensity by properly accounting for PPAs. The goal of our work is to shed quantitative and qualitative light on the renewable energy attribution and the incorrect carbon intensity estimation problems.

cs.DC↗

arXiv2024

The Green Mirage: Impact of Location- and Market-based Carbon Intensity Estimation on Carbon Optimization Efficacy

In recent years, there has been an increased emphasis on reducing the carbon emissions from electricity consumption. Many organizations have set ambitious targets to reduce the carbon footprint of their operations as a part of their sustainability goals. The carbon footprint of any consumer of electricity is computed as the product of the total energy consumption and the carbon intensity of electricity. Third-party carbon information services provide information on carbon intensity across regions that consumers can leverage to modulate their energy consumption patterns to reduce their overall carbon footprint. In addition, to accelerate their decarbonization process, large electricity consumers increasingly acquire power purchase agreements (PPAs) from renewable power plants to obtain renewable energy credits that offset their "brown" energy consumption. There are primarily two methods for attributing carbon-free energy, or renewable energy credits, to electricity consumers: location-based and market-based. These two methods yield significantly different carbon intensity values for various consumers. As there is a lack of consensus which method to use for carbon-free attribution, a concurrent application of both approaches is observed in practice. In this paper, we show that such concurrent applications can cause discrepancies in the carbon savings reported by carbon optimization techniques. Our analysis across three state-of-the-art carbon optimization techniques shows possible overestimation of up to 55.1% in the carbon reductions reported by the consumers and even increased emissions for consumers in some cases. We also find that carbon optimization techniques make different decisions under the market-based method and location-based method, and the market-based method can yield up to 28.2% less carbon savings than those claimed by the location-based method for consumers without PPAs.

cs.DC↗

arXiv2022

CU-Net: Real-Time High-Fidelity Color Upsampling for Point Clouds

Point cloud upsampling is essential for high-quality augmented reality, virtual reality, and telepresence applications, due to the capture, processing, and communication limitations of existing technologies. Although geometry upsampling to densify a point cloud's coordinates has been well studied, the upsampling of the color attributes has been largely overlooked. In this paper, we propose CU-Net, the first deep-learning point cloud color upsampling model that enables low latency and high visual fidelity operation. CU-Net achieves linear time and space complexity by leveraging a feature extractor based on sparse convolution and a color prediction module based on neural implicit function. Therefore, CU-Net is theoretically guaranteed to be more efficient than most existing methods with quadratic complexity. Experimental results demonstrate that CU-Net can colorize a photo-realistic point cloud with nearly a million points in real time, while having notably better visual performance than baselines. Besides, CU-Net can adapt to arbitrary upsampling ratios and unseen objects without retraining. Our source code is available at https://github.com/UMass-LIDS/cunet.

cs.CV↗

arXiv2020

BOLA: Near-Optimal Bitrate Adaptation for Online Videos

Modern video players employ complex algorithms to adapt the bitrate of the video that is shown to the user. Bitrate adaptation requires a tradeoff between reducing the probability that the video freezes (rebuffers) and enhancing the quality of the video. A bitrate that is too high leads to frequent rebuffering, while a bitrate that is too low leads to poor video quality. Video providers segment videos into short segments and encode each segment at multiple bitrates. The video player adaptively chooses the bitrate of each segment to download, possibly choosing different bitrates for successive segments. We formulate bitrate adaptation as a utility-maximization problem and devise an online control algorithm called BOLA that uses Lyapunov optimization to minimize rebuffering and maximize video quality. We prove that BOLA achieves a time-average utility that is within an additive term O(1/V) of the optimal value, for a control parameter V related to the video buffer size. Further, unlike prior work, BOLA does not require prediction of available network bandwidth. We empirically validate BOLA in a simulated network environment using a collection of network traces. We show that BOLA achieves near-optimal utility and in many cases significantly higher utility than current state-of-the-art algorithms. Our work has immediate impact on real-world video players and for the evolving DASH standard for video transmission. We also implemented an updated version of BOLA that is now part of the standard reference player dash.js and is used in production by several video providers such as Akamai, BBC, CBS, and Orange.

cs.NI↗

arXiv2019

OCCAM: An Optimization-Based Approach to Network Inference

We study the problem of inferring the structure of a communication network based only on network measurements made from a set of hosts situated at the network periphery. Our novel approach called "OCCAM" is based on the principle of occam's razor and finds the "simplest" network that explains the observed network measurements. OCCAM infers the internal topology of a communication network, including the internal nodes and links of the network that are not amenable to direct measurement. In addition to network topology, OCCAM infers the routing paths that packets take between the hosts. OCCAM uses path metrics measurable from the hosts and expresses the observed measurements as constraints of a mixed-integer bilinear optimization problem that can then be feasibly solved to yield the network topology and the routing paths. We empirically validate OCCAM on a wide variety of real-world ISP networks and show that its inferences agree closely with the ground truth. Specifically, OCCAM infers the topology with an average network similarity score of 93% and infers routing paths with a path edit distance of 0.20. Further, OCCAM is robust to error in its measured path metric inputs, producing high quality inferences even when 20-30% of its inputs are erroneous. Our work is a significant advance in network tomography as it proposes and empirically evaluates the first method that infers the complete network topology, rather than just logical routing trees from sources.

cs.NI↗

arXiv2016

Go-With-The-Winner: Client-Side Server Selection for Content Delivery

Content delivery networks deliver much of the web and video content in the world by deploying a large distributed network of servers. We model and analyze a simple paradigm for client-side server selection that is commonly used in practice where each user independently measures the performance of a set of candidate servers and selects the one that performs the best. For web (resp., video) delivery, we propose and analyze a simple algorithm where each user randomly chooses two or more candidate servers and selects the server that provided the best hit rate (resp., bit rate). We prove that the algorithm converges quickly to an optimal state where all users receive the best hit rate (resp., bit rate), with high probability. We also show that if each user chose just one random server instead of two, some users receive a hit rate (resp., bit rate) that tends to zero. We simulate our algorithm and evaluate its performance with varying choices of parameters, system load, and content popularity.

cs.NI↗

arXiv2013

Dynamic Provisioning in Next-Generation Data Centers with On-site Power Production

The critical need for clean and economical sources of energy is transforming data centers that are primarily energy consumers to also energy producers. We focus on minimizing the operating costs of next-generation data centers that can jointly optimize the energy supply from on-site generators and the power grid, and the energy demand from servers as well as power conditioning and cooling systems. We formulate the cost minimization problem and present an offline optimal algorithm. For "on-grid" data centers that use only the grid, we devise a deterministic online algorithm that achieves the best possible competitive ratio of $2-α_{s}$, where $α_{s}$ is a normalized look-ahead window size. For "hybrid" data centers that have on-site power generation in addition to the grid, we develop an online algorithm that achieves a competitive ratio of at most \textmd{\normalsize {\small $\frac{P_{\max} (2-α_{s})}{c_{o}+c_{m}/L} [1+2\frac{P_{\max}-c_{o}}{P_{\max}(1+α_{g})}]$}}, where $α_{s}$ and $α_{g}$ are normalized look-ahead window sizes, $P_{\max}$ is the maximum grid power price, and $L$, $c_{o}$, and $c_{m}$ are parameters of an on-site generator. Using extensive workload traces from Akamai with the corresponding grid power prices, we simulate our offline and online algorithms in a realistic setting. Our offline (resp., online) algorithm achieves a cost reduction of 25.8% (resp., 20.7%) for a hybrid data center and 12.3% (resp., 7.3%) for an on-grid data center. The cost reductions are quite significant and make a strong case for a joint optimization of energy supply and energy demand in a data center. A hybrid data center provides about 13% additional cost reduction over an on-grid data center representing the additional cost benefits that on-site power generation provides over using the grid alone.

cs.DS↗

arXiv2012

Optimizing MapReduce for Highly Distributed Environments

MapReduce, the popular programming paradigm for large-scale data processing, has traditionally been deployed over tightly-coupled clusters where the data is already locally available. The assumption that the data and compute resources are available in a single central location, however, no longer holds for many emerging applications in commercial, scientific and social networking domains, where the data is generated in a geographically distributed manner. Further, the computational resources needed for carrying out the data analysis may be distributed across multiple data centers or community resources such as Grids. In this paper, we develop a modeling framework to capture MapReduce execution in a highly distributed environment comprising distributed data sources and distributed computational resources. This framework is flexible enough to capture several design choices and performance optimizations for MapReduce execution. We propose a model-driven optimization that has two key features: (i) it is end-to-end as opposed to myopic optimizations that may only make locally optimal but globally suboptimal decisions, and (ii) it can control multiple MapReduce phases to achieve low runtime, as opposed to single-phase optimizations that may control only individual phases. Our model results show that our optimization can provide nearly 82% and 64% reduction in execution time over myopic and single-phase optimizations, respectively. We have modified Hadoop to implement our model outputs, and using three different MapReduce applications over an 8-node emulated PlanetLab testbed, we show that our optimized Hadoop execution plan achieves 31-41% reduction in runtime over a vanilla Hadoop execution. Our model-driven optimization also provides several insights into the choice of techniques and execution parameters based on application and platform characteristics.

cs.DC↗