arXiv subjects

Peter Steenkiste

Publications and source records attributed to Peter Steenkiste.

8 recordsLinked to original sources

arXiv2025

ECCO: Leveraging Cross-Camera Correlations for Efficient Live Video Continuous Learning

Recent advances in video analytics address real-time data drift by continuously retraining specialized, lightweight DNN models for individual cameras. However, the current practice of retraining a separate model for each camera suffers from high compute and communication costs, making it unscalable. We present ECCO, a new video analytics framework designed for resource-efficient continuous learning. The key insight is that the data drift, which necessitates model retraining, often shows temporal and spatial correlations across nearby cameras. By identifying cameras that experience similar drift and retraining a shared model for them, ECCO can substantially reduce the associated compute and communication costs. Specifically, ECCO introduces: (i) a lightweight grouping algorithm that dynamically forms and updates camera groups; (ii) a GPU allocator that dynamically assigns GPU resources across different groups to improve retraining accuracy and ensure fairness; and (iii) a transmission controller at each camera that configures frame sampling and coordinates bandwidth sharing with other cameras based on its assigned GPU resources. We conducted extensive evaluations on three distinctive datasets for two vision tasks. Compared to leading baselines, ECCO improves retraining accuracy by 6.7%-18.1% using the same compute and communication resources, or supports 3.3 times more concurrent cameras at the same accuracy.

cs.DC↗

arXiv2025

Toward Co-adapting Machine Learning Job Shape and Cluster Topology

Allocating resources to distributed machine learning jobs in multi-tenant torus-topology clusters must meet each job's specific placement and communication requirements, which are typically described using shapes. There is an inherent tension between minimizing network contention and maximizing cluster utilization when placing various-shaped jobs. While existing schedulers typically optimize for one objective at the expense of the other, we demonstrate that both can be achieved simultaneously. Our proposed approach, RFold, adapts both job shapes and the underlying cluster topology at runtime. This is accomplished by combining two techniques: (1) identifying homomorphic job shapes that support the jobs communication needs, and (2) reconfiguring the optical circuit switch-enabled topology to support more diverse job shapes. Preliminary evaluation performed on a 4096-node torus cluster simulator indicates that RFold can improve absolute cluster utilization by 57% and reduce job completion time by up to 11x relative to existing methods

cs.DC↗

arXiv2022

A study of HTTP/2's Server Push Performance Potential

Modern web pages have complex structures comprised of up to hundreds of different resources, such as scripts and images. Server push is an HTTP/2 feature enabling servers to preemptively send resources to clients before they realize they need them to render a page. The objective is to reduce the amount of time the browser has to wait for data to be transferred, and consequently total page load time. Our goal in this work is to quantify how much server push can actually reduce web page load times. We approach the problem from both theoretical and experimental perspectives. We start by deriving an upper bound for the load time reduction afforded by server push. Then we proceed to actually evaluate an idealized push implementation on the Alexa Top 100 global sites, against a non-push HTTP/2 baseline. Our results show a linear relationship between latency and the benefit of server push. Moreover, pages with taller dependency trees tend to benefit the most from it.

cs.NI↗

arXiv2022

Wi-Fi Assist: Enhancing Vehicular Wi-Fi Connectivity with an Infrastructure-driven Approach

Vehicles access the Internet via cellular networks, instead of Wi-Fi networks. This choice has been mostly justified by the ubiquitous coverage of cellular networks: Wi-Fi coverage has been shown to be inadequate in the past, even in urban areas. We argue that providing Internet connectivity to vehicles via Wi-Fi is worth a revisit. Motivated by improvements in Wi-Fi network coverage in recent years, we propose Wi-Fi Assist, an add-on to current Wi-Fi infrastructures which differs from existing solutions in two key ways: (1) it is heavily infrastructure-driven; and (2) defines an interface for low-latency cooperation between different WLAN service sets, managed by different service providers.

cs.NI↗

arXiv2020

Sketchy With a Chance of Adoption: Can Sketch-Based Telemetry Be Ready for Prime Time?

Sketching algorithms or sketches have emerged as a promising alternative to the traditional packet sampling-based network telemetry solutions. At a high level, they are attractive because of their high resource efficiency and accuracy guarantees. While there have been significant recent advances in various aspects of sketching for networking tasks, many fundamental challenges remain unsolved that are likely stumbling blocks for adoption. Our contribution in this paper is in identifying and formulating these research challenges across the ecosystem encompassing network operators, platform vendors/developers, and algorithm designers. We hope that these serve as a necessary fillip for the community to enable the broader adoption of sketch-based telemetry.

cs.NI↗

arXiv2019

PARI: A Probabilistic Approach to AS Relationships Inference

Over the last two decades, several algorithms have been proposed to infer the type of relationship between Autonomous Systems (ASes). While the recent works have achieved increasingly higher accuracy, there has not been a systematic study on the uncertainty of AS relationship inference. In this paper, we analyze the factors contributing to this uncertainty and introduce a new paradigm to explicitly model the uncertainty and reflect it in the inference result. We also present PARI, an exemplary algorithm implementing this paradigm, that leverages a novel technique to capture the interdependence of relationship inference across AS links.

cs.NI↗

arXiv2013

TVR - Tall Vehicle Relaying in Vehicular Networks

Vehicle-to-Vehicle (V2V) communication is a core technology for enabling safety and non-safety applications in next generation Intelligent Transportation Systems. Due to relatively low heights of the antennas, V2V communication is often influenced by topographic features, man-made structures, and other vehicles located between the communicating vehicles. On highways, it was shown experimentally that vehicles can obstruct the line of sight (LOS) communication up to 50 percent of the time; furthermore, a single obstructing vehicle can reduce the power at the receiver by more than 20 dB. Based on both experimental measurements and simulations performed using a validated channel model, we show that the elevated position of antennas on tall vehicles improves communication performance. Tall vehicles can significantly increase the effective communication range, with an improvement of up to 50 percent in certain scenarios. Using these findings, we propose a new V2V relaying scheme called Tall Vehicle Relaying (TVR) that takes advantage of better channel characteristics provided by tall vehicles. TVR distinguishes between tall and short vehicles and, where appropriate, chooses tall vehicles as next hop relays. We investigate TVR's system-level performance through a combination of link-level experiments and system-level simulations and show that it outperforms existing techniques.

cs.NI↗

arXiv2011

Can User-Level Probing Detect and Diagnose Common Home-WLAN Pathologies?

Common WLAN pathologies include low signal-to-noise ratio, congestion, hidden terminals or interference from non-802.11 devices and phenomena. Prior work has focused on the detection and diagnosis of such problems using layer-2 information from 802.11 devices and special-purpose access points and monitors, which may not be generally available. Here, we investigate a userlevel approach: is it possible to detect and diagnose 802.11 pathologies with strictly user-level active probing, without any cooperation from, and without any visibility in, layer-2 devices? In this paper, we present preliminary but promising results indicating that such diagnostics are feasible.

cs.NI↗