arXiv subjects

Sangtae Ha

Publications and source records attributed to Sangtae Ha.

17 recordsLinked to original sources

arXiv2026

Practical Post-Quantum Cryptography for Bandwidth Constrained or Non-Terrestrial Networks, and Power Constrained Devices

Post-quantum (PQ) cryptographic algorithms, particularly for authentication, are more complex than classical algorithms and require larger certificates, signatures, and keys. Establishing a PQ-secure network connection increases bandwidth, memory, computation time, and energy consumption. These costs are especially severe in Non-Terrestrial Networks (NTNs), where long propagation delays, intermittent connectivity, constrained link budgets, satellite handovers, limited terminal resources, and bandwidth-constrained satellite-to-ground links amplify the overhead of certificate-based PQ authentication. Consequently, applications such as key rotation and key management may be unable to achieve acceptable handshake reliability or support NIST PQ Security Categories above Category 1. Similar limitations affect low-power IoT devices and bandwidth- or energy-constrained terrestrial networks, where PQ authentication may restrict devices to Category 1 security or prevent ambient-powered endpoints from supporting PQ authentication altogether. This paper investigates an alternative cryptographic framework that replaces PQ digital certificates with shared secret keys (SSKs). The framework leverages shared-secret ecosystems that do not rely on asymmetric key distribution, such as 5G/6G, and combines a Key Distribution Center (KDC) (e.g., Kerberos) with a preshared-key PQ handshake (e.g., DTLS-PSK) and ephemeral PQ key establishment (e.g., ML-KEM). Compared with certificate-based PQ authentication (e.g., ML-DSA), the proposed approach reduces handshake bandwidth, endpoint RAM, computation time, and energy use while preserving PQ-secure AEAD, including forward secrecy and replay resistance. Applications include NTN-based key rotation and management, uncrewed aerial vehicle (UAV) command-and-control systems, embedded medical sensors, and supply-chain monitoring and asset-tracking platforms.

cs.CR↗

arXiv2026

HARMONY: Bridging the Personalization-Generalization Gap by Mitigating Representation Skew in Heterogeneous Split Federated Learning

Mobile devices face diverse resource constraints and non-IID data class distributions, requiring fast on-device inference for local in-distribution (ID) classes and on-demand remote support for client-specific out-of-distribution (OOD) classes. Hybrid split federated learning (Hybrid SFL) couples personalized client-side front ends (supporting early exit) with a generalized server-side backend for fallback inference, balancing accuracy and cost. However, under client architectural heterogeneity, the existing hybrid SFL suffers from representation skew, where features from customized extractors fail to align in the shared space, leading to a sharp degradation in the server model responsible for OOD prediction. We propose HARMONY, the first hybrid SFL framework to support heterogeneous client architectures. HARMONY modifies meta-learning to simulate diverse extractors across parameters and architectures, and to learn to personalize. To mitigate representation skew, HARMONY conducts server-side contrastive learning to align extracted features, neither sacrificing clients' personalization nor sharing raw labels. Compared to the state of the art across multiple datasets and model families, HARMONY improves test accuracy by up to 43.0%/28.3% without/with OOD, respectively, while maintaining acceptable latency.

cs.LG↗

arXiv2024

LLMem: Estimating GPU Memory Usage for Fine-Tuning Pre-Trained LLMs

Fine-tuning pre-trained large language models (LLMs) with limited hardware presents challenges due to GPU memory constraints. Various distributed fine-tuning methods have been proposed to alleviate memory constraints on GPU. However, determining the most effective method for achieving rapid fine-tuning while preventing GPU out-of-memory issues in a given environment remains unclear. To address this challenge, we introduce LLMem, a solution that estimates the GPU memory consumption when applying distributed fine-tuning methods across multiple GPUs and identifies the optimal method. We conduct GPU memory usage estimation prior to fine-tuning, leveraging the fundamental structure of transformer-based decoder models and the memory usage distribution of each method. Experimental results show that LLMem accurately estimates peak GPU memory usage on a single GPU, with error rates of up to 1.6%. Additionally, it shows an average error rate of 3.0% when applying distributed fine-tuning methods to LLMs with more than a billion parameters on multi-GPU setups.

cs.AI↗

arXiv2023

FIS-ONE: Floor Identification System with One Label for Crowdsourced RF Signals

Floor labels of crowdsourced RF signals are crucial for many smart-city applications, such as multi-floor indoor localization, geofencing, and robot surveillance. To build a prediction model to identify the floor number of a new RF signal upon its measurement, conventional approaches using the crowdsourced RF signals assume that at least few labeled signal samples are available on each floor. In this work, we push the envelope further and demonstrate that it is technically feasible to enable such floor identification with only one floor-labeled signal sample on the bottom floor while having the rest of signal samples unlabeled. We propose FIS-ONE, a novel floor identification system with only one labeled sample. FIS-ONE consists of two steps, namely signal clustering and cluster indexing. We first build a bipartite graph to model the RF signal samples and obtain a latent representation of each node (each signal sample) using our attention-based graph neural network model so that the RF signal samples can be clustered more accurately. Then, we tackle the problem of indexing the clusters with proper floor labels, by leveraging the observation that signals from an access point can be detected on different floors, i.e., signal spillover. Specifically, we formulate a cluster indexing problem as a combinatorial optimization problem and show that it is equivalent to solving a traveling salesman problem, whose (near-)optimal solution can be found efficiently. We have implemented FIS-ONE and validated its effectiveness on the Microsoft dataset and in three large shopping malls. Our results show that FIS-ONE outperforms other baseline algorithms significantly, with up to 23% improvement in adjusted rand index and 25% improvement in normalized mutual information using only one floor-labeled signal sample.

cs.NI↗

arXiv2023

Semi-supervised Learning with Network Embedding on Ambient RF Signals for Geofencing Services

In applications such as elderly care, dementia anti-wandering and pandemic control, it is important to ensure that people are within a predefined area for their safety and well-being. We propose GEM, a practical, semi-supervised Geofencing system with network EMbedding, which is based only on ambient radio frequency (RF) signals. GEM models measured RF signal records as a weighted bipartite graph. With access points on one side and signal records on the other, it is able to precisely capture the relationships between signal records. GEM then learns node embeddings from the graph via a novel bipartite network embedding algorithm called BiSAGE, based on a Bipartite graph neural network with a novel bi-level SAmple and aggreGatE mechanism and non-uniform neighborhood sampling. Using the learned embeddings, GEM finally builds a one-class classification model via an enhanced histogram-based algorithm for in-out detection, i.e., to detect whether the user is inside the area or not. This model also keeps on improving with newly collected signal records. We demonstrate through extensive experiments in diverse environments that GEM shows state-of-the-art performance with up to 34% improvement in F-score. BiSAGE in GEM leads to a 54% improvement in F-score, as compared to the one without BiSAGE.

cs.NI↗

arXiv2022

A Large-Scale Study of a Sleep Tracking and Improving Device with Closed-loop and Personalized Real-time Acoustic Stimulation

Various intervention therapies ranging from pharmaceutical to hi-tech tailored solutions have been available to treat difficulty in falling asleep commonly caused by insomnia in modern life. However, current techniques largely remain ill-suited, ineffective, and unreliable due to their lack of precise real-time sleep tracking, in-time feedback on the therapies, an ability to keep people asleep during the night, and a large-scale effectiveness evaluation. Here, we introduce a novel sleep aid system, called Earable, that can continuously sense multiple head-based physiological signals and simultaneously enable closed-loop auditory stimulation to entrain brain activities in time for effective sleep promotion. We develop the system in a lightweight, comfortable, and user-friendly headband with a comprehensive set of algorithms and dedicated own-designed audio stimuli. We conducted multiple protocols from 883 sleep studies on 377 subjects (241 women, 119 men) wearing either a gold-standard device (PSG), Earable, or both concurrently. We demonstrate that our system achieves (1) a strong correlation (0.89 +/- 0.03) between the physiological signals acquired by Earable and those from the gold-standard PSG, (2) an 87.8 +/- 5.3% agreement on sleep scoring using our automatic real-time sleep staging algorithm with the consensus scored by three sleep technicians, and (3) a successful non-pharmacological stimulation alternative to effectively shorten the duration of sleep falling by 24.1 +/- 0.1 minutes. These results show that the efficacy of Earable exceeds existing techniques in intentions to promote fast falling asleep, track sleep state accurately, and achieve high social acceptance for real-time closed-loop personalized neuromodulation-based home sleep care.

eess.SY↗

arXiv2022

GRAFICS: Graph Embedding-based Floor Identification Using Crowdsourced RF Signals

We study the problem of floor identification for radiofrequency (RF) signal samples obtained in a crowdsourced manner, where the signal samples are highly heterogeneous and most samples lack their floor labels. We propose GRAFICS, a graph embedding-based floor identification system. GRAFICS first builds a highly versatile bipartite graph model, having APs on one side and signal samples on the other. GRAFICS then learns the low-dimensional embeddings of signal samples via a novel graph embedding algorithm named E-LINE. GRAFICS finally clusters the node embeddings along with the embeddings of a few labeled samples through a proximity-based hierarchical clustering, which eases the floor identification of every new sample. We validate the effectiveness of GRAFICS based on two large-scale datasets that contain RF signal records from 204 buildings in Hangzhou, China, and five buildings in Hong Kong. Our experiment results show that GRAFICS achieves highly accurate prediction performance with only a few labeled samples (96% in both micro- and macro-F scores) and significantly outperforms several state-of-the-art algorithms (by about 45% improvement in micro-F score and 53% in macro-F score).

cs.NI↗

arXiv2022

A Fresh Look at ECN Traversal in the Wild

The Explicit Congestion Notification (ECN) field has taken on new importance due to Low Latency, Low Loss, and Scalable throughput (L4S) technology designed for extremely latency-sensitive applications (such as cloud games and cloud-rendered VR/AR). ECN and L4S need to be supported by the client and server but also all devices in the network path. We have identified that "ECN bleaching", where an intermediate network device clears or "bleaches" the ECN flags, occurs and quantified how often that happens, why it happens and identified where in the network it happens. In this research, we conduct a comprehensive measurement study on end-to-end traversal of the ECN field using probes deployed on the Internet across different varied clients and servers. Using these probes, we identify and locate instances of ECN bleaching on various network paths on the Internet. In our six months of measurements, conducted in late 2021 and early 2022, we found the prevalence varied considerably from network to network. One cloud provider and two cellular providers bleach the ECN field as a matter of policy. Of the rest, we found 1,112 out of 129,252 routers, 4.17% of paths we measured showed ECN bleaching.

cs.NI↗

arXiv2022

CPrune: Compiler-Informed Model Pruning for Efficient Target-Aware DNN Execution

Mobile devices run deep learning models for various purposes, such as image classification and speech recognition. Due to the resource constraints of mobile devices, researchers have focused on either making a lightweight deep neural network (DNN) model using model pruning or generating an efficient code using compiler optimization. Surprisingly, we found that the straightforward integration between model compression and compiler auto-tuning often does not produce the most efficient model for a target device. We propose CPrune, a compiler-informed model pruning for efficient target-aware DNN execution to support an application with a required target accuracy. CPrune makes a lightweight DNN model through informed pruning based on the structural information of subgraphs built during the compiler tuning process. Our experimental results show that CPrune increases the DNN execution speed up to 2.73x compared to the state-of-the-art TVM auto-tune while satisfying the accuracy requirement.

cs.LG↗

arXiv2022

TVConv: Efficient Translation Variant Convolution for Layout-aware Visual Processing

As convolution has empowered many smart applications, dynamic convolution further equips it with the ability to adapt to diverse inputs. However, the static and dynamic convolutions are either layout-agnostic or computation-heavy, making it inappropriate for layout-specific applications, e.g., face recognition and medical image segmentation. We observe that these applications naturally exhibit the characteristics of large intra-image (spatial) variance and small cross-image variance. This observation motivates our efficient translation variant convolution (TVConv) for layout-aware visual processing. Technically, TVConv is composed of affinity maps and a weight-generating block. While affinity maps depict pixel-paired relationships gracefully, the weight-generating block can be explicitly overparameterized for better training while maintaining efficient inference. Although conceptually simple, TVConv significantly improves the efficiency of the convolution and can be readily plugged into various network architectures. Extensive experiments on face recognition show that TVConv reduces the computational cost by up to 3.1x and improves the corresponding throughput by 2.3x while maintaining a high accuracy compared to the depthwise convolution. Moreover, for the same computation cost, we boost the mean accuracy by up to 4.21%. We also conduct experiments on the optic disc/cup segmentation task and obtain better generalization performance, which helps mitigate the critical data scarcity issue. Code is available at https://github.com/JierunChen/TVConv.

cs.CV↗

arXiv2018

Elasticizing Linux via Joint Disaggregation of Memory and Computation

In this paper, we propose a set of operating system primitives which provides a scaling abstraction to cloud applications in which they can transparently be enabled to support scaled execution across multiple physical nodes as resource needs go beyond that available on a single machine. These primitives include stretch, to extend the address space of an application to a new node, push and pull, to move pages between nodes as needed for execution and optimization, and jump, to transfer execution in a very lightweight manner between nodes. This joint disaggregation of memory and computing allows for transparent elasticity, improving an application's performance by capitalizing on the underlying dynamic infrastructure without needing an application re-write. We have implemented these primitives in a Linux 2.6 kernel, collectively calling the extended operating system, ElasticOS. Our evaluation across a variety of algorithms shows up to 10x improvement in performance over standard network swap.

cs.DC↗

arXiv2017

FluidMem: Memory as a Service for the Datacenter

Disaggregating resources in data centers is an emerging trend. Recent work has begun to explore memory disaggregation, but suffers limitations including lack of consideration of the complexity of cloud-based deployment, including heterogeneous hardware and APIs for cloud users and operators. In this paper, we present FluidMem, a complete system to realize disaggregated memory in the datacenter. Going beyond simply demonstrating remote memory is possible, we create an entire Memory as a Service. We define the requirements of Memory as a Service and build its implementation in Linux as FluidMem. We present a performance analysis of FluidMem and demonstrate that it transparently supports remote memory for standard applications such as MongoDB and genome sequencing applications.

cs.OS↗

arXiv2017

AMUSE: Empowering Users for Cost-Aware Offloading with Throughput-Delay Tradeoffs

To cope with recent exponential increases in demand for mobile data, wireless Internet service providers (ISPs) are increasingly changing their pricing plans and deploying WiFi hotspots to offload their mobile traffic. However, these ISP-centric approaches for traffic management do not always match the interests of mobile users. Users face a complex, multi-dimensional tradeoff between cost, throughput, and delay in making their offloading decisions: while they may save money and receive a higher throughput by waiting for WiFi access, they may not wait for WiFi if they are sensitive to delay. To navigate this tradeoff, we develop AMUSE (Adaptive bandwidth Management through USer-Empowerment), a functional prototype of a practical, cost-aware WiFi offloading system that takes into account a user's throughput-delay tradeoffs and cellular budget constraint. Based on predicted future usage and WiFi availability, AMUSE decides which applications to offload to what times of the day. Since nearly all traffic flows from mobile devices are TCP flows, we introduce a new receiver-side bandwidth allocation mechanism to practically enforce the assigned rate of each TCP application. Thus, AMUSE users can optimize their bandwidth rates according to their own cost-throughput-delay tradeoff without relying on support from different apps' content servers. Through a measurement study of 20 smartphone users' traffic usage traces, we observe that though users already offload a large amount of some application types, our framework can offload a significant additional portion of users' cellular traffic. We implement AMUSE and evaluate its effectiveness with 3G and WiFi usage data obtained from a trial with 37 mobile users. Our results show that AMUSE improves user utility, when compared with AMUSE, other offloading algorithms yield 14\% and 27\% lower user utilities for light and heavy users, respectively.

cs.NI↗

arXiv2015

The Cloud Needs a Reputation System

Today's cloud apps are built from many diverse services that are managed by different parties. At the same time, these parties, which consume and/or provide services, continue to rely on arcane static security and entitlements models. In this paper, we introduce Seit, an inter-tenant framework that manages the interactions between cloud services. Seit is a software-defined reputation-based framework. It consists of two primary components: (1) a set of integration and query interfaces that can be easily integrated into cloud and service providers' management stacks, and (2) a controller that maintains reputation information using a mechanism that is adaptive to the highly dynamic environment of the cloud. We have fully implemented Seit, and integrated it into an SDN controller, a load balancer, a cloud service broker, an intrusion detection system, and a monitoring framework. We evaluate the efficacy of Seit using both an analytical model and a Mininet-based emulated environment. Our analytical model validate the isolation and stability properties of Seit. Using our emulated environment, we show that Seit can provide improved security by isolating malicious tenants, reduced costs by adapting the infrastructure without compromising security, and increased revenues for high quality service providers by enabling reputation to impact discovery.

cs.NI↗

arXiv2014

Offering Supplementary Network Technologies: Adoption Behavior and Offloading Benefits

To alleviate the congestion caused by rapid growth in demand for mobile data, wireless service providers (WSPs) have begun encouraging users to offload some of their traffic onto supplementary network technologies, e.g., offloading from 3G or 4G to WiFi or femtocells. With the growing popularity of such offerings, a deeper understanding of the underlying economic principles and their impact on technology adoption is necessary. To this end, we develop a model for user adoption of a base technology (e.g., 3G) and a bundle of the base plus a supplementary technology (e.g., 3G + WiFi). Users individually make their adoption decisions based on several factors, including the technologies' intrinsic qualities, negative congestion externalities from other subscribers, and the flat access rates that a WSP charges. We then show how these user-level decisions translate into aggregate adoption dynamics and prove that these converge to a unique equilibrium for a given set of exogenously determined system parameters. We fully characterize these equilibria and study adoption behaviors of interest to a WSP. We then derive analytical expressions for the revenue-maximizing prices and optimal coverage factor for the supplementary technology and examine some resulting non-intuitive user adoption behaviors. Finally, we develop a mobile app to collect empirical 3G/WiFi usage data and numerically investigate the profit-maximizing adoption levels when a WSP accounts for its cost of deploying the supplemental technology and savings from offloading traffic onto this technology.

cs.NI↗

arXiv2013

Mind Your Own Bandwidth: An Edge Solution to Peak-hour Broadband Congestion

Motivated by recent increases in network traffic, we propose a decentralized network edge-based solution to peak-hour broadband congestion that incentivizes users to moderate their bandwidth demands to their actual needs. Our solution is centered on smart home gateways that allocate bandwidth in a two-level hierarchy: first, a gateway purchases guaranteed bandwidth from the Internet Service Provider (ISP) with virtual credits. It then self-limits its bandwidth usage and distributes the bandwidth among its apps and devices according to their relative priorities. To this end, we design a credit allocation and redistribution mechanism for the first level, and implement our gateways on commodity wireless routers for the second level. We demonstrate our system's effectiveness and practicality with theoretical analysis, simulations and experiments on real traffic. Compared to a baseline equal sharing algorithm, our solution significantly improves users' overall satisfaction and yields a fair allocation of bandwidth across users.

cs.NI↗

arXiv2013

A Survey of Smart Data Pricing: Past Proposals, Current Plans, and Future Trends

Traditionally, network operators have used simple flat-rate broadband data plans for both wired and wireless network access. But today, with the popularity of mobile devices and exponential growth of apps, videos, and clouds, service providers are gradually moving towards more sophisticated pricing schemes. This decade will therefore likely witness a major change in the ways in which network resources are managed, and the role of economics in allocating these resources. This survey reviews some of the well-known past broadband pricing proposals (both static and dynamic), including their current realizations in various consumer data plans around the world, and discusses several research problems and open questions. By exploring the benefits and challenges of pricing data, this paper attempts to facilitate both the industrial and the academic communities' efforts in understanding the existing literature, recognizing new trends, and shaping an appropriate and timely research agenda.

cs.NI↗