Retracted: Minimizing Memory Usage for Resource Constrained Devices using Deep Convolutional Neural Networks DOI
Yadvendra Singh, Ajay Singh

2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon), Journal Year: 2022, Volume and Issue: unknown, P. 1 - 6

Published: Oct. 16, 2022

The embedded design of IoT systems usually depend on resource limitations that include memory capacity, low power consumption and dependable cost. end nodes manage the limited devices, like edge servers. gateway modules link cloud-based system interconnect nodes, including sensors actuators. term "resource constrained device" refers to a mobile (wireless) device runs solely battery powered or wireless media offers confined set computational storage-based capabilities. Systems with funds offer an effective means computation maximum data output least amount input. Due fact they use less energy, these are typically economical. network application entry point is kind resource-constrained devices known as server. This article's analysis centered model based source energy uses CNN reduce characteristics. settings increase accuracy while lowering execution latency. Knowledge Distillation Method presented in order maintain greater computation. distillation method divides smaller trained sets into predictions from larger CNN. approach shrinks by appropriately reducing it. Comparable bigger CNN, anticipates outcomes actions. A simpler predicts roughly par Numerous applications machine learning natural-language processing, artificial intelligence, object recognition, neural net graphs, knowledge technique.

Language: Английский

Empowering WebAssembly with Thin Kernel Interfaces DOI

Arjun Ramesh,

Tianshu Huang, Ben L. Titzer

et al.

Published: March 26, 2025

Language: Английский

Citations

1
Research on WebAssembly Runtimes: A Survey DOI Open Access
Yixuan Zhang, Mugeng Liu, Haoyu Wang

et al.

ACM Transactions on Software Engineering and Methodology, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 23, 2025

WebAssembly (abbreviated as Wasm) was initially introduced for the Web and quickly extended its reach into various domains beyond Web. To create Wasm applications, developers can compile high-level programming languages binaries or manually write textual format of translate it by toolchain. Regardless whether is utilized within outside Web, execution supported runtime. Such a runtime provides secure, memory-efficient, sandboxed environment to execute binaries. This paper comprehensive survey research on runtimes with 103 collected papers related following traditional systematic literature review process. It characterizes existing studies from two different angles, including internal (Wasm design, testing, analysis) external (applying domains). also proposes future directions about runtimes.

Language: Английский

Citations

0
WebAssembly for Container Runtime: Are We There Yet? DOI Open Access
Mugeng Liu,

Haiyang Shen,

Yixuan Zhang

et al.

ACM Transactions on Software Engineering and Methodology, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 7, 2025

To pursue more efficient software deployment with containers, WebAssembly (abbreviated as Wasm) has long been regarded a promising alternative to native container runtime (such Docker container) due its features of secure memory sandbox, lightweight isolation, portability, and multi-language support. However, it remains unknown whether how much Wasm indeed brings benefits for containerized applications. fill the knowledge gap, this paper presents first measurement study on Wasm-based (i.e., by comparison standalone execution performance in terms startup, computation, system interface access, resource consumption. Surprisingly, we find that does not achieve better versus expected introduces significant overhead compared runtime. Through comparison, identify main causes degradation containers. Some stem from heavy containerization similar while others are inherently caused VMs WASI interface. Our findings can help developers, developers community improve efficiency utilizing runtime, ultimately optimizing performance.

Language: Английский

Citations

0
DrWASI : LLM-assisted Differential Testing for WebAssembly System Interface Implementations DOI Open Access
Yixuan Zhang, Ningyu He,

Jianting Gao

et al.

ACM Transactions on Software Engineering and Methodology, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 11, 2025

WebAssembly (Wasm) is an emerging binary format that serves as a compilation target for over 40 programming languages. Wasm runtimes provide execution environments enhance portability by abstracting away operating systems and hardware details. A key component in these the System Interface (WASI), which manages interactions with systems, like file operations. Considering critical role of runtimes, community has aimed to detect their implementation bugs. However, no work focused on WASI-specific bugs can affect original functionalities running binaries cause unexpected results. To fill void, we present DrWASI , first general-purpose differential testing framework WASI implementations. Our approach uses large language model generate seeds applies variant environment mutation strategies expand enrich test case corpus. We then perform across major runtimes. By leveraging dynamic static information collected during after execution, identify evaluation shows uncovered 33 unique bugs, all confirmed 7 fixed developers. This research represents pioneering step exploring promising yet under-explored area ecosystem, providing valuable insights stakeholders.

Language: Английский

Citations

0
Optimizing WebAssembly Bytecode for IoT Devices Using Deep Reinforcement Learning DOI
Kaijie Gong, R. X. Yang,

Haoyu Li

et al.

ACM Transactions on Internet Technology, Journal Year: 2025, Volume and Issue: unknown

Published: April 19, 2025

WebAssembly has shown promising potential on various IoT devices to achieve the desired features such as multi-language support and seamless device-cloud integration. The execution performance of bytecode is directly influenced by compilation sequences. While existing research explored optimization sequences for native code, these approaches are not suitable due its unique instruction format control flow graph structure. In this work, we propose WasmRL, a novel efficient deep reinforcement learning (DRL)-based compiler framework tailored bytecode. We conduct fine-grained analysis characteristics instructions associated flags. observe that same sequence may yield contrasting outcomes in code. Motivated our observation, introduce WebAssembly-specific DRL state representation simultaneously captures impact runtime performance. To enhance training efficiency model, tree-based action space refinement method. Furthermore, develop pluggable cross-platform strategy optimize across different devices. evaluate WasmRL extenssively PolybenchC, MiBench, Shootout public datasets real-world applications. Experimental results show: (1) model trained specific device achieves 1.4x/1.1x speedups over -O3 seen/unseen programs; (2) 1.21x/1.06x improvements respectively. code been available at https://github.com/CarrollAdmin/WasmRL.

Language: Английский

Citations

0
WBSan: WebAssembly Bug Detection for Sanitization and Binary-Only Fuzzing DOI

Wu Xiao,

Junzhou He,

Lein-Saing Huang

et al.

Published: April 22, 2025

Language: Английский

Citations

0
Case study of WebAssembly Runtimes for AI Applications on the Edge DOI

Saif eddine Khelifa,

Miloud Bagaa,

Ahmed Ouameur Messaoud

et al.

Published: Feb. 19, 2024

In the realm of Artificial Intelligence (AI), need for immediate response times has given rise to Cloud Edge Computing Continuum (CECC). This new paradigm, aided by emerging technologies, addresses latency and network delays while promoting portability, security, efficiency, thereby enhancing Quality Service (QoS). A noteworthy technology in this context is WebAssembly (Wasm), originally conceived amplify web performance. It transitioned CECC, primarily due key enablers like System Interface (Wasi) Wasm runtime. Besides offering heightened security through its sandboxing mechanism, WebAssembly's compact code paves way rapid cold start seamless migration AI applications. However, with nascent integration into several questions arise. Prominent among them efficiency deploying tasks binary format, particularly performance runtimes AI-centric potential factors affecting such executions. Addressing these queries, our study examines various deep-learning models on standalone runtimes. Our findings indicate that, smaller networks optimized parameters, approach native performance, presenting just a 1.1x overhead average. Contrarily, an extensive parameter set exhibited pronounced overheads. We also identified multiple factors, associated both run-times neural networks, insights future research endeavors.

Language: Английский

Citations

3
Empowering In-Browser Deep Learning Inference on Edge Through Just-In-Time Kernel Optimization DOI Creative Commons
Fucheng Jia, Shiqi Jiang, Ting Cao

et al.

Published: June 3, 2024

Web is increasingly becoming the primary platform to deliver AI services onto edge devices, making in-browser deep learning (DL) inference more prominent. Nevertheless, heterogeneity of combined with underdeveloped state hardware acceleration practices, hinders current from achieving its full performance potential on target devices.

Language: Английский

Citations

3
WebAssembly Beyond the Web: A Review for the Edge-Cloud Continuum DOI
Sangeeta Kakati, Mats Brorsson

Published: June 23, 2023

The cloud computing environment has changed over the past years, transitioning from a centralized architecture including big data centers to dispersed and heterogeneous that incorporates edge followed by device processing units. This transformation calls for cross-platform, interoperable solution, feature WebAssembly (Wasm) offers. Wasm can be used as compact effective representation of server-less functions or micro-services deployment at edge. In settings, where various hardware software systems might employed, this is especially crucial. Developers create applications operate on any Wasm-compatible without spending time worrying about platform-specific challenges using common runtime environment.In survey, we indicate main opportunities runtimes in edge-cloud continuum, such performance optimisation, security, interoperability with other programming languages platforms. We provide comprehensive overview current landscape outside web, possible standardization efforts best practices these runtimes, thus serving valuable resource researchers practitioners field.

Language: Английский

Citations

4
LUNAR: A Native Table Engine for Embedded Devices DOI Open Access
Xiaopeng Fan,

Song Yan,

Yuchen Huang

et al.

Published: June 13, 2023

Embedded systems have evolved tremendously in recent years. We perform a study on SQLite and find that the multiple layers of abstraction drastically reduce bandwidth utilization. To minimize loss I/O path, we propose Lunar, novel native table storage engine. Lunar performs cross-layer design across database file system to avoid pitfalls multi-layer while providing SQL-compatible APIs. It employs type-aware layout considers access patterns different data types. Then, designs variable-size allocator fragmentation optimize RAM usage. Further, considering limited resources embedded devices, modular architecture enables selecting modules demand. also offers optional consistency modes make trade-off between resource consumption consistency. Experiments show achieves higher utilization, outperforming state-of-the-art approaches consuming fewer resources.

Language: Английский

Citations

3