Piezoelectric energy (PE) harvesters can harvest vibration energy and convert it into electrical energy, which provides a promising solution to supply sustainable power for wireless sensor networks (WSN) applications....
详细信息
This paper presents a novel multi-bit dualpolarization frequency-reconfigurable intelligent surface optimized for sub-6 GHz applications. For the TE mode, phase variation is observed from 1.75 GHz to 2.1 GHz, with the...
详细信息
Generative AI like the Large Language Models (LLMs) has become more available for the general consumer in recent years. Publicly available services, e.g., ChatGPT, perform token generation on networked cloud server ha...
详细信息
ISBN:
(数字)9798331504847
ISBN:
(纸本)9798331504854
Generative AI like the Large Language Models (LLMs) has become more available for the general consumer in recent years. Publicly available services, e.g., ChatGPT, perform token generation on networked cloud server hardware, effectively removing the hardware entry cost for end users. However, the reliance on network access for these services, privacy and security risks involved, and sometimes the needs of the application make it necessary to run LLMs locally on edge devices. A significant amount of research has been done on optimization of LLMs and other transformer-based models on non-networked, resource-constrained devices, but they typically target older hardware. Our research intends to provide a ‘baseline’ characterization of more recent commercially available embedded hardware for LLMs, and to provide a simple utility to facilitate batch testing LLMs on recent Jetson hardware. We focus on the latest line of NVIDIA Jetson devices (Jetson Orin), and a set of publicly available LLMs (Pythia) ranging between 70 million and 1.4 billion parameters. Through detailed experimental evaluation with varying software and hardware parameters, we showcase tradeoff spaces and optimization choices. Additionally, we design our testing structure to facilitate further research that involves performing batch LLM testing on NVIDIA Jetson hardware.
The successful and complete downlink transmission of critical data from an unmanned aerial vehicle (UAV) base station, such as control command and intelligence information, is essential for ground users to perform spe...
详细信息
Deep learning and big data analysis are among the most important research topics in the fields of biomedical applications and digital healthcare. With the fast development of artificial intelligence (AI) and Internets...
详细信息
360-degree video has become increasingly popular in content consumption. However, finding the viewing direction for important content within each frame poses a significant challenge. Existing approaches rely on either...
详细信息
Linters are static analysis tools supporting developers by automatically detecting possible code errors, bad coding practices, violations of coding conventions, and styling issues while offering actionable advice on h...
详细信息
Edge computing, an extension of cloud and IoT technologies, introduces unique challenges for intrusion-detection systems (IDS). This study explores IDS architecture in the context of edge computing. We proposed a spec...
详细信息
The compact key sizes and the low computational latency of the Elliptic Curve Cryptography (ECC) family of curves sparked high interest in their integration into network protocols. The recently suggested Curve448, ass...
详细信息
The idea of Smart Cities (SCs) was developed to digitize conventional urban living areas and redevelop using digital equipment to enhance lifestyle and resident security. Smart cities cover a wide range of application...
详细信息
暂无评论