Nonvolatile Binary CNN Accelerator with Extremely Low Standby Power using RRAM for IoT Applications

Cai, Yujie; Zhou, Keji; Xue, Xiaoyong; Wang, Mingyu; Zeng, Xiaoyang

doi:10.1109/asicon47005.2019.8983658

Cited by 1 publication

(2 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The development of mobile Artificial Intelligence (AI) and Machine Learning (ML) applications lead to the advancement of the hardware for user equipment (UE) [1]. Although nowadays, some AI applicat ions, e.g., ResNet [2] and MobileNet [3] may be handled by UEs locally, other computational-intensive applications, e.g., augmented reality (AR) [4] services may still require to be processed in the central cloud.…”

Section: Introductionmentioning

confidence: 99%

“…Fortunately, edge servers are deployed over the station (like Wi-Fi Hotspots), so the smart glasses can use them to accelerate the object recognition service by offloading complex DNN computations to a nearby edge server. The problems caused by the above instance includes: (1) The programs required to execute user tasks are not pre-installed on the edge servers. Thus the user needs to offload the program files to the edge server first; (2) It takes too much time to offload the whole program file at one time, so partial offloading strategy is much more practical.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Joint Program Partitioning and Resource Allocation for Completion Time Minimization in Multi-MEC Systems

Taizhou

Zhang

Wang

et al. 2022

IEEE Trans. Netw. Sci. Eng.

View full text Add to dashboard Cite

This paper considers a practical mobile edge computing (MEC) system, where edge server does not preinstall the program required to perform user offloaded computing tasks. A partial program offloading (PPO) scheme is proposed, which can divide a user program into two parts, where the first part is executed by the user itself and the second part is transferred to an edge server for remote execution. However, the execution of the latter part requires the results of the previous part (called intermediate result) as the input. We aim to minimize the overall time consumption of a multi-server MEC system to complete all user offloaded tasks. It is modeled as a mixed integer nonlinear programming (MINLP) problem which considers user-and-server association, program partitioning, and communication resource allocation in a joint manner. An effective algorithm is developed to solve the problem by exploiting its structural features. First, the task completion time of a single server is minimized given the computing workload and available resource. Then, the working time of the edge servers are balanced by updating user-and-server association and communication resource allocation. Numerical results show that significant performance improvement can be achieved by the proposed scheme.

show abstract