A deep reinforcement learning based offloading scheme in ad-hoc mobile clouds

Le, Duc Van; Tham, Chen-Khong

doi:10.1109/infcomw.2018.8406881

Cited by 67 publications

(30 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [ 17 ], the authors addressed the problem of optimal computation offloading of adhoc mobile applications to the cloud using cellular networks. They have used deep reinforcement learning to reach optimal offloading decisions.…”

Section: Related Workmentioning

confidence: 99%

“…Furthermore, the gain of energy saving from a computation perspective is compromised by extra energy dissipation due to communication. This topic is a typical research problem pertaining to adopting optimal computation offloading strategies in cloud robotics that were addressed with different approaches in recent works, including Game Theory [ 14 , 15 ], Markov Decision Processes [ 16 , 17 ], and computational intelligence [ 18 ].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DeepBrain: Experimental Evaluation of Cloud-Based Computation Offloading and Edge Computing in the Internet-of-Drones for Deep Learning Applications

Koubâa

Ammar

Alahdab

et al. 2020

Sensors

View full text Add to dashboard Cite

Unmanned Aerial Vehicles (UAVs) have been very effective in collecting aerial images data for various Internet-of-Things (IoT)/smart cities applications such as search and rescue, surveillance, vehicle detection, counting, intelligent transportation systems, to name a few. However, the real-time processing of collected data on edge in the context of the Internet-of-Drones remains an open challenge because UAVs have limited energy capabilities, while computer vision techniquesconsume excessive energy and require abundant resources. This fact is even more critical when deep learning algorithms, such as convolutional neural networks (CNNs), are used for classification and detection. In this paper, we first propose a system architecture of computation offloading for Internet-connected drones. Then, we conduct a comprehensive experimental study to evaluate the performance in terms of energy, bandwidth, and delay of the cloud computation offloading approach versus the edge computing approach of deep learning applications in the context of UAVs. In particular, we investigate the tradeoff between the communication cost and the computation of the two candidate approaches experimentally. The main results demonstrate that the computation offloading approach allows us to provide much higher throughput (i.e., frames per second) as compared to the edge computing approach, despite the larger communication delays.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

DeepBrain: Experimental Evaluation of Cloud-Based Computation Offloading and Edge Computing in the Internet-of-Drones for Deep Learning Applications

Koubâa

Ammar

Alahdab

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

“…Other studies on the offloading decision to an edge node have been conducted [10], [16], [21], [22], [24], [26], [27]. The most relevant work to our approach is ST-CODA [10].…”

Section: Related Work and Contributionmentioning

confidence: 99%

Time-Optimized Task Offloading Decision Making in Mobile Edge Computing

Alghamdi

Anagnostopoulos

Pezaros

2019

2019 Wireless Days (WD)

View full text Add to dashboard Cite

Mobile Edge Computing application domains such as vehicular networks, unmanned aerial vehicles, data analytics tasks at the edge and augmented reality have recently emerged. Under such domains, while mobile nodes are moving and have certain tasks to be offloaded to Edge Servers, choosing an appropriate time and an ideally suited server to guarantee the quality of service can be challenging. We tackle the offloading decision making problem by adopting the principles of Optimal Stopping Theory to minimize the execution delay in a sequential decision manner. A performance evaluation is provided by using real data sets compared with the optimal solution. The results show that our approach significantly minimizes the execution delay for task execution and the results are very close to the optimal solution.Index Terms-Mobile edge computing, tasks offloading, optimal stopping theory, sequential decision making.To deal with the disadvantages of MCC, the Mobile Edge Computing (MEC) paradigm has emerged. The rationale architecture of this concept is to offer cloud services closer to mobile devices by placing many data-centres at the edge of the network. The network edge refers to different places e.g., mobile network at the Base Station (BS), or indoor places such Wi-Fi and 3G/4G access points [17].Motivation & Challenge: An essential use case of MEC is the computing task/data offloading. Computation offloading is the task of sending a computation task and data to a remote server for delegating this computation [1]. As the new emerging applications require intensive computation processes, computation offloading provides a solution to overcome the limitation of the mobile device. Examples of applications that can benefit from computing offloading are mobile AR [6], gaming, Internet of Things (IoT) applications, data analytics tasks at the edge [15], VN [25] and Unmanned Aerial Vehicles (UAV) [28] . A study showed the benefit of using offloading for the Percipio AR application on a real MEC testbed [7]. The study showed that the computation offloading reduces latency up to 88% and energy consumption of mobile devices up to 93%.Computation offloading faces several challenges, the most significant ones being: (i) the decision of when to offload tasks/data to a MEC server and (ii) mobility patterns/behaviour of the users in such MEC environments. The decision making of tasks/data offloading is of high importance as it is expected to directly affect the Quality of Service (QoS) of the user application including the inherent latency due to the offloading process. Therefore, different parameters have to be considered when to decide to offload tasks and/or data including: the current MEC server load and the transmission / communication status between the mobile node and the MEC server. The decision can be spatial or temporal as stated in [10]. The spatial decision refers to realising the computing tasks locally at the mobile device, in the cloud, or at the edge server [11]. The temporal decision refers to a situation where it is a...

show abstract

“…But the proposed model is based on map-reduce which is implied for data-intensive applications. Deep reinforcement learning centered offloading procedure for the consumer to obtain the ideal offloading strategy in an ad-hoc itinerant cloud is proposed in [18]. The depositing problem is voiced as an MDP with the core objective of creating an optimal divesting action choice at each structure state as such that the convenience obtained by task completing is maximized by minimizing the energy usage, computing delay [18].…”

Section: Related Workmentioning

confidence: 99%

“…Deep reinforcement learning centered offloading procedure for the consumer to obtain the ideal offloading strategy in an ad-hoc itinerant cloud is proposed in [18]. The depositing problem is voiced as an MDP with the core objective of creating an optimal divesting action choice at each structure state as such that the convenience obtained by task completing is maximized by minimizing the energy usage, computing delay [18]. This is an inefficient and ineffective model for the case of vehicular networks since a large number of data transmission sessions are not achievable in practical scenarios due to irregular access mechanism which is used by Wireless LAN standards like 802.11p.…”

Section: Related Workmentioning

confidence: 99%

Scalable and Energy Efficient Task Offloading Schemes for Vehicular Cloud Computing

Pasha¹,

Khan²

2018

IJCNC

View full text Add to dashboard Cite

Smart vehicles of today on road are equipped with advanced computational units, multiple communication technologies, intelligent sensing platforms, and human-computer interaction devices which utilize Vehicular Edge Networks to support services offered by the remote cloud. This being named as Opportunistic Vehicular Edge Computing recently, has the possibility to supplement the services provided by the Edge gadgets. Many Vehicular Edge Computing architectures have been proposed as of late which support task offloading. One among the premier difficulties in these networks is efficiently utilizing the resources available at the vehicular nodes. The present work uses APEATOVC, a conveyed and versatile protocol for economical, efficient and effective task offloading in these networks which address the adaptability of vehicular clouds. The results obtained by extensive simulations are presented to assess and contrast its performance with existing protocols.

show abstract

A deep reinforcement learning based offloading scheme in ad-hoc mobile clouds

Cited by 67 publications

References 12 publications

DeepBrain: Experimental Evaluation of Cloud-Based Computation Offloading and Edge Computing in the Internet-of-Drones for Deep Learning Applications

DeepBrain: Experimental Evaluation of Cloud-Based Computation Offloading and Edge Computing in the Internet-of-Drones for Deep Learning Applications

Time-Optimized Task Offloading Decision Making in Mobile Edge Computing

Scalable and Energy Efficient Task Offloading Schemes for Vehicular Cloud Computing

Contact Info

Product

Resources

About