Just-in-time provisioning for cyber foraging

Ha, Kiryong; Pillai, Padmanabhan; Richter, Wolfgang; Abe, Yoshihisa; Satyanarayanan, Mahadev

doi:10.1145/2462456.2464451

Cited by 133 publications

(86 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Four main steps are involved for the service initiation time: bind the MD with the cloudlet, transfer the VM overlay, decompress the VM overlay, and apply the VM overlay to the base VM to launch the VM. To minimize the service initiation time, Ha, et al [15] applied four optimization techniques (Deduplication, Bridging the Semantic Gap, Pipelining, and Early Start) to minimize the delay significantly.…”

Section: Fog Computingmentioning

confidence: 99%

See 1 more Smart Citation

Nearby live virtual machine migration using cloudlets and multipath TCP

2016

View full text Add to dashboard Cite

A nearby virtual machine (VM) based cloudlet is proposed for mobile cloud computing (MCC) to enhance the performance of real-time resource-intensive mobile applications. Generally, when a mobile device (MD) discovers a cloudlet in the vicinity, it takes time to set up a VM inside the cloudlet before data offloading from the MD to the VM starts. The time between the discovery of the cloudlet and actual offloading of data is considered as the service initiation time. When multiple cloudlets are present in a nearby geographical location, initiating a service with each cloudlet may be frustrating for cloudlet users that moving from one location to another. In order to eliminate the delay caused by the service initiation time after moving away from the source cloudlet, this paper proposes a seamless live VM migration between neighbouring cloudlets. A seamless live VM migration is achieved with the prior knowledge of the migrating VM IP address in the destination cloudlet and more importantly with multipath TCP (MPTCP). We have performed a number of experiments to validate the proposed approach using Linux KVM hypervisor. The experimental results demonstrate the feasibility of the proposed approach and also show performance improvement. Specifically, there is almost zero downtime at the destination cloudlet after the migration is completed.

show abstract

Section: Fog Computingmentioning

confidence: 99%

“…Before analyzing the results as shown in Figs 12,13,14,15,16, and 17, some terms are defined below for clarity.…”

Section: Latency and Performance Analysismentioning

confidence: 99%

Nearby live virtual machine migration using cloudlets and multipath TCP

2016

View full text Add to dashboard Cite

show abstract

“…Gabriel [22,23] is a system that uses cloudlets for face and object recognition. These cloudlets run the object recognition pipeline and the client ships every frame to the cloudlet.…”

Section: Mobile Computation Offloadingmentioning

confidence: 99%

“…Offloading is a well-known idea [13,23,22], but in the context of continuous recognition, it must be applied with care because wireless network latencies are too high. For example, if it takes 700 milliseconds to transfer a frame and recognize its objects at a server (measured time on an LTE network), when the results arrive they will be over 20 frames old, and the located object may no longer be at the reported position.…”

Section: Introductionmentioning

confidence: 99%

Glimpse

Chen

Ravindranath

Deng

et al. 2015

Proceedings of the 13th ACM Conference on Embedded Networked Sensor Systems

279

View full text Add to dashboard Cite

Glimpse is a continuous, real-time object recognition system for camera-equipped mobile devices. Glimpse captures full-motion video, locates objects of interest, recognizes and labels them, and tracks them from frame to frame for the user. Because the algorithms for object recognition entail significant computation, Glimpse runs them on server machines. When the latency between the server and mobile device is higher than a frame-time, this approach lowers objectrecognition accuracy. To regain accuracy, Glimpse uses an active cache of video frames on the mobile device. A subset of the frames in the active cache are used to track objects on the mobile, using (stale) hints about objects that arrive from the server from time to time. To reduce network bandwidth usage, Glimpse computes trigger frames to send to the server for recognizing and labeling. Experiments with Android smartphones and Google Glass over Verizon, AT&T, and a campus Wi-Fi network show that with hardware face detection support (available on many mobile devices), Glimpse achieves precision between 96.4% to 99.8% for continuous face recognition, which improves over a scheme performing hardware face detection and server-side recognition without Glimpse's techniques by between 1.8-2.5×. The improvement in precision for face recognition without hardware detection is between 1.6-5.5×. For road sign recognition, which does not have a hardware detector, Glimpse achieves precision between 75% and 80%; without Glimpse, continuous detection is non-functional (0.2%-1.9% precision). Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions@acm.org. Categories and Subject Descriptors

show abstract

“…Systems using VMs "copy" entire applications from mobile devices to nearby infrastructure. Either the mobile application is executed entirely in the dedicated VM, such as in the VM-based cloudlets in [12] and [13] or the decision is made on a per-routine basis, per example in CloneCloud [14] by using profiling and code analysis. COMET [15] uses a Distributed Shared Memory (DSM) approach which allows for easy migration of individual application threads between VMs.…”

Section: Related Workmentioning

confidence: 99%

Allocation Algorithms for Autonomous Management of Collaborative Cloudlets

Bohez

Verbelen

Simoens

et al. 2014

2014 2nd IEEE International Conference on Mobile Cloud Computing, Services, and Engineering

View full text Add to dashboard Cite

Abstract-Mobile applications are evolving towards support for advanced interactivity and resource-demanding multimedia features. Mobile platforms are however struggling to cope with these new innovative application concepts, such as Augmented Reality, due to the inherent limitations on their available resources, such as CPU, memory and battery power. Offloading resource-intensive calculations to nearby infrastructure or devices, also known as cloudlets, has emerged as a viable alternative in offering interactive and resource-intensive applications to mobile users.This concept of resource sharing provides promising opportunities for collaborative scenarios in which not only data processing, but also the data itself are shared between multiple users. In this paper, we investigate the challenges posed by offloading collaborative mobile applications. We describe the problem of autonomous management of collaborative cloudlets and propose two heuristic algorithms, Simulated Annealing and Steepest Descent, in order to solve this optimization problem. We observe that these heuristics yield an effectiveness of more than 90% for execution times that are three orders of magnitude lower when compared to a guaranteed-optimal approach.

show abstract

Just-in-time provisioning for cyber foraging

Cited by 133 publications

References 53 publications

Nearby live virtual machine migration using cloudlets and multipath TCP

Nearby live virtual machine migration using cloudlets and multipath TCP

Glimpse

Allocation Algorithms for Autonomous Management of Collaborative Cloudlets

Contact Info

Product

Resources

About