Huber Flores

Mobile code offloading: from concept to practice and beyond

The emerging mobile cloud has expanded the horizon of application development and deployment with techniques such as code offloading. While offloading has been widely considered for saving energy and increasing responsiveness of mobile devices, the technique still faces many challenges pertaining to practical usage. In this article, we adopt a systemic approach for analyzing the components of a generic code offloading architecture. Based on theoretical and experimental analysis, we identify the key limitations for code offloading in practice and then propose solutions to mitigate these limitations. We develop a generic architecture to evaluate the proposed solutions. The results provide insights regarding the evolution and deployment of code offloading.

Adaptive code offloading for mobile cloud applications: exploiting fuzzy sets and evidence-based learning

Mobile cloud computing is arising as a prominent domain that is seeking to bring the massive advantages of the cloud to the resource constrained smartphones, by following a delegation or offloading criteria. In a delegation model, a mobile device consumes services from multiple clouds by efficiently utilizing solutions like middleware. In the offloading model, a mobile application is partitioned and analyzed so that the most computational expensive operations at code level can be identified and offloaded for remote processing. While code offloading is studied extensively for the development of mobile cloud applications, much of the advantages of cloud computing are still left unexploited and poorly considered in these approaches. Cloud computing may introduce many other dynamic variables like performance metrics, parallelization of tasks, elasticity etc., to current code offloading models that could affect the overall offloading decision process. To address this, we propose a fuzzy decision engine for code offloading, that considers both mobile and cloud variables. The cloud parameters and rules are introduced asynchronously to the mobile, using notification services. The paper also proposes a strategy to enrich the offloading decision process with evidence-based learning methods, by exploiting cloud processing capabilities over code offloading traces.

Computational Offloading or Data Binding? Bridging the Cloud Infrastructure to the Proximity of the Mobile User

Mobile and cloud computing are converging as the prominent technologies that are leading the change to the post personal computing (PC) era. Computational offloading and data binding are the core techniques that foster to elastically augment the capabilities of low-power devices, such as smartphones. Mobile applications may be bonded to cloud resources by following a task delegation or code offloading criteria. In a delegation model, a handset can utilize the cloud in a service-oriented manner to delegate asynchronously a resource-intensive mobile task by direct invocation of the service. In contrast, in an offloading model, a mobile application is partitioned and analyzed so that the most computational expensive operations at code level can be identified and offloaded to a remote cloud-based surrogate. We compared in this paper, the mobile cloud computing models for offloading and delegation. We utilized our own frameworks for computational offloading and data binding in the analysis. While in principle, offloading and delegation are viable methods to augment the capabilities of the mobile devices with cloud power, they enrich the mobile applications from different perspectives at diverse computational scales.

Mobile code offloading: should it be a local decision or global inference?

Mobile and cloud computing are converging as the prominent technologies that are leading the change to the post-pc era. Mobile devices are looking towards cloud-aware techniques, driven by their growing interest to provide ubiquitous PC-like functionality to mobile users. These functionalities mainly target at increasing storage and computational capabilities. On one hand, storage limitations of the devices have been overcome by many cloud services provided in the Internet, which are built under different protocols (e.g. SyncML). On the other hand, binding computational cloud services such as Amazon EC2 to low-power devices such as smartphones have been proven feasible with latest mobile technologies [2, 1], mostly due to virtualization technologies and their synchronization primitives, enabling transparent migration and execution of intermediate code. Moreover, multiple research works have proposed different offloading strategies to empower the mobile application with cloud resources. Most of these solutions try to overcome the problem by granting the mobile, a local context logic, which is used to decide whether a mobile component is offloaded or not. However, given this context, we can argue that much of the advantages of cloud computing are left unexploited and poorly considered. A cloud does not just mean a virtual machine or a pool of servers which are accessible from the Internet. It has its own intrinsic features like utility computing, fine-grained billing, parallelization of tasks, etc. So an ideal mobile cloud framework should take advantage of several of these features. Cloud computing may introduce many other dynamic variables to current code offloading models that could affect the overall offloading decision process. For instance, performance metrics (e.g. CPU load) of the instance/cluster at the cloud may be useful by the mobile in order to determine 1) whether a server is not that busy so that it can handle an incoming request and 2) a dynamic execution plan that allows to parallelize mobile operations in a single machine with multiple cores or in different machines, each with a single core. Consequently, a code offloading model should not just target mobility aspects, but also target oscillating changes in cloud infrastructure. On the basis of these assumptions, we envisioned in this work, an ”Evidence-based mobile code offloading approach”[3] that enables to transform raw code offloading traces, which are obtained by the huge amount of devices that connect

Mobile Sensor Data Classification for Human Activity Recognition using MapReduce on Cloud

Abstract Mobiles are equipped with different sensors like accelerometer, magnetic field, and air pressure meter, which help in the process of extracting context of the user like location, situation etc. However, processing the extracted sensor data is generally a resource intensive task, which can be offloaded to the public cloud from mobiles. This paper specifically targets at extracting useful information from the accelerometer sensor data. The paper proposes the utilization of parallel computing using MapReduce on the cloud for training and recognizing human activities based on classifiers that can easily scale in performance and accuracy. The sensor data is extracted from the mobile, offloaded to the cloud and processed using three different classification algorithms, Iterative Dichotomizer 3, Naive Bayes Classifier and K-Nearest-Neighbors. The MapReduce based algorithms are mentioned in detail along with one of their performance on Amazon cloud. The recognized activities can be used in mobile applications like our Zompopo that utilizes the information in creating an intelligent calendar.

Mobile cloud messaging supported by XMPP primitives

The increasing demand of the smartphones for processing power, storage space and energy saving is leading the rapid adaption of the mobile cloud computing (MCC) domain. Asynchronous mobile cloud communication is a two way process which in most cases is managed via REST (from mobile to cloud) and push technologies (from cloud to mobile). There are several push notification mechanisms provided by the cloud vendors and mobile operators for delivering messages to the mobile applications (e.g. GCM, APNS, MPNS, etc.). However, these mechanisms have certain constraints such as being platform specific, limited message payload and the number of the messages that can be sent to a single handset. This paper explored the possibility of implementing a messaging framework that can be used as open notification mechanism running at the cloud for handling asynchronous delegation of mobile tasks to cloud. The framework is developed by extending a XMPP-based IM infrastructure, which can be integrated with any application server and reconfigured on demand as the mobile workload grows or shrinks. The detailed analysis of the framework shows that it is possible to send messages to the smartphones with IM, at a quality of service similar to that of other push notification mechanisms.

Social-aware device-to-device communication: a contribution for edge and fog computing?

The exploitation of the opportunistic infrastructure via Device-to-Device (D2D) communication is a critical component towards the adoption of new paradigms such as edge and fog computing. While a lot of work has demonstrated the great potential of D2D communication, it is still unclear whether the benefits of the D2D approach can really be leveraged in practice. In this paper, we develop a software sensor, namely Detector, which senses the infrastructure in proximity of a mobile user. We analyze and evaluate D2D on the wild, i.e., not in simulations. We found that in a realistic environment, a mobile is always co-located in proximity to at least one other mobile device throughout the day. This suggests that a device can schedule tasks processing in coordination with other devices, potentially more powerful, instead of handling the processing of the tasks by itself.

Proximal and social-aware device-to-device communication via audio detection on cloud

Device-to-Device (D2D) communication is a potential strategy to release the mobile network from unnecessary data transfer, accelerate the responsiveness of end-to-end apps, and decentralize the provisioning of traditional services. D2D coordination is a critical challenge, which cannot be overcome without the explicit intervention of the user as D2D communication represents a threat for users privacy. However, social attributes can be leveraged to equip the devices with trusted mechanisms that can automate D2D communication. In this paper, we build and design a mobile cloud system that relies on audio data obtained from users environment to determine whether a set of devices are located in proximity. Audio analysis is performed on the cloud using classical machine learning principles, and the cloud instance (server) also informs the devices about the coordination plan to establish D2D communication. The framework is evaluated using a smartphone app for sharing files and the evaluation shows that the approach is feasible in practice.

Large-scale offloading in the Internet of Things

Large-scale deployments of IoT devices are subject to energy and performance issues. Fortunately, offloading is a promising technique to enhance those aspects. However, several problems still remain open regarding cloud deployment and provisioning. In this paper, we address the problem of provisioning offloading as a service in large-scale IoT deployments. We design and develop an AutoScaler, an essential component for our offloading architecture to handle offloading workload. In addition, we also develop an offloading simulator to generate dynamic offloading workload of multiple devices. With this toolkit, we study the effect of task acceleration in different cloud servers and analyze the capacity of several cloud servers to handle multiple concurrent requests. We conduct multiple experiments in a real testbed to evaluate the system and present our experiences and lessons learned. From the results, we find that the AutoScaler component introduces a very small overhead of ≈150 milliseconds in the total response time of a request, which is a fair price to pay to empower the offloading architectures with multi-tenancy ability and dynamic horizontal scaling for IoT scenarios.

TestAWARE: A Laboratory-Oriented Testing Tool for Mobile Context-Aware Applications

Although mobile context instrumentation frameworks have simplified the development of mobile context-aware applications, it remains challenging to test such applications. In this paper, we present TestAWARE that enables developers to systematically test context-aware applications in laboratory settings. To achieve this, TestAWARE is able to download, replay and emulate contextual data on either physical devices or emulators. To support both white -box and black-box testing, TestAWARE has been implemented as a novel structure with a mobile client and code library. In blackbox testing scenarios, developers can manage data replay through the mobile client, without writing testing scripts or modifying the source code of the targeted application. In white-box testing scenarios, developers can manage data replay and test functional/non-functional properties of the targeted application by writing testing scripts using the code library. We evaluated TestAWARE by quantifying its maximal data replay speed, and by conducting a user study with 13 developers. We show that TestAWARE can overcome data synchronisation challenges, and found that PC-based emulators can replay data significantly faster than physical smartphones and tablets. The user study highlights the usefulness of TestAWARE in the systematic testing of mobile context-aware applications in laboratory settings.