Automatic Software Tuning of Parallel Programs for Energy-Aware Executions

in Proc. of 13th Intl. Conf. on Parallel Processing and Applied Mathematics (PPAM 2019) (2019, December)

For large scale systems, such as data centers, energy efficiency has proven to be key for reducing capital, operational expenses and environmental impact. Power drainage of a system is closely related to ... [more ▼]

For large scale systems, such as data centers, energy efficiency has proven to be key for reducing capital, operational expenses and environmental impact. Power drainage of a system is closely related to the type and characteristics of workload that the device is running. For this reason, this paper presents an automatic software tuning method for parallel program generation able to adapt and exploit the hardware features available on a target computing system such as an HPC facility or a cloud system in a better way than traditional compiler infrastructures. We propose a search based approach combining both exact methods and approximated heuristics evolving programs in order to find optimized configurations relying on an ever-increasing number of tunable knobs i.e., code transformation and execution options (such as the num- ber of OpenMP threads and/or the CPU frequency settings). The main objective is to outperform the configurations generated by traditional compiling infrastructures for selected KPIs i.e., performance, energy and power usage (for both for the CPU and DRAM), as well as the runtime. First experimental results tied to the local optimization phase of the proposed framework are encouraging, demonstrating between 8% and 41% improvement for all considered metrics on a reference benchmark- ing application (i.e., Linpack). This brings novel perspectives for the global optimization step currently under investigation within the presented framework, with the ambition to pave the way toward automatic tuning of energy-aware applications beyond the performance of the current state-of-the-art compiler infrastructures. [less ▲]

A Multilayer Low-Altitude Airspace Model for UAV Traffic Management

in Samir Labib, Nader; Danoy, Grégoire; Musial, Jedrzej (Eds.) et al 9th ACM Symposium on Design and Analysis of Intelligent Vehicular Networks and Applications (DIVANet '19) (2019, November)

Over the recent years, Unmanned Aerial Vehicles' (UAVs) technology developed rapidly. In turn shedding light on a wide range of potential civil and commercial applications ranging from mapping and ... [more ▼]

Over the recent years, Unmanned Aerial Vehicles' (UAVs) technology developed rapidly. In turn shedding light on a wide range of potential civil and commercial applications ranging from mapping and surveillance, parcel delivery to more demanding ones that require UAVs to operate in heterogeneous swarms. However, with the great advantages UAVs bring, they are expected to soon dominate the shared, low-altitude airspace over populated cities, introducing multiple new research challenges in safely managing the unprecedented traffic demands. The main contribution of this work is addressing the complex problem of UAV traffic management at an abstract level by proposing a structure for the uncontrolled low-altitude airspace. The paper proposes a model of the airspace as a weighted multilayer network of nodes and airways and presents a set of experimental simulations of UAV traffic for the verification and validation of the model. Finally, the paper outlines our intended future work. [less ▲]

Crowdsensed Data Learning-Driven Prediction of Local Businesses Attractiveness in Smart Cities

in IEEE Symposium on Computers and Communications (ISCC), Barcelona, Spain, 2019 (2019, July)

Urban planning typically relies on experience-based solutions and traditional methodologies to face urbanization issues and investigate the complex dynamics of cities. Recently, novel data-driven ... [more ▼]

Urban planning typically relies on experience-based solutions and traditional methodologies to face urbanization issues and investigate the complex dynamics of cities. Recently, novel data-driven approaches in urban computing have emerged for researchers and companies. They aim to address historical urbanization issues by exploiting sensing data gathered by mobile devices under the so-called mobile crowdsensing (MCS) paradigm. This work shows how to exploit sensing data to improve traditionally experience-based approaches for urban decisions. In particular, we apply widely known Machine Learning (ML) techniques to achieve highly accurate results in predicting categories of local businesses (LBs) (e.g., bars, restaurants), and their attractiveness in terms of classes of temporal demands (e.g., nightlife, business hours). The performance evaluation is conducted in Luxembourg city and the city of Munich with publicly available crowdsensed datasets. The results highlight that our approach does not only achieve high accuracy, but it also unveils important hidden features of the interaction of citizens and LBs. [less ▲]

A GP Hyper-Heuristic Approach for Generating TSP Heuristics

in 33rd IEEE International Parallel & Distributed Processing Symposium (IPDPS 2019) (2019, May 20)

A Survey on Mobile Crowdsensing Systems: Challenges, Solutions, and Opportunities

in IEEE Communications Surveys and Tutorials (2019), 21(3, thirdquarter 2019), 2419-2465

Mobile crowdsensing (MCS) has gained significant attention in recent years and has become an appealing paradigm for urban sensing. For data collection, MCS systems rely on contribution from mobile devices ... [more ▼]

Mobile crowdsensing (MCS) has gained significant attention in recent years and has become an appealing paradigm for urban sensing. For data collection, MCS systems rely on contribution from mobile devices of a large number of participants or a crowd. Smartphones, tablets, and wearable devices are deployed widely and already equipped with a rich set of sensors, making them an excellent source of information. Mobility and intelligence of humans guarantee higher coverage and better context awareness if compared to traditional sensor networks. At the same time, individuals may be reluctant to share data for privacy concerns. For this reason, MCS frameworks are specifically designed to include incentive mechanisms and address privacy concerns. Despite the growing interest in the research community, MCS solutions need a deeper investigation and categorization on many aspects that span from sensing and communication to system management and data storage. In this paper, we take the research on MCS a step further by presenting a survey on existing works in the domain and propose a detailed taxonomy to shed light on the current landscape and classify applications, methodologies, and architectures. Our objective is not only to analyze and consolidate past research but also to outline potential future research directions and synergies with other research areas. [less ▲]

Transforming Collaboration Data into Network Layers for Enhanced Analytics

Scientific Conference (2019, February 25)

We consider the problem of automatically generating networks from data of collaborating researchers. The objective is to apply network analysis on the resulting network layers to reveal supplemental ... [more ▼]

We consider the problem of automatically generating networks from data of collaborating researchers. The objective is to apply network analysis on the resulting network layers to reveal supplemental patterns and insights of the research collaborations. In this paper, we describe our data-to-networks method, which automatically generates a set of logical network layers from the relational input data using a linkage threshold. We, then, use a series of network metrics to analyze the impact of the linkage threshold on the individual network layers. Moreover, results from the network analysis also provide beneficial information to improve the network visualization. We demonstrate the feasibility and impact of our approach using real-world collaboration data. We discuss how the produced network layers can reveal insights and patterns to direct the data analytics more intelligently. [less ▲]

Security, reliability and regulation compliance in Ultrascale Computing System

in Zomaya, A. Y.; Carretero, J.; Jeannot, E. (Eds.) Ultrascale Computing Systems (2019)

Ultrascale Computing Systems (UCSs) are envisioned as large-scale complex systems joining parallel and distributed computing systems that will be two to three orders of magnitude larger than today’s ... [more ▼]

Ultrascale Computing Systems (UCSs) are envisioned as large-scale complex systems joining parallel and distributed computing systems that will be two to three orders of magnitude larger than today’s systems (considering the number of Central Process Unit (CPU) cores). It is very challenging to find sustainable solutions for UCSs due to their scale and a wide range of possible applications and involved technologies. For example, we need to deal with heterogeneity and cross fertilization among HPC, large-scale distributed systems, and big data management. One of the challenges regarding sustainable UCSs is resilience. Another one, which attracted less interest in the literature but becomes more and more crucial with the expected convergence with the Cloud computing paradigm, is the notion of regulation in such system to assess the Quality of Service (QoS) and Service Level Agreement (SLA) proposed for the use of these platforms. This chapter covers both aspects through the reproduction of two articles: [1] and [2]. [less ▲]

Multi-Objective Scientific-Workflow Scheduling With Data Movement Awareness in Cloud.

in IEEE Access (2019), 7

Due to serving several purposes simultaneously, running scientific workflows on dynamic environments such as cloud computing, has become multi-objective scheduling. Among these purposes, Cost and Makespan ... [more ▼]

Due to serving several purposes simultaneously, running scientific workflows on dynamic environments such as cloud computing, has become multi-objective scheduling. Among these purposes, Cost and Makespan are probably the most two primitive objectives. Another critical factor in a large-scale scientific workflow is tremendous amount of data during execution. Therefore, this work also includes Data Movement as an additional objective as it has a major impact on network utilization and energy consumption in network equipment in cloud data center. In considering these three objectives, this work proposes a framework for scheduling solutions which combines a new nodes clustering technique in Directed Acyclic Graph (DAG) model known as Multilevel Dependent Node Clustering (MDNC) and the multiobjective optimization, Extreme Nondominated Sorting Genetic Algorithm-III (E-NSGA-III). E-NSGAIII is the recent extension of Nondominated Sorting Genetic Algorithm (NSGA-III). Five well-known scientific workflows, CyberShake, Epigenomics, LIGO, Montage, and SIPHT are selected as testbeds, while the commonly known Hypervolume is chosen as the performance metric. In this work, MDNC is also experimented with both NSGA-III. Comparison among three approaches, E-NAGA-III alone, E-NAGA-III with Peer-to-Peer clustering and E-NAGA-III with MDNC are carried out. The superiority of the proposed framework among them and its limitation are discussed. [less ▲]

Toward Real-world Vehicle Placement Optimization in Round-trip Carsharing

in Proceedings of the Genetic and Evolutionary Computation Conference (2019)

Profiling Performance of Application Partitioning for Wearable Devices in Mobile Cloud and Fog Computing

in IEEE Access (2019), 7

Wearable devices have become essential in our daily activities. Due to battery constrains the use of computing, communication, and storage resources is limited. Mobile Cloud Computing (MCC) and the ... [more ▼]

Wearable devices have become essential in our daily activities. Due to battery constrains the use of computing, communication, and storage resources is limited. Mobile Cloud Computing (MCC) and the recently emerged Fog Computing (FC) paradigms unleash unprecedented opportunities to augment capabilities of wearables devices. Partitioning mobile applications and offloading computationally heavy tasks for execution to the cloud or edge of the network is the key. Offloading prolongs lifetime of the batteries and allows wearable devices to gain access to the rich and powerful set of computing and storage resources of the cloud/edge. In this paper, we experimentally evaluate and discuss rationale of application partitioning for MCC and FC. To experiment, we develop an Android-based application and benchmark energy and execution time performance of multiple partitioning scenarios. The results unveil architectural trade-offs that exist between the paradigms and devise guidelines for proper power management of service-centric Internet of Things (IoT) applications. [less ▲]