Cost-performance Tradeoffs Research Articles

Selective laser melting of CP-Ti to overcome the low cost and high performance trade-off

In this study, commercially pure titanium (CP-Ti) parts were successfully fabricated by selective laser melting (SLM) using cost-effective hydride-dehydride (HDH) Ti powders for the first time modified by jet milling. Jet milling effectively improves the particle-shape sphericity, suppresses the impurity pick-up, and produces localized plastic deformation. The flowability of the jet-milled powders is tremendously improved to 29.7 s/50 g that satisfies the SLM processing well, while the oxygen content only increases by 0.02 wt.% (the raw oxygen level: 0.15 wt.%). The oxide layer in the powder surface is determined with the thickness of ∼8 nm and TiO being the predominant phase before and after jet milling. The SLM-made (SLMed) CP-Ti achieves dominant martensitic α’ phase with the fracture tensile strength up to 731.5 ± 5.7 MPa and elongation of 20.5 ± 1.1%, comparable with those using expensive atomized powders. Contrary to the conventional metallurgical mechanism for Ti which suffers the cost-performance dilemma, this work presents SLMed CP-Ti with excellent synergy of strength and ductility while using the cost-affordable HDH Ti powders.

Internet backbones in space

Several "NewSpace" companies have launched the first of thousands of planned satellites for providing global broadband Internet service. The resulting low-Earth-orbit (LEO) constellations will not only bridge the digital divide by providing service to remote areas, but they also promise much lower latency than terrestrial fiber for long-distance routes. We show that unlocking this potential is non-trivial: such constellations provide inherently variable connectivity, which today's Internet is ill-suited to accommodate. We therefore study cost-performance tradeoffs in the design space for Internet routing that incorporates satellite connectivity examining four solutions ranging from naively using BGP to an ideal, clean-slate design. We find that the optimal solution is provided by a path-aware networking architecture in which end-hosts obtain information and control over network paths. However, a pragmatic and more deployable approach inspired by the design of content distribution networks can also achieve stable and close-to-optimal performance.

Cross-Layer Co-Optimization of Network Design and Chiplet Placement in 2.5-D Systems

2.5-D integration technology is gaining attention and popularity in manycore computing system design. 2.5-D systems integrate homogeneous or heterogeneous chiplets in a flexible and cost-effective way. The design choices of 2.5-D systems impact overall system performance, manufacturing cost, and thermal feasibility. This article proposes a cross-layer co-optimization methodology for 2.5-D systems. We jointly optimize the network topology and chiplet placement across logical, physical, and circuit layers to improve system performance, reduce manufacturing cost, and lower operating temperature, while ensuring thermal safety and routability. We also propose a novel gas-station link, which enables pipelined interchiplet links in passive interposers. Our cross-layer methodology achieves better performance-cost tradeoffs of 2.5-D systems and yields better solutions in optimizing interchiplet network and 2.5-D system designs than prior methods. Compared to single-chip systems, 2.5-D systems designed using our new approach achieve 88% higher performance at the same manufacturing cost, or 29% lower cost with the same performance. Compared to the closest state-of-the-art, our new approach achieves 40%–68% (49% on average) iso-cost performance improvement and 30%–38% (32% on average) iso-performance cost reduction.

Motor/Generator Applications in Electrified Vehicle Chassis—A Survey

Electrifying the ground vehicles is a promising approach to alleviating the energy and climate issues. The adoption of an electric vehicle (EV) and a hybrid EV (HEV) has introduced significant impacts on various aspects of vehicle design, especially on the chassis systems. When vehicles tend to be lighter, smarter, and autonomous, more electroactuators with a wide-bandwidth control and an energy-regeneration capability are deployed instead of traditional hydraulic ones to improve the vehicle dynamic and energy performance. In this article, an overview of the chassis electrification technologies regarding different configurations of electric/hybrid powertrains, active suspension, and steering systems, along with a comprehensive review of key research fields on the design and control of the discussed systems, is presented. For related technologies, the comparative discussions and design considerations serve as a reference for practical research and development. Although the electric solutions in different subsystems show great potential to further benefit the essential chassis functions, a fully electrified chassis still has a long way to go considering the technical maturity and the cost–performance tradeoff. This article also highlights the technical gaps and supplies future directions for the chassis electrification.

Performance-Cost Tradeoff of Using Mobile Roadside Units for V2X Communication

Roadside unit (RSU) is a communication device for vehicular networks that provides vehicle-to-infrastructure (V2I) connectivity to nearby vehicles. It allows broadcasting of traffic and safety information, Internet connectivity, shared storage, advertisement, as well as supporting and enhancing vehicle-to-vehicle (V2V) communication for future autonomous vehicles. However, high cost of its deployment and management has so far prevented RSUs from being used widely in practice despite its utility and importance. In this paper, we investigate the performance-cost tradeoff and viability of using buses as mobile RSUs (mRSUs). Although there have been some prior works that suggest the use of cars, public transportation, and controlled vehicles as mRSUs to address the cost problem of static RSUs (sRSUs), their assumptions were mostly rather idealistic. We aim to provide a more realistic view of the problem. Through real-world measurements, experiments, analysis, and simulation, we show how mRSUs can replace sRSUs while maintaining the same level of throughput, contact time, and inter-contact time. Our results provide a basis for judging whether it is beneficial to complement sRSUs with mRSUs depending on the deployment environment and cost-performance tradeoff.

Optimal Cloud Resource Allocation With Cost Performance Tradeoff Based on Internet of Things

Internet of Things (IoT) still faces many challenges, one of which is how to efficiently and effectively allocate the cloud resources in order to obtain the desired quality of service (QoS) for the IoT users and communication domains. To face the above challenges, this paper first identifies that there is an issue of cost performance tradeoff, which is stemmed from various limited resources in the IoT systems and the competing services requirements. We second propose a nonlinear optimization model that attacks the resource allocation problem. Within the constraints of resources and service demands, this model seeks to maximize the suggested IoT cost performance ratio. We then find that the problem can be treated as a quasiconcave maximization problem. We are able to reduce the original problem to a pseudoconcave maximization problem and thereby identify a local solution, which is proved to be exactly the desired global one. We therefore propose a feasible direction method and design its corresponding algorithm to yield the desired solution. Finally, we provide a numerical example to demonstrate the theoretical findings of resource allocation and the proposed algorithm for the cloud computing based on IoT systems. The mathematical results and method proposed in this paper can act as designing guidelines for resource allocation and managements, computing scheduling, and networking protocol designing in the cloud infrastructure of IoT.

Analysis-Synthesis Learning With Shared Features: Algorithms for Histology Image Classification.

The diversity of tissue structure in histopathological images makes feature extraction for classification a challenging task. Dictionary learning within a sparse representation-based classification (SRC) framework has been shown to be successful for feature discovery. However, there exist stiff practical challenges: 1) computational complexity of SRC can be onerous in the decision stage since it involves solving a sparsity constrained optimization problem and often over a large number of image patches; and 2) images from distinct classes continue to share several geometric features. We propose a novel analysis-synthesis model learning with shared features algorithm (ALSF) for classifying such images more effectively. In the ALSF, a joint analysis and synthesis learning model is introduced to learn the classifier and the feature extractor at the same time. Unlike SRC, no explicit optimization is needed in the inference phase leading to much reduced computation. Crucially, we introduce the learning of a low-rank shared dictionary and a shared analysis operator, which more accurately represents both similarities and differences in histopathological images from distinct classes. We also develop an extension of ALSF with a sparsity constraint, whose presence or absence facilitates a cost-performance tradeoff. The ALSF is evaluated on three challenging and well-known datasets: 1) spleen tissue images; 2) brain tumor images; and 3) breast cancer tissue dataset, provided by different organizations. Experimental results demonstrate both complexity and performance benefits of the ALSF over state-of-the-art alternatives. Modeling shared features with appropriate quantitative constraints lead to significantly improved classification in histopathology.

Cost-Efficient Tasks and Data Co-Scheduling with AffordHadoop

With today's massive jobs spanning thousands of tasks each, cost-optimality has become more important than ever. Modern distributed data processing paradigms can be significantly more sensitive to cost than makespan, especially for long jobs deployed in commercial clouds. This paper posits that minimized dollar costs can not be achieved unless data and tasks are scheduled simultaneously. In this paper, we introduce the problem of cost-efficient co-scheduling for highly data-intensive jobs in cloud, such as MapReduce. We show that while the problem is polynomial in some cases, its general problem is NP-Hard. We propose to tackle the problem by using integer programming techniques coupled with heuristic reduction and optimization to enable a near-realtime solution. AffordHadoop, a pluggable co-scheduler for Hadoop, is implemented as an example of such a co-scheduler. AffordHadoop can save up to 48 percent of the overall dollar costs when compared to existing schedulers and provides significant flexibility in fine-tuning the cost-performance tradeoff.

Adaptive beamforming and user association in heterogeneous cloud radio access networks: A mobility-aware performance-cost trade-off

Heterogeneous Cloud Radio Access Network (H-CRAN) is a promising network architecture for the future 5G mobile communication system to address the increasing demand for mobile data traffic. In this work, we consider the design of efficient joint beamforming and user clustering (user-to-Remote Radio Head (RRH) association) in the downlink of a H-CRAN where users have different mobility profiles. Given the rapidly time-varying nature of such wireless environment, it becomes very challenging to enable optimized beamforming and user clustering without incurring large Channel State Information (CSI) and signaling overheads. The main objective of this work is to investigate and evaluate the trade-off between system throughput and the incurred costs in terms of complexity and signaling overhead, including the impact of different CSI feedback strategies given different user mobility profiles. We propose the Adaptive Beamforming and User Clustering (ABUC) algorithm which adapts its feedback parameters, namely the period of dynamic user clustering and the type of CSI feedback, in function of user mobility. Furthermore, we design a reinforcement-learning framework which enables the proposed ABUC algorithm to optimize its scheduling parameters on-the-fly, given each user mobility profile. Based on computer simulations, an analysis of the effect of mobility on system performance metrics is presented and conclusions are drawn regarding the algorithm’s adequate parameter tuning for different mobility scenarios.11This is an extended version of the conference paper presented at IEEE PIMRC 2017 [1].

A Roadmap to Low-Cost Hydrogen with Hydroxide Exchange Membrane Electrolyzers.

Hydrogen is an ideal alternative energy carrier to generate power for all of society's energy demands including grid, industrial, and transportation sectors. Among the hydrogen production methods, water electrolysis is a promising method because of its zero greenhouse gas emission and its compatibility with all types of electricity sources. Alkaline electrolyzers (AELs) and proton exchange membrane electrolyzers (PEMELs) are currently used to produce hydrogen. AELs are commercially mature and are used in a variety of industrial applications, while PEMELs are still being developed and find limited application. In comparison with AELs, PEMELs have more compact structure and can achieve higher current densities. Recently, however, an alternative technology to PEMELs, hydroxide exchange membrane electrolyzers (HEMELs), has gained considerable attention due to the possibility to use platinum group metal (PGM)-free electrocatalysts and cheaper membranes, ionomers, and construction materials and its potential to achieve performance parity with PEMELs. Here, the state-of-the-art AELs and PEMELs along with the current status of HEMELs are discussed in terms of their positive and negative aspects. Additionally discussed are electrocatalyst, membrane, and ionomer development needs for HEMELs and benchmark electrocatalysts in terms of the cost-performance tradeoff.

HMMN: a cost-effective derivative of midimew-connected mesh network

A horizontal midimew-connected mesh network (HMMN) is constructed using multiple mesh-connected basic modules, whereby the basic modules are hierarchically interconnected by a horizontal midimew network for a higher level of hierarchy. In the HMMN, the vertical direction is toroidal connected and the horizontal direction is midimew links connected. This is why it is called HMMN. The architecture of the HMMN, the static network performance, and the dynamic communication performance were evaluated in our previous studies and were found to be good. However, the cost-effectiveness and cost-performance trade-off analysis of the HMMN have not been evaluated yet. Therefore, in this paper, we have evaluated packing density, message traffic density, cost-effective factor, time cost-effective factor, and static cost-performance trade-off factor of the proposed HMMN and compared these with those of mesh, torus, TESH, and MMN networks. Accordingly, the HMMN yields better results than the mesh, TESH, and MMN networks with respect to the above parameters. Even in regard to the extra cost of wrap-around links in all end-to-end nodes, the HMMN is more cost-effective than the torus network. However, this additional path between the end-to-end node using wrap-around links results in a low diameter, which in turn yields a slightly better density parameter and static cost-performance trade of factor for the torus network than for the HMMN.

Autoscaling tiered cloud storage in Anna

In this paper, we describe how we extended a distributed key-value store called Anna into an autoscaling, multi-tier service for the cloud. In its extended form, Anna is designed to overcome the narrow cost-performance limitations typical of current cloud storage systems. We describe three key aspects of Anna's new design: multi-master selective replication of hot keys, a vertical tiering of storage layers with different cost-performance tradeoffs, and horizontal elasticity of each tier to add and remove nodes in response to load dynamics. Anna's policy engine uses these mechanisms to balance service-level objectives around cost, latency and fault tolerance. Experimental results explore the behavior of Anna's mechanisms and policy, exhibiting orders of magnitude efficiency improvements over both commodity cloud KVS services and research systems.

Surface-Property Recognition With Force Sensors for Stable Walking of Humanoid Robot

In this paper, we propose a surface-identification system for stable humanoid-robot walking on various types of surfaces using force sensors mounted under the robot feet. For experimental identification analysis of the surface condition, we measured the sensor-output data using five different types of test surfaces. To achieve fast dynamic recognition capability of changing surface conditions, we applied an overlapped sliding-window method for the incoming sensor-data stream to generate dynamically four distinguishable well-known features from the raw sensor data. The multi-class k-nearest-neighbor (MC-kNN) classifier rather than a binary classifier is used for online classification of the measured robot-walking pattern and classification-accuracy evaluation. Further, we combine the four studied feature descriptors into a fused multi-feature descriptor rather than invoking each feature descriptor independently, increasing the classification performance. Our analysis results verify that 90.4% maximum overall accuracy with 91.49% average precision can be achieved, demonstrating the realization of a better cost-performance trade-off than in other previous research works. The obtained results are useful for balancing the robot body through optimized controlling of the robot motors according to the recognized different surfaces during robot motion.

Parametric Critical Path Analysis for Event Networks With Minimal and Maximal Time Lags

High-end manufacturing systems are cyber-physical systems, where productivity depends on the close cooperation of mechanical (physical) and scheduling (cyber) aspects. Mechanical and control constraints impose minimal and maximal time differences between events in the product flow. Sequence-dependent constraints are used by a scheduler to optimize system productivity while satisfying operational requirements. The numerous constraints in a schedule are typically related to a relatively small set of parameters, such as speeds, lengths, or settling times. We contribute a parametric critical path algorithm that identifies bottlenecks in terms of the feasible parameter combinations. This algorithm allows analysis of schedules to identify bottlenecks in terms of the underlying cause of constraints. We also contribute a way to find Pareto-optimal cost-performance tradeoffs and their associated parameter combinations. These results are used to quantify the impact of relaxing constraints that hinder system productivity.

Follow Me at the Edge: Mobility-Aware Dynamic Service Placement for Mobile Edge Computing

Mobile edge computing is a new computing paradigm, which pushes cloud computing capabilities away from the centralized cloud to the network edge. However, with the sinking of computing capabilities, the new challenge incurred by user mobility arises: since end-users typically move erratically, the services should be dynamically migrated among multiple edges to maintain the service performance, i.e., user-perceived latency. Tackling this problem is non-trivial since frequent service migration would greatly increase the operational cost. To address this challenge in terms of the performance-cost trade-off, in this paper we study the mobile edge service performance optimization problem under long-term cost budget constraint. To address user mobility which is typically unpredictable, we apply Lyapunov optimization to decompose the long-term optimization problem into a series of real-time optimization problems which do not require a priori knowledge such as user mobility. As the decomposed problem is NP-hard, we first design an approximation algorithm based on Markov approximation to seek a near-optimal solution. To make our solution scalable and amenable to future 5G application scenario with large-scale user devices, we further propose a distributed approximation scheme with greatly reduced time complexity, based on the technique of best response update. Rigorous theoretical analysis and extensive evaluations demonstrate the efficacy of the proposed centralized and distributed schemes.

Package Design Optimization for Intel SoC Xeon-D

Xeon-D brings the high performance of Xeon processors into a dense, low-power system-on-chip. This paper addresses the importance of cost–performance tradeoff analysis for the Xeon-D package. It describes how to determine the low-cost package factors (size, footprint, pin map, and layer count) without compromising the performance of the system. The 10-GbE signal integrity design and high-speed differential cost-performance optimization are discussed. Low-power architecture and package power delivery features including fully integrated voltage regulator are presented in this paper as well.

Performance evaluation of a permanent magnet-assisted synchronous reluctance machine with hybrid magnets of ferrite magnets and rare-earth PMs

In this paper, a novel permanent magnet-assisted synchronous reluctance machine (PMASynRM) with rare-earth PMs and ferrite magnets is proposed. The performance of PMASynRM is discussed with respected to the different magnet ratio of rare-earth PMs and ferrite magnets. Some characteristics including the flux density, output torque, cogging torque, output power, power factor, torque ripple, loss, efficiency, and demagnetization are calculated by 2-D finite element analysis (FEA). The analysis results show that the excellent performance can be obtained by using hybrid magnet of rare-earth PMs and ferrite magnets with the suitable magnet ratio, and provide some desirable cost-performance trade-off.

Optimizing the Cost-Performance Tradeoff for Coflows Across Geo-Distributed Datacenters

For data analytics jobs running across geographically distributed datacenters, coflows have to go through the inter-datacenter network over relatively low bandwidth and high cost links. In this case, optimizing cost-performance tradeoffs for such coflows becomes crucial. Ideally, decreasing the coflow completion time (CCT) can significantly improve the network performance, meanwhile, reducing the transmission cost introduced by these coflows is another fundamental goal for datacenter operators. Unfortunately, minimizing both the CCT and the transmission cost are conflicting objectives that cannot be achieved concurrently. Prior methods have significant limitations when exploring such tradeoffs, because they either merely decrease the average CCT or reduce the transmission cost independently. In this paper, we focus on a cost-performance tradeoff problem for coflows running across the inter-datacenter network. Specifically, we formulate an optimization problem, so as to minimize a combination of both the average CCT and the average transmission cost. This problem is inherently hard to solve due to the unknown information of future coflows. Therefore, we present Lever , an online coflow-aware optimization framework, to balance these two conflicting objectives. Without any prior knowledge of future coflows, Lever has been proved to have a non-trivial competitive ratio in solving this cost-performance tradeoff problem. Results from large-scale simulations demonstrate that Lever can significantly reduce the average transmission cost, and at the same time, speed up the completion of these coflows, compared with the state-of-the-art solutions.

Distributing the computation in combinatorial optimization experiments over the cloud

Combinatorial optimization is an area of great importance since many of the real-world problems have discrete parameters which are part of the objective function to be optimized. Development of combinatorial optimization algorithms is guided by the empirical study of the candidate ideas and their performance over a wide range of settings or scenarios to infer general conclusions. Number of scenarios can be overwhelming, especially when modeling uncertainty in some of the problem’s parameters. Since the process is also iterative and many ideas and hypotheses may be tested, execution time of each experiment has an important role in the efficiency and successfulness. Structure of such experiments allows for significant execution time improvement by distributing the computation. We focus on the cloud computing as a cost- efficient solution in these circumstances. In this paper we present a system for validating and comparing stochastic combinatorial optimization algorithms. The system also deals with selection of the optimal settings for computational nodes and number of nodes in terms of performance-cost tradeoff. We present applications of the system on a new class of project scheduling problem. We show that we can optimize the selection over cloud service providers as one of the settings and, according to the model, it resulted in a substantial cost-savings while meeting the deadline.

MLscale: A machine learning based application-agnostic autoscaler

Autoscaling is the practice of automatically adding or removing resources for an application deployment to meet performance targets in response to changing workload conditions. However, existing autoscaling approaches typically require expert application and system knowledge to reduce resource costs and performance target violations, thus limiting their applicability. We present MLscale, an application-agnostic, machine learning based autoscaler that is composed of: (i) a neural network based online (black-box) performance modeler, and (ii) a regression based metrics predictor to estimate post-scaling application and system metrics. Implementation results for diverse applications across several traces highlight MLscale's application-agnostic behavior and show that MLscale (i) reduces resource costs by about 41%, on average, compared to the optimal static policy, (ii) is within 14%, on average, of the cost of the optimal dynamic policy, and (iii) provides similar cost-performance tradeoffs, without requiring any tuning, when compared to carefully tuned threshold-based policies.