Multi-resource scheduling for FPGA systems

In modern cloud data centers, reconfigurable devices (FPGAs) are used as an alternative to Graphics Processing Units to accelerate data-intensive computations (e.g., machine learning, image and signal processing). Currently, FPGAs are configured to execute fixed workloads, repeatedly over long periods of time. This conflicts with the needs, proper to cloud computing, to flexibly allocate different workloads and to offer the use of physical devices to multiple users. This raises the need for novel, efficient FPGA scheduling algorithms that can decide execution orders close to the optimum in a short time. In this context, we propose a novel scheduling heuristic where groups of tasks that execute together are interposed by hardware reconfigurations. Our contribution is based on gathering tasks around a high-latency task that hides the latency of tasks, within the same group, that run in parallel and have shorter latencies. We evaluated our solution on a benchmark of 37500 random workloads, synthesized from realistic designs (i.e., topology, resource occupancy). For this testbench, on average, our heuristic produces optimum makespan solutions in 47.4% of the cases. It produces acceptable solutions for moderately constrained systems (i.e., the deadline falls within 10% of the optimum makespan) in 90.1% of the cases.     

Mining fleet management systems: a review of models and algorithms

The objective of this paper is to review and document the mine fleet management systems’ models and algorithms. The purpose is to understand the algorithms behind the fleet management systems and the proposed academic solutions in this area to identify any gaps in the current literature and to open up opportunities to establish research questions that need to be addressed in an integrated simulation and optimisation operational planning research framework. In this paper, we review industrial fleet management systems and the main academic algorithms behind such systems. The fleet management systems are divided into three subsequently related problems to review: shortest path, production optimisation and real-time dispatching. Finally, the limitations of current algorithms for fleet management systems are documented in terms of mining practice feasibility and optimality of the solution on large-scale problems. The results of this literature review enable us to evaluate the logical links between major components of an integrated simulation and optimisation operational planning framework with current theory of fleet management systems.     

Test sample allocation method for testability verification test

An extended failure mode effect and criticality analysis (FMECA)-based sample allocation method for testability verification is presented in this study to deal with the poor representativeness of test sample sets and the randomness of the testability evaluation results caused by unreasonable selection of failure samples. First, the fault propagation intensity is introduced as part of the extended information of FMECA, and the sample allocation impact factors of component units and failure modes are determined under this framework. Then, the failure mode similarity and impact factor support are defined, and the game decision method for weighing the relationship between similarity and support is proposed to obtain the weight of failure mode impact factor. Finally, a two-step allocation framework of test samples is formulated to realize the sample allocation of component units and failure modes. This method is applied to the testability verification test of a launch control system. Results show that this method can obtain more representative test samples compared with the traditional sample allocation method while effectively reducing randomness of single testability evaluation result.     

Rate-distortion-complexity optimization for x265

Rate-distortion optimization (RDO) is conventionally based on the analysis of rate-distortion (R-D) curve to minimize the coding distortion under the coding bits constraint. However, it is necessary to consider the computational complexity in the RDO process. In this paper, we obtain the Confidence LEvel - Computational complexity (CLEC) curves which indicate the characteristics of coding tree units (CTUs). Based on the CLEC curves, a rate-distortion-complexity optimization (RDCO) algorithm is proposed to optimize R-D under given computational complexity and achieve the optimal coding performance for x265. Experimental results demonstrate that the proposed algorithm can achieve a wide range of encoding speed under a given quantization parameter (QP) whereas the original x265 can only achieve a few fixed encoding speeds, and the proposed algorithm can reduce the BD-rate and increase the BD-PSNR by 6.59% and 0.13 dB on average under the same requirements of encoding speeds as the original x265.     

A knowledge-based archive multi-objective simulated annealing algorithm to optimize series–parallel system with choice of redundancy strategies

Redundancy allocation problem (RAP) is one of the best-developed problems in reliability engineering studies. This problem follows to optimize the reliability of a system containing s sub-systems under different constraints, including cost, weight, and volume restrictions using redundant components for each sub-system. Various solving methodologies have been used to optimize this problem, including exact, heuristic, and meta-heuristic algorithms. In this paper, an efficient multi-objective meta-heuristic algorithm based on simulated annealing (SA) is developed to solve multi-objective RAP (MORAP). This algorithm is knowledge-based archive multi-objective simulated annealing (KBAMOSA). KBAMOSA applies a memory matrix to reinforce the neighborhood structure to achieve better quality solutions. The results analysis and comparisons demonstrate the performance of the proposed algorithm for solving MORAP.     

Simulation optimization approach for hybrid flow shop scheduling problem in semiconductor back-end manufacturing

This study presents a simulation optimization approach for a hybrid flow shop scheduling problem in a real-world semiconductor back-end assembly facility. The complexity of the problem is determined based on demand and supply characteristics. Demand varies with orders characterized by different quantities, product types, and release times. Supply varies with the number of flexible manufacturing routes but is constrained in a multi-line/multi-stage production system that contains certain types and numbers of identical and unrelated parallel machines. An order is typically split into separate jobs for parallel processing and subsequently merged for completion to reduce flow time. Split jobs that apply the same qualified machine type per order are compiled for quality and traceability. The objective is to achieve the feasible minimal flow time by determining the optimal assignment of the production line and machine type at each stage for each order. A simulation optimization approach is adopted due to the complex and stochastic nature of the problem. The approach includes a simulation model for performance evaluation, an optimization strategy with application of a genetic algorithm, and an acceleration technique via an optimal computing budget allocation. Furthermore, scenario analyses of the different levels of demand, product mix, and lot sizing are performed to reveal the advantage of simulation. This study demonstrates the value of the simulation optimization approach for practical applications and provides directions for future research on the stochastic hybrid flow shop scheduling problem.     

A resilient buffer allocation scheme in active queue management: a stochastic cooperative game theoretic approach

During the last decade, a plentiful number of active queue management schemes have been proposed, but their main objectives are simply allocating the buffer resource to all flows evenly, or protecting responsive flows from being degraded by unresponsive flows. However, the sending rates of the responsive flows can be determined diversely, and not all unresponsive flows have aggressively high sending rates. Furthermore, it is rational to reserve a portion of the buffer resource for certain privileged traffic. Grounded by these evidences, in this paper, we present a resilient active queue management algorithm, named Prior‐Core‐based Buffer Allocation considering diverse congestion control algorithms, fair‐unresponsive flows, and some privileged traffic. Our approach is based on stochastic cooperative game theory, where the payoffs yielded by cooperation are described by random variables, and the core is defined only over the distribution of these random payoffs; the core in this situation is called the prior‐core. As a result, it is shown that our buffer allocation, yielded by the prior‐core, achieves completely fair allocation for those flows whose requirement does not exceed the fair‐share regardless of the responsiveness, whereas aggressive flows are restricted according to availability of the buffer; all these are verified through ns‐2 simulation experiments. Copyright © 2014 John Wiley & Sons, Ltd.     

社区居家养老服务设施的规划配置研究

社区和居家养老是我国当前和未来的主要养老模式，然而，与机构养老设施相比较而言，我国社区居家养老服务设施的发展明显滞后，具体表现为设施的建成基础薄弱、空间资源的配置要素模糊、规范和管控手段缺位等。在概述现状与问题的基础上，系统地探讨了我国社区层面养老服务资源的配置主体、配置对象和配置思路。针对养老服务需求和供给主体日益多元化的挑战，提出结合社区实际情况灵活配置服务模块的建议，采用总量控制和复合设置的规划手段落实配建，从而实现服务设施综合效益最大化的建设目标。     

Throughput-based fair bandwidth allocation in OBS networks

Fair bandwidth allocation (FBA) has been studied in optical burst switching (OBS) networks, with the main idea being to map the max-min fairness in traditional IP networks to the fair-loss probability in OBS networks. This approach has proven to be fair in terms of the bandwidth allocation for differential connections, but the use of the ErlangB formula to calculate the theoretical loss probability has made this approach applicable only to Poisson flows. Furthermore, it is necessary to have a reasonable fairness measure to evaluate FBA models. This article proposes an approach involving throughput-based-FBA, called TFBA, and recommends a new fairness measure that is based on the ratio of the actual throughput to the allocated bandwidth. An analytical model for the performance of the output link with TFBA is also proposed.     

A methodology to benchmark flexible payload architectures in a megaconstellation use case

This paper proposes a methodology to benchmark satellite payload architectures and find the optimal trade‐offs between high flexibility and low complexity. High flexibility would enable the satellite to adapt to various distributions of user terminals on the ground and fulfill the data rate demand of these users. Besides, low complexity is required to keep satellite networks competitive in the context of emerging 5G networks. To estimate the flexibility of a payload, an indicator to characterize the non‐uniformity of user distributions is proposed. Each benchmarked payload may be characterized by a graph relating the throughput to this parameter further denoted . The payload provides the same throughput trends for different scenarios of user distributions with the same parameter. As a consequence, the average capacity of the system may be estimated by (a) calculating the probability distribution of over the orbit and (b) integrating the throughput based on this payload response. It thus results in a straightforward way for benchmarking payloads directly on an estimation of the averaged capacity, accounting for the user distribution over the earth. A simulation platform has been developed to characterize the payload throughput including the implementation of a resource allocation algorithm that accounts for constraints of various payloads. Using this definition and the developed tool, we benchmark a bent‐pipe architecture, a beam hopping architecture and a hybrid beam‐steering architecture for a LEO megaconstellation use case. The methodology showcases the interest for investigating different payload architectures depending on realistic traffic scenario analysis.