A multi-dimensional job scheduling

With the advent of new computing technologies, such as cloud computing and contemporary parallel processing systems, the building blocks of computing systems have become multi-dimensional. Traditional scheduling systems based on a single-resource optimization, like processors, fail to provide near optimal solutions. The efficient use of new computing systems depends on the efficient use of several resource dimensions. Thus, the scheduling systems have to fully use all resources. In this paper, we address the problem of multi-resource scheduling via multi-capacity bin-packing. We propose the application of multi-capacity-aware resource scheduling at host selection layer and queuing mechanism layer of a scheduling system. The experimental results demonstrate performance improvements of scheduling in terms of waittime and slowdown metrics.     

Optimal assignment of dynamic priorities in queuing systems with two types of customers

We consider queuing systems with two types of customers. For such systems, we develop numerical procedures for computation of optimal dynamic priorities in the case of multiplicative priority functions. The optimality criterion is based on the total queue length for customers of both types. Our technique allows one to take into account waiting-time bounds. The optimization problem is formulated in the language of linear-fractional programming. To illustrate our technique, we present some numerical results at the end of the paper.     

改进粒子群优化的关联交叉口协调控制

提出了基于改进粒子群优化的关联交叉口协调控制方法。建立了关于排队长度的交通流模型和协调控制目标函数,利用改进粒子群算法对各交叉口绿信比和考虑双向绿波的相位差进行求解,实现了关联交叉口的最优控制。以实际采集的几个关联交叉口的交通数据仿真表明,相比单向绿波控制和感应控制,所提方法可有效减少延误和平均排队长度。     

闭排队网络基于并行仿真的灵敏度估计和优化算法

基于Markov性能势理论，对一类闭排队网络的灵敏度估计和优化，建立了一种行之有效的并行仿真算法。采用公共随机数，使所有的处理器使用相同的样本轨道，以减少各个处理器之间的通讯时间。在一台SPMD并行计算机上的仿真实例表明，该并行仿真算法对于闭排队网络的优化能显著地提高运算速度。     

Optimization of queuing system via stochastic control

The problem of access and service rate control in queuing systems as a general optimization problem for controlled Markov process with finite state space is considered. By using the dynamic programming approach we obtain the explicit form of the optimal control in the case of minimizing cost given as a mixture of an average queue length, number of lost jobs, and service resources. The problem is considered on a finite time interval in the case of nonstationary input flow. In this case we suggest the general procedure of the numerical solution which can be applied to a problems with constraints.     

Joint location and dispatching decisions for Emergency Medical Services

The main purpose of Emergency Medical Service systems is to save lives by providing quick response to emergencies. The performance of these systems is affected by the location of the ambulances and their allocation to the customers. Previous literature has suggested that simultaneously making location and dispatching decisions could potentially improve some performance measures, such as response times. We developed a mathematical formulation that combines an integer programming model representing location and dispatching decisions, with a hypercube model representing the queuing elements and congestion phenomena. Dispatching decisions are modeled as a fixed priority list for each customer. Due to the model’s complexity, we developed an optimization framework based on Genetic Algorithms. Our results show that minimization of response time and maximization of coverage can be achieved by the commonly used closest dispatching rule. In addition, solutions with minimum response time also yield good values of expected coverage. The optimization framework was able to consistently obtain the best solutions (compared to enumeration procedures), making it suitable to attempt the optimization of alternative optimization criteria. We illustrate the potential benefit of the joint approach by using a fairness performance indicator. We conclude that the joint approach can give insights of the implicit trade-offs between several conflicting optimization criteria.     

多服务员时两种等待队列性能的比较

利用排队论的相关知识,对计算机系统中常见的多服务员情况下的排队现象进行分析,通过理论推导、并用实际数据表明了在多服务员模式下,单一共享的排队等待队列的性能要优于多个独自的排队等待队列。     

Probability distribution functions for servicing two types of requests

A method for calculating the parameters of queuing systems with heterogeneous requests was obtained in our paper [1]. In [2], the accuracy of this method for a network of queuing systems was studied. This method is applicable for the case of the estimation of queuing systems with respect to an average value (averaged over the parameters of the distribution averages). This paper develops the method for calculating the parameters of queuing systems by using the integral characteristics (distribution functions of random quantities). This approach allows one to use more accurate methods for calculating the characteristics of queuing systems with different types of requests. In addition, this method provides more detailed specification of the characteristics of a queuing service system, in particular, the examination of the characteristics of output streams.     

Performance analysis and optimization for chunked network coding based wireless cooperative downloading systems

Dense network coding (NC) is widely used in wireless cooperative downloading systems. Wireless devices have limited computing resources. Researchers have recently found that dense NC is not suitable because of its high coding complexity, and it is necessary to use chunked NC in wireless environments. However, chunked NC can cause more communications, and the amount of communications is affected by the chunk size. Therefore, setting a suitable chunk size to improve the overall perfor-mance of chunked NC is a prerequisite for applying it in wireless cooperative downloading systems. Most of the existing studies on chunked NC focus on centralized wireless broadcasting systems, which are different from wireless cooperative downloading systems with distributed features. Accordingly, we study the performance of chunked NC based wireless cooperative downloading systems. First, an analysis model is established using a Markov process taking the distributed features into consideration, and then the block collection completion time of encoded blocks for cooperative downloading is optimized based on the analysis model. Furthermore, queuing theory is used to model the decoding process of the chunked NC. Combining queuing theory with the analysis model, the decoding completion time for cooperative downloading is optimized, and the optimal chunk size is derived. Numerical simulation shows that the block collection completion time and the decode completion time can be largely reduced after optimization.     

一种改进的M/D/1模型及其应用的研究

M/D/1模型(M表示到达时间间隔为指数分布,D表示服务时间间隔为定长分布,1表示单服务台)是一种常见的排队模型,在生活的很多领域都会遇到。但是实际生活中的排队过程常常不是平稳过程,顾客到达率往往跟时间有显著相关的关系。针对这一个问题,以上海世博会的排队系统为例,将全天排队过程划分为若干时间区间,将每个时间区间内的排队过程视为平稳过程。并对其全天到达率做线性假设,给出排队系统各参数指标的递推公式。进而对全天排队情况进行预测,预测结果表明该方法行之有效。     

基于消息队列的LightGBM超参数优化

为了提高LightGBM超参数优化效率,同时得到全局最优模型,提出了以消息队列方式并行优化LightGBM超参数方法。根据超参数的预选范围,将每一组超参数发送到队列中,各节点从队列获取到消息后以并行方式进行模型训练并验证准确率,最后选出准确率最高的模型计算待预测的数据集。实验结果表明,与传统的网格搜索、贝叶斯法、随机搜索方法以及消息队列串行优化相比,消息队列并行优化超参数方法时间最短,AUC值最大。     

A hybrid approach to minimize state space explosion problem for the solution of two stage tandem queues

Two stage open queuing networks are used for modeling the subsystem-behaviour in computers and communication networks, mass storage devices, memory servers, and queuing analysis of wireless mobile cellular networks. The queuing analysis of wireless systems is essential in order to quantify the impact of different factors on quality of service (QoS); performance measures so that wireless protocols can be designed and/or tuned in an optimal manner. In that sense two stage open queuing systems are particularly important to model handoff phenomena, especially for the integration of two different systems such as cellular and wireless local area networks (WLANs). Analytical solutions for two-dimensional Markov processes suffer from the state space explosion problem. The numerical difficulties caused by large state spaces, make it difficult to handle multiple servers at the second stage of a tandem queuing system together with server failures and repairs. This study presents a new approach to analytical modeling of open networks offering improvements in alleviating this problem. The proposed solution is a hybrid version, which combines well known spectral expansion, and hierarchical Markov reward rate approaches. Using this approach, two-stage open networks with multiple servers, break-downs, and repairs at the second stage and feedback can be modeled as three-dimensional Markov processes and solved for performability measures. Comparative results show that the new algorithm used for solution, provides a high degree of accuracy, and it is computationally more efficient than the existing approaches. The proposed model is capable of solving other three-dimensional Markov processes.     

A Queuing Model for Evaluating the Transfer Latency of Peer-to-Peer Systems

This paper presents a queuing model to evaluate the latency associated with file transfers or replications in peer-to-peer (P2P) computer systems. The main contribution of this paper is a modeling framework for the peers that accounts for the file size distribution, the search time, load distribution at peers, and number of concurrent downloads allowed by a peer. We propose a queuing model that models the nodes or peers in such systems as M/G/1/K processor sharing queues. The model is extended to account for peers which alternate between online and offline states. The proposed queuing model for the peers is combined with a single class open queuing network for the routers interconnecting the peers to obtain the overall file transfer latency. We also show that in scenarios with multipart downloads from different peers, a rate proportional allocation strategy minimizes the download times.     

Investigation of the M2/G2/1/∞,N queue with restricted admission of priority customers and its application to HSDPA mobile systems

This paper investigates a queuing system for QoS optimization of multimedia traffic consisting of aggregated streams with diverse QoS requirements transmitted to a mobile terminal over a common downlink shared channel. The queuing system, proposed for buffer management of aggregated single-user traffic in the base station of High-Speed Downlink Packet Access (HSDPA), allows for optimum loss/delay/jitter performance for end-user multimedia traffic with delay-tolerant non-real-time streams and partially loss tolerant real-time streams. In the queuing system, the real-time stream has non-preemptive priority in service but the number of the packets in the system is restricted by a constant. The non-real-time stream has no service priority but is allowed unlimited access to the system. Both types of packets arrive in the stationary Poisson flow. Service times follow general distribution depending on the packet type. Stability condition for the model is derived. Queue length distribution for both types of customers is calculated at arbitrary epochs and service completion epochs. Loss probability for priority packets is computed. Waiting time distribution in terms of Laplace–Stieltjes transform is obtained for both types of packets. Mean waiting time and jitter are computed. Numerical examples presented demonstrate the effectiveness of the queuing system for QoS optimization of buffered end-user multimedia traffic with aggregated real-time and non-real-time streams.     

Performance bounds for distributed systems with workload variabilities and uncertainties

Bounding techniques for queuing network models used to analyze the performance of parallel and distributed computer systems accept single values as model inputs. Uncertainties or variabilities in service demands may exist in many types of systems. Using models with a single aggregate mean value for each parameter for such systems can lead to inaccurate or even incorrect results. This paper proposes to use histograms for characterizing model parameters that are associated with uncertainty and/or variability. The adaptation of the well-known asymptotic bounds as well as balanced job bounds for single class queuing networks to histogram parameters is presented in the paper.     

一种使用历史信息的长期公平队列调度算法

经典的基于时延的队列调度仅关注调度时刻的队列延时,对历史信息并无记忆性,从而在业务流突变时排队延时公平性无法得到保障。本文提出一种长期公平调度算法,不仅综合考虑队列长度、到达速率等即时参量,而且考虑历史参量--历史延时累计量,使得调度更加理性,排队延时不会随流量突变而突变。另外,3个调度参量的关系式并非人为给出,而是通过对长期公平性的建模和对长期优化问题的推导得来。仿真对比了WRR, RPF, EDF算法,证实该算法中的队列延时不仅具有较高的公平性还具有较高的稳定性。     

Dynamic neural-based buffer management for queuing systems with self-similar characteristics

Buffer management in queuing systems plays an important role in addressing the tradeoff between efficiency measured in terms of overall packet loss and fairness measured in terms of individual source packet loss. Complete partitioning (CP) of a buffer with the best fairness characteristic and complete sharing (CS) of a buffer with the best efficiency characteristic are at the opposite ends of the spectrum of buffer management techniques. Dynamic partitioning buffer management techniques aim at addressing the tradeoff between efficiency and fairness. Ease of implementation is the key issue when determining the practicality of a dynamic buffer management technique. In this paper, two novel dynamic buffer management techniques for queuing systems accommodating self-similar traffic patterns are introduced. The techniques take advantage of the adaptive learning power of perceptron neural networks when applied to arriving traffic patterns of queuing systems. Relying on the water-filling approach, our proposed techniques are capable of coping with the tradeoff between packet loss and fairness issues. Computer simulations reveal that both of the proposed techniques enjoy great efficiency and fairness characteristics as well as ease of implementation.     

Stochastic Model for Communicating Retrial Queuing Systems with Cyclic Control in Random Environment1

The cybernetic approach is used to develop a mathematical model for communicating queuing systems. Conflicting input flows of the first queuing system and one of the input flows of the second queuing system are formed in a synchronous Markov random environment with a finite number of states. Another input flow of the second queuing system consists of retrials arriving from the first queuing system. The transition of a customer from the first queuing system to the second one takes a random amount of time. Servicing is performed by a cyclic algorithm with fixed duration.     

基于排队论模型的淮南市泉山华润苏果超市收银服务系统优化

对淮南市田家庵区华润苏果超市的收银服务系统进行分析研究，通过数据的采集、计算与分析对比，将排队论与数据模型相结合进行分析，建立了超市收银系统的排队模型，合理布置了收银台的数量，有效提高了超市工作效率和顾客满意度并且降低了超市运营成本，证明排队论在超市服务系统优化中具有实际用途。