Average System Response Time Research Articles

Research on vehicle performance test based on DirectInput data acquisition

Aiming at the high cost of high-performance driving simulator at present, a low-cost vehicle simulation system with universal applicability was developed based on Microsoft Visual C++6.0 software platform. A data acquisition interface program based on DirectInput components and the real-time data acquisition program of driving operation was designed. Based on the dynamic model of vehicle simulation, a data output conversion model was established for the driver operation input collected in real time. For real-time vehicle dynamics simulation model, the real-time simulation control process was realized, the simulation description and decoding quality of the ground, hanging, no suspension quality coordinate system and the real-time simulation for buses with nine degrees of freedom dynamic model were established. Finally, the real-time and reliability of the simulation system were verified through the simulation test. The experiment results show that the average response time of the simulation system is 11.465ms. Emergency braking time is 4.40s with an error of 4.762%. The braking distance is 70.145m and the error is 5.959%.

Data-Driven Derivation of an Analytic Model for Parallel Servers With Job Replication

The job replication problem has been studied recently as a mechanism to improve performance and availability of systems with $n$ n parallel servers, each with its own queue. A dispatcher using some policy sends $d ~(1 \leq d \leq n)$ d ( 1 ≤ d ≤ n ) copies of a job to $d$ d of the servers. Copies are eliminated from the system as soon as the first copy completes from any of the $d$ d servers. This article introduces a data-driven method to derive closed-form expressions for the average response time and other metrics of jobs as a function of the degree of replication $d$ d . This method consists of developing a simulator for the system in order to generate a very large number of datasets for a wide range of input parameters. A statistical and visualization analysis of the data provides the analytical models. It is important to emphasize the difference between using simulation methods to obtain the value of metrics (e.g., average response time) of a computer system given values of input parameters and using our data-driven method to obtain closed-form expressions that relate output metrics to input parameters. The latter is the focus of our approach. The analysis presented here covers results for homogeneous and heterogeneous servers with exponentially distributed service times and for homogeneous servers with hypo-exponentially and hyper-exponentially distributed service times. This article also presents a closed-form equation for the optimal replication degree for the case of homogeneous servers with hypo-exponentially distributed service times.

Estimating the Response Time of a Cloud Computing System with the Help of Neural Networks

The article presents a new approach to assessing the average response time of a cloud computing system and its dispersion. A fork-join system or a system with request splitting was chosen as a queuing model, and artificial neural networks were used as a method for estimating a variable of interest. The analysis showed that the estimates obtained were more accurate than those previously known. Besides, the proposed approach allows expanding the analysis of the cloud system to the case of a model with a non-Poisson input stream and non-exponential service time, as well as obtaining estimates for a larger number of performance indicators of the cloud system, which was not previously possible.

Mean Field Evolutionary Dynamics in Dense-User Multi-Access Edge Computing Systems

Multi-access edge computing (MEC) can use the distributed computing resources to serve the large numbers of mobile users in the next generation of communication systems. In this new architecture, a limited number of mobile edge servers will serve a relatively large number of mobile users. Heterogeneous servers can provide either single resource or multiple different resources to the massive number of selfish mobile users. To achieve high quality of service (QoS) and low latency under these two cases, we construct two system models and formulate our problems as two non-cooperative population games. Then we apply our proposed mean field evolutionary approach with two different strategy graphs to solve the load balancing problems under those two cases. Finally, to evaluate the performance of our algorithms, we employ the following performance indicators: overall response time (average response time of the whole system), individual response time (response time of each server), and fairness index (equality of users' response time).

MNFTL

The write constraints of Multi-Level Cell (MLC) NAND flash memory make most of the existing flash translation layer (FTL) schemes inefficient or inapplicable. In this article, we solve several fundamental problems in the design of MLC flash translation layer. The objective is to reduce the garbage collection overhead to reduce the average system response time. We make the key observation that the valid pages copy is the essential garbage collection overhead. Based on this observation, we propose two approaches, namely, concentrated mapping and postponed reclamation, to effectively reduce the valid pages copy. Besides, we propose a progressive garbage collection that can well utilize the system idle time to reclaim more spaces. We conduct a series of experiments on an embedded developing board with a set of benchmarks. The experimental results show that our scheme can achieve a significant reduction in the average system response time compared with the previous work.

CHANCE: Capacitor Charging Management Scheme in Energy Harvesting Systems

The energy efficiency of emerging nonvolatile processors equipped with FRAM-SRAM memory makes them a promising solution for energy harvesting systems. To enable correct functionality and forward progress with an unreliable power supply, the system must accumulate sufficient energy in the capacitor to execute tasks atomically, even in the worst case scenario. Due to the large gap between the average and worst case energy consumption of tasks, state-of-the-art approaches like eM-map require a large capacitor to execute tasks on the SRAM. However, the size, cost, and charging time of the capacitor are major concerns in the energy harvesting systems. In this article, we proposed CHANCE, a capacitor charging management scheme that improves the capacitor size and average response time of an energy harvesting system. CHANCE analyses the energy consumption of tasks to set an appropriate capacitor size to make a balance between capacitor charging time and failure rate for each task. The results show that CHANCE improves the response time of state-of-the-art approaches up to 68% with a five times smaller capacitor.

Solving the Problem of Determining the Maximum Fault Tolerance of Distributed Information Processing Systems with a Limit on Their Reactivity

The use of exponential Queuing networks for calculating the reliability characteristics of distributed in-formation processing systems is proposed. This allows you to naturally tie the reliability characteristics of the system (failure rates) with its technical and operational characteristics (the average response time of the system to user requests). And also reduce the problem to calculating stationary probabilities of Queuing network States. The article formulates and solves the optimization problem of finding the maximum fault tolerance of distributed information processing systems implemented on the basis of a two-level and three-level client-server architecture with a fixed value of the average system response time to user requests. The problem of determining the fault tolerance indicators of distributed information processing systems is reduced to the problem of deter-mining the stationary probabilities of Queuing network States. The article presents the results of numerical ex-periments. The analysis of the obtained results shows the prospects of using these models to obtain estimates of the reliability characteristics distributed information processing systems in various subject areas.

Flat and hierarchical system deployment for edge computing systems

In this paper, we consider the server allocation problem for edge computing system deployment where each edge cloud is modeled as an M/M/c queue. Our goal is to minimize the overall average system response time of application requests generated by all mobile devices/users. We consider two approaches for edge cloud deployment: the flat deployment, where all edge clouds are co-located with the base stations, and the hierarchical deployment, where edge clouds can be co-located with other system components besides the base stations. In flat deployment, we demonstrate that the allocation of edge cloud servers should be balanced across all the base stations, if the application request arrival rates at the base stations are equal to each other; if the application request arrival rates are not the same, we propose a Largest Weighted Reduction Time First (LWRTF) algorithm to assign servers to edge clouds. Numerical comparisons of the proposed algorithm against several other reasonably designed heuristics verify that algorithm LWRTF has very good performances in terms of minimizing the average system response time. By theoretical analysis and numerical evaluations, we also show that, the hierarchical deployment approach has great potentials in minimizing the overall average system response time compared to the flat deployment approach. We also investigate the server allocation problem in hierarchical deployment and derive important insights to guide practical edge cloud server allocation in real-world systems.

The optimal placement of information resources on the nodes of a distributed information processing system based on a two-tier and three-tier client-server architecture

This paper presents the results of numerical experiments. The models the optimal allocation of information resources on the nodes of the distributed information processing system (IPS) were compared. They are implemented on the basis of a two-tier and three-tier client-server architecture. The article presents a conceptual model of the functioning of a distributed IPS based on a three-tier client-server architecture. It also shows a mathematical model for determining the stationary probabilities of network states and the average system response time to user requests.

Models of Optimal Allocation Information Resources on the Nodes of a Distributed Enterprise Information Processing System Using the Decomposition Approximation Method

In this article, using the decomposition approximation method based on the Norton theorem, the authors developed mathematical models of the functioning a distributed information processing system based on a local computer network of a two- level client-server and a three- level client-server architecture with an arbitrary function distribution of service time applications in the network nodes. A developed conceptual model of an equivalent two-node exponential queuing network and the basic mathematical relationships for calculating the intensity of service in the composition center, as well as expressions for calculating the average system response time to user requests are also presented. For the given mathematical models, using the heuristic algorithm developed earlier, the problem optimal allocation of information resources across nodes the computer network was solved by the criterion of the minimum average system response time to user requests. The results of numerical experiments are given, the analysis the results of which allows us to state about the effectiveness the developed models.

An Effective Attendance Monitoring System with Fraud Prevention Technique for Educational Institutions

In this article, an Attendance Monitoring System (AMS) is proposed for educational institutions. The novel aspect of this work is to integrate a password-protected attendance counting module with the system. The module can effectively limit the fraudulent activities in attendance monitoring process. Besides that, some constraints are proposed to increase the security of the AMS. Based on the constraints a new system is designed to maintain the attendance data and the user information. The aim of this research is to design an effective AMS using the Radio Frequency Identification (RFID) technology to prevent fraudulence. Besides, the attention is given to the usability and the efficiency of the system. For this purpose, a usability testing named System Usability Scale (SUS) is used after implementing the system in the real world. The usability test result shows that the proposed system has an average SUS score of 71.92 out of 100. The score indicates that the proposed system has the adjective rating, “good” (range 71.4 – 85.5). A performance testing is conducted by calculating the average response time for the system. The response time is collected by varying the number of RFID tags and the lengths of the serial cable. The average response time of the proposed system is found 1.11 second which is better than the traditional paper-based system.

Kriging-based simulation optimization: An emergency medical system application

Metamodeling is a common subject in simulation optimization literature. It aims to estimate the actual value (simulated) even before the point is evaluated by a simulation model. However, most publications do not apply metamodeling to models with real world complexity and size. This paper sought to apply Kriging to minimize the average response time of a Medical Emergency System by allocating ambulances throughout several city bases. Kriging is considered the state-of-art technique in metamodeling as it provides, in addition to the new point estimation, the level of prediction uncertainty. The optimization process followed the Efficient Global Optimization algorithm (EGO) and the Reinterpolation Procedure to deal with a stochastic simulation model. Finally, EGO was used to obtain a curve that reflected the relationship between the minimum response time and the total number of ambulances allocated to the city, representing significant information for healthcare public systems managers.

Response time analysis of a live-cube compact storage system with two storage classes

ABSTRACTWe study a next generation of storage systems: live-cube compact storage systems. These systems are becoming increasingly popular, due to their small physical and environmental footprint paired with a large storage space. At each level of a live-cube system, multiple shuttles take care of the movement of unit loads in the x and y directions. When multiple empty locations are available, the shuttles can cooperate to create a virtual aisle for the retrieval of a desired unit load. A lift takes care of the movement across different levels in the z-direction. Two-class-based storage, in which high turnover unit loads are stored at storage locations closer to the Input/Output point, can result in a short response time. We study two-class-based storage for a live-cube system and derive closed-form formulas for the expected retrieval time. Although the system needs to be decomposed into several cases and sub-cases, we eventually obtain simple-to-use closed-form formulas to evaluate the performance of systems with any configuration and first zone boundary. Continuous-space closed-form formulas are shown to be very close to the results obtained for discrete-space live-cube systems. The numerical results show that two-class-based storage can reduce the average response time of a live-cube system by up to 55% compared with random storage for the instances tested.

An Adaptive Demand-Based Caching Mechanism for NAND Flash Memory Storage Systems

During past decades, the capacity of NAND flash memory has been increasing dramatically, leading to the use of nonvolatile flash in the system’s memory hierarchy. The increasing capacity of NAND flash memory introduces a large RAM footprint to store the logical to physical address mapping. The demand-based approach can effectively reduce and well control the RAM footprint. However, extra address translation overhead is also introduced which may degrade the system performance. In this article, we present CDFTL, an adaptive Caching mechanism for Demand-based Flash Translation Layer, for NAND flash memory storage systems. CDFTL adopts both the fine-grained entry-based caching mechanism to exploit temporal locality and the coarse-grained translation-page-based caching mechanism to exploit spatial locality of workloads. By selectively caching the on-demand address mappings and adaptively changing the space configurations of two granularities, CDFTL can effectively utilize the RAM space and improve the cache hit ratio. We evaluate CDFTL under a real hardware embedded platform using a variety of I/O traces. Experimental results show that our technique can achieve an 11.13% reduction in average system response time and a 35.21% reduction in translation block erase counts compared with the previous work.

Collaborative location recommendations with dynamic time periods

This study proposes a location-based collaborative filtering recommendation system with dynamic time periods (LCFDTPs) for recommending timely and suitable points of interest (POIs) to mobile users. The system expedites calculating similarity based on POI recency and enables mobile users to promptly obtain recommended items that closely match their current space–time conditions by selecting different strategies for dissimilar situations. The performance of the proposed system was evaluated through simulations. The simulation results revealed that compared with three existing strategies, the proposed LCFDTP system demonstrated higher recommendation accuracy and coverage and a shorter average recommendation time. When the users’ moving velocity was set to 50 km/h and the query radius was set to 2 km, the recommendation precision of the proposed system was 61% higher than those of the other strategies. Moreover, the recommendation coverage and average response time of the LCFDTP system were 9% higher and 62% shorter than those of the other compared strategies, respectively. The LCFDTP system can also improve the POI recommendation quality by applying location-based services, thus enhancing user satisfaction.

QuLa: Queue and latency-aware service selection and routing in service-centric networking

Due to an explosive growth in services running in different datacenters, there is need for service selection and routing to deliver user requests to the best service instance. In current solutions, it is generally the client that must first select a datacenter to forward the request to before an internal load-balancer of the selected datacenter can select the optimal instance. An optimal selection requires knowledge of both network and server characteristics, making clients less suitable to make this decision. Information-Centric Networking (ICN) research solved a similar selection problem for static data retrieval by integrating content delivery as a native network feature. We address the selection problem for services by extending the ICN-principles for services. In this paper we present Queue and Latency, a network-driven service selection algorithm which maps user demand to service instances, taking into account both network and server metrics. To reduce the size of service router forwarding tables, we present a statistical method to approximate an optimal load distribution with minimized router state required. Simulation results show that our statistical routing approach approximates the average system response time of source-based routing with minimized state in forwarding tables.

A self-adaptive scheduling algorithm for reduce start time

MapReduce is by far one of the most successful realizations of large-scale data-intensive cloud computing platforms. When to start the reduce tasks is one of the key problems to advance the MapReduce performance. The existing implementations may result in a block of reduce tasks. When the output of map tasks become large, the performance of a MapReduce scheduling algorithm will be influenced seriously. Through analysis for the current MapReduce scheduling mechanism, this paper illustrates the reasons of system slot resources waste, which results in the reduce tasks waiting around, and proposes an optimal reduce scheduling policy called SARS (Self Adaptive Reduce Scheduling) for reduce tasks’ start times in the Hadoop platform. It can decide the start time point of each reduce task dynamically according to each job context, including the task completion time and the size of map output. Through estimating job completion time, reduce completion time, and system average response time, the experimental results illustrate that, when comparing with other algorithms, the reduce completion time is decreased sharply. It is also proved that the average response time is decreased by 11% to 29%, when the SARS algorithm is applied to the traditional job scheduling algorithms FIFO, FairScheduler, and CapacityScheduler.

On-Demand Block-Level Address Mapping in Large-Scale NAND Flash Storage Systems

The density of flash memory chips has doubled every two years in the past decade and the trend is expected to continue. The increasing capacity of NAND flash memory leads to large RAM footprint on address mapping management. This paper proposes a novel Demand-based block-level Address mapping scheme with a two-level Caching mechanism (DAC) for large-scale NAND flash storage systems. The objective is to reduce RAM footprint without excessively compromising system response time. In our technique, the block-level address mapping table is stored in fixed pages (called the translation pages) in the flash memory. Considering temporal locality that workloads exhibit, we maintain one cache in RAM to store the on-demand address mapping entries. Meanwhile, by exploring both spatial locality and access frequency of workloads with another two caches, the second-level cache is designed to cache selected translation pages. In such a way, both the most-frequently-accessed and sequentially accessed address mapping entries can be stored in the cache so the cache hit ratio can be increased and the system response time can be improved. To the best of our knowledge, this is the first work to reduce the RAM cost by employing the demand-based approach on block-level address mapping schemes. The experiments have been conducted on a real embedded platform. The experimental results show that our technique can effectively reduce the RAM footprint while maintaining similar average system response time compared with previous work.

Dynamic power management for QoS-aware applications

Reducing the power requirement of large IT infrastructures is becoming a major concern. Energy savings can be achieved with hardware and/or software solutions; in particular, modern CPUs can operate at different power levels that can be selected by software: low power modes reduce energy consumption at the cost of lowering also the CPU processing rate. In this paper we address the problem of reducing energy consumption of a large-scale distributed application subject to Service Level Agreements requiring a maximum allowed response time. Specifically, we propose Energy Aware reconfiguration of software SYstems (EASY), an on-line algorithm for dynamically adjusting the processing speed of individual devices such that the average system response time is kept below a predefined threshold, and the total power consumption is minimized. EASY uses a queueing networks performance model to proactively drive the reconfiguration process, so that the number of individual reconfiguration actions is reduced. We formulate the energy conservation problem as a Mixed Integer Programming problem, for which we propose a heuristic solution technique. Numerical experiments show that the heuristic produces almost optimal results at a substantially lower computational cost. Therefore, EASY can be effectively applied on-line to make a large system energy-proportional.

Improving the Human–Computer Dialogue With Increased Temporal Predictability

An experiment was conducted to investigate the impacts of length and variability of system response time (SRT) on user behavior and user experience (UX) in sequential computing tasks. Length is widely considered to be the most important aspect of SRTs in human-computer interaction. Research on temporal attention shows that humans adjust to temporal structures and that performance substantially improves with temporal predictability. Participants performed a sequential task with simulated office software. Duration and variability, that is, the number of different SRTs, was manipulated. Lower variability came at the expense of on average higher durations. User response times, task execution times, and failure rates were measured to assess user performance. UX was measured with a questionnaire. A reduction in variability improved user performance significantly. Whereas task load and failure rates remained constant, responses were significantly faster. Although a reduction in variability came along with, on average, increased SRTs, no difference in UX was found. Considering SRT variability when designing software can yield considerable performance benefits for the users. Although reduced variability comes at the expense of overall longer SRTs, the interface is not subjectively evaluated to be less satisfactory or demanding. Time design should aim not only at reducing average SRT length but also at finding the optimum balance of length and variability. Our findings can easily be applied in any user interface for sequential tasks. User performance can be improved without loss of satisfaction by selectively prolonging particular SRTs to reduce variability.