The discrete optimal transport (OT) problem, which offers an effective computational tool for comparing two discrete probability distributions, has recently attracted much attention and played essential roles in many modern applications. This paper proposes to solve the discrete OT problem by applying a squared smoothing Newton method via the Huber smoothing function for solving the corresponding KKT system directly. The proposed algorithm admits appealing convergence properties and is able to take advantage of the solution sparsity to greatly reduce computational costs. Moreover, the algorithm can be extended to solve problems with similar structures, including the Wasserstein barycenter (WB) problem with fixed supports. To verify the practical performance of the proposed method, we conduct extensive numerical experiments to solve a large set of discrete OT and WB benchmark problems. Our numerical results show that the proposed method is efficient compared to state-of-the-art linear programming (LP) solvers. Moreover, the proposed method consumes less memory than existing LP solvers, which demonstrates the potential usage of our algorithm for solving large-scale OT and WB problems.

Modern computing systems especially cloud-based and cloud-centric systems always consist of a mass of components running in large distributed environments with complicated interactions. They are vulnerable to performance problems due to the highly dynamic runtime environment changes (e.g., overload and resource contention) or software bugs (e.g., memory leak). Unfortunately, it is notoriously difficult to diagnose the root causes of these performance problems in a fine granularity due to complicated interactions and a large cardinality of potential cause set. In this paper, we build an automated, black-box and end-to-end cause inference system named CauseInfer to pinpoint the root causes or at least provide some hints. CauseInfer can automatically map a distributed system to a two-layer hierarchical causality graph and infer the root causes along the causal paths in the causality graph. CauseInfer models the fault propagation paths in an explicit way and works without instrumentation to the running production system, which makes CauseInfer more effective and practical than previous approaches. The experimental evaluations in two benchmark systems show that CauseInfer can identify the root causes in a high accuracy. Compared to several state-of-the-art approaches, CauseInfer can achieve over 10 percent improvement. Moreover, CauseInfer is lightweight and flexible enough to readily scale out in large distributed systems. With CauseInfer, the mean time to recovery (MTTR) of the cloud systems can be significantly reduced.

Even well-designed software systems suffer from chronic performance degradation, also known as “software aging”, due to internal (e.g., software bugs) or external (e.g., resource exhaustion) impairments. These chronic problems often fly under the radar of software monitoring systems before causing severe impacts (e.g., system failures). Therefore, it is a challenging issue how to timely predict the occurrence of failures caused by these problems. Unfortunately, the effectiveness of prior approaches are far from satisfactory due to the insufficiency of aging indicators adopted by them. To accurately predict failures caused by software aging which are named as Aging-Related Failure (ARFs), this paper presents a novel entropy-based aging indicator, namely Multidimensional Multi-scale Entropy (MMSE) which leverages the complexity embedded in runtime performance metrics to indicate software aging. To the best of our knowledge, this is the first time to leverage entropy to predict ARFs. Based upon MMSE, we implement three failure prediction approaches encapsulated in a proof-of-concept prototype named ARF-Predictor. The experimental evaluations in a Video on Demand (VoD) system, and in a real-world production system, AntVision, show that ARF-Predictor can predict ARFs with a very high accuracy and a low Ahead-Time-To-Failure (ATTF). Compared to previous approaches, ARF-Predictor improves the prediction accuracy by about 5 times and reduces ATTF even by 3 orders of magnitude. In addition, ARF-Predictor is light-weight enough to satisfy the real-time requirement.

Time series prediction problems can play an important role in many areas, and multi-step ahead time series forecast, like river flow forecast, stock price forecast, could help people to make right decisions. Many predictive models do not work very well in multi-step ahead predictions. LSTM (Long Short-Term Memory) is an iterative structure in the hidden layer of the recurrent neural network which could capture the long-term dependency in time series. In this paper, we try to model different types of data patterns, use LSTM RNN for multi-step ahead prediction, and compare the prediction result with other traditional models.

In recent years, with the large-scale growth of the high-energy physics experimental data, the performance of metadata retrieval based on disk storage has been gradually reduced, which can not meet the retrieval performance requirements of EB-level high-energy physics experimental metadata. To solve this problem, a method of converting traditional directory structure storage into RAMCloud storage is proposed. The core idea of this method is to use Key-Value non-relational database to re-design the traditional directory tree, separate directory structure and directory node content, and add a secondary index for parent directory, which can give full play to Key-Value retrieval and memory storage advantages, improve search efficiency. Through the implementation of the test, showed that the method has a better performance. Compared to the storage based on Mysql, the retrieval time drops significantly in the case of increased data.

Modern applications especially cloud-based or cloud-centric applications always have many components running in the large distributed environment with complex interactions. They are vulnerable to suffer from performance or availability problems due to the highly dynamic runtime environment such as resource hogs, configuration changes and software bugs. In order to make efficient software maintenance and provide some hints to software bugs, we build a system named CauseInfer, a low cost and blackbox cause inference system without instrumenting the application source code. CauseInfer can automatically construct a two layered hierarchical causality graph and infer the causes of performance problems along the causal paths in the graph with a series of statistical methods. According to the experimental evaluation in the controlled environment, we find out CauseInfer can achieve an average 80% precision and 85% recall in a list of top two causes to identify the root causes, higher than several state-of-the-art methods and a good scalability to scale up in the distributed systems.

The increasingly wide range of RFID applications and the characteristics of RFID data, such as massive, continuous, require RFID middleware to provide efficient, real-time data processing capability, and to support to extract meaningful complex event from meaningless simple low-level events. OSGi is a kind of lightweight, loosely coupled, service-oriented application development framework. System based on OSGi can install, start, stop and uninstall certain components of the system in the run-time environment if application needs, because the system has some features like component-based, reconfigurable and dynamic management. This paper discusses the overall technology solutions of RFID middleware into which the OSGi technology is added. By adding complex event processing layer based on OSGi into RFID middleware, the middleware can detect complex event defined by the application and accept the query modification without stopping. Considering out-of-order event from event source, the paper presents the improved detection algorithm for out-of-order event which is the extension of the traditional data stream approach and causes complex event detection more practical and precise.

Software rejuvenation is a preventive and proactive technology to counteract the phenomenon of software aging and system failures, and to improve the system reliability. In this paper we present a mixed software rejuvenation policy for an operational software system with multiple degradation states, which considers both the history information and the current running state. By this policy, the system is rejuvenated when it achieves to a degradation threshold or it comes to the pre-determined rejuvenation interval. For comparison, standard rejuvenation policy is also discussed. Continuous-time Markov chains are used to describe the multiple degradation states model. To evaluate these polices expediently, we utilize deterministic and stochastic Petri nets (DSPN) to solve the models. Numerical results show that the deployment of software rejuvenation in the system leads to significant improvement in availability and throughput. And the mixed rejuvenation policy is better than the standard rejuvenation policy.

With the chip multi-processor (CMP) being more and more widespread used in the laptop, desktop and data center area, the power-performance scheduling issues are becoming challenges to the researchers. In this paper, we propose a multi-objective fuzzy genetic algorithm to optimize the energy saving scheduling tasks on heterogeneous CMP system. According to the characteristic of heterogeneous CMP system, we present a novel encoding and decoding scheme of genetic algorithm, improve the crossover operator and the mutation operator. Based on that, we improve the genetic algorithm architecture by using the relative fuzzy membership grade fitness and the elitist strategy. Simulation results demonstrate that using our algorithm can save both the execution time and system energy cost at the same time.

With the increased interest in the Web services composition, more and more enterprises and businesses depend on this paradigm. Open, distributed and dynamic properties of the schema, there is a pressing need for secure services in daily transactions. Orchestration and choreography language provide basic services standards and interaction, collaboration, and negotiation standards among services, but they are not give any secure manners or secure operation styles and specifications. Despite the interest of such security mechanisms, a formal module of them is still lacking. For giving general guide to implement secure orchestration and choreography language, we give a formal approach to carry out those goals. To this target, we emphasize on those by designing an extension of the Spi calculus with Secure Global Calculus. The Spi calculus precisely identifies orchestration secure properties of each principal from a local viewpoint. The secure global calculus describes an interaction secure choreography scenario from a vantage point of view. We called our method SpiG4WSC calculus. We believe that the combination of strong practical needs for dynamic secure Web services composition and the theoretical foundations will lead to a bridge between practice and theories. The contribution of this paper are (1) giving the syntax and semantic of SpiG4WSC calculus; (2)applying the calculus to give a model to presenting the secure orchestration, emphasizing on the formal basis for secure services; (3)describing the secure choreography, giving the formal frame for interaction processes.

Service Oriented Architecture (SOA) and Service Oriented Computing (SOC) are prevailing paradigms for sharing and reusing resources. Service composition is a methodology widely used in SOA and SOC to build new value-added services on primitive services on-the-fly to support online Business-to-Business collaborations. Requirements of customers to these composite services include functionality and non-functionality. Since many services can have the similar functionality, the non-functionality of composite services, such as Quality of Services (QoS), is the important metrics to distinguish a service from each other and find an optimal program to meet the requirements of customers. This paper firstly proposes a system model from the view of resource and value, and then we introduce an improved heuristic algorithm for the selection of composite services with multiple constraints. The simulation experiments show an outperforming result of proposal algorithm in both utility performance and time cost comparing with the other heuristic algorithms.

There are two important factors in the power-performance issues of chip multi-processor(CMP) system: the execution time of tasks and the system energy consumption. Most of exist energy saving methods are not designed to reduce the system energy while cut the execution time down. This paper represents a multi-objective hybrid genetic algorithm (MHGA) which can make the execution time of tasks minimize while reducing the system power consumption. We analyze the problem of energy saving task scheduling on CMP system and a novel coding scheme of genetic algorithm. Based on that, we improve the crossover and mutation operator of genetic algorithm. We propose the multi-objective genetic algorithm by using simulated annealing algorithm to enhance the search ability. Simulation results demonstrate that using our algorithm can make the efficiency of task scheduling on CMP increase, make both the execution time of task and energy consumption of system decrease.

Nowadays, smart spaces occupy an essential part of ubiquitous computing environment. The spaces integrated with wireless sensors networks, actuators and context-aware services become part of our daily life. Smart spaces are equipped with a large number of wireless sensors that aim to collect large quantities of context information, during the process, there exists a large amount of collisions and energy consumption. Therefore, this paper provides a novel multi-rate based local framing pre-schedule scheme to further reduce collisions and improve energy efficiency in CSMA/TDMA hybrid MAC layer of wireless sensor network. This MAC combines CSMA and TDMA functionalities together while obviates their shortcomings. Having been assigned, slot 0 is preserved as the pre-schedule slot, to inform neighbor nodes the schedule of the senders. During the pre-schedule slot, each node knows exactly the schedule of other neighbor nodes. Multi-rate and power scaling are applied to achieve further energy saving by adpoting an acceptable rate rather than maximum rate. Data rate is dynamically adjusted according to the traffic load of sending nodes, in an energy efficient data rate, to save energy. Being compared with Z-MAC in terms of performances, local framing pre-schedule and multi-rate in this experiment achieved further energy efficiency. Index Terms--MAC, CSMA, TDMA, Mult-Rate, Wireless Sensor Networks

Insulin pump system is a safety-critical embedded system controlling the amount of injection of insulin to diabetics based upon their blood glucose levels, and the high reliability of the software used in the pump is crucial. One way to achieve the high reliability of software is to build an accurate and complete model through effective analysis and specification, and to implement the system based upon the specification. In this paper, we describe how the SOFL formal engineering method is applied to develop a specific insulin pump system in practice. In particular, we focus on the issue of how the three-step modeling approach advocated by the SOFL method, which includes informal, semi-formal, and formal specifications, is utilized to obtain a precise and valid specification of the embedded software for the insulin pump system. We also discuss how the specification benefits the implementation of the system, and report our experience and lessons learned.

By using personal information in a pervasive computing environment, context-aware applications can provide appropriate services for people. This personal information is often involved in personal privacy. In order to protect personal privacy concerns about personal information, privacy role is proposed to control access personal information. We also construct an information system about the privacy decision of personal information disclosure based on people's interaction history. In the initial period of personal information disclosure, the privacy decision is made by people and the information system is constructed based on the decision data. Then privacy disclosure policies are extracted from this information system using rough set theory. According to deducing from the privacy disclosure policies and people's context information, the contextaware application is assigned to an adequate privacy role. It reduces the distraction of privacy decision for people. A case study further shows the proposed method is effective. Finally, it provides about the overload performance of privacy role analysis personaengine.

In pervasive environment, context-aware service provider can use personal information to customize the adequate services for end users. Personal information is people's privacy concern. Therefore, people need privacy control methods. In this paper, we analyzed privacy control from two aspects: personal privacy model about information disclosure and the function of context-aware service provider. According to the analysis, we designed the components about context-aware privacy control in order to minimize the burden of personal privacy decision about personal information disclosure. The simulation experiment shows that it is possible method for the proposed privacy protection mechanism. Finally, we also proposed conceptual context-aware architecture to control personal information disclosure.

According to the characteristics of the operational behavior and runtime state of application sever, the resource consumption time series are observed and modeled by fuzzy wavelet network (FWN) with fuzzy logic inference and learning capability. The objective is to model the extracted data series of systematic performance parameters to predict software aging in application server. The dimensionality of input variables of FWN is reduced by principal components analysis (PCA), and the structure and parameters of FWN are optimized with adaptive genetic algorithm (GA). Judging by the model, we can get the aging threshold before application server failed and preventively maintenance the application server before systematic parameter value reaches the threshold. The experiments are carried out to validate the efficiency of the proposed model and show that the aging prediction model based on FWN with adaptive genetic algorithm is superior to the neural network (NN) model and wavelet network (WN) model in the aspects of convergence rate and prediction precision