As heterogeneous multicore real-time systems are increasingly applied in safety critical systems, it is important to ensure the correctness of these systems. One key attribute of real-time systems is the schedulability that guarantees to satisfy the timing requirements. This paper presents a method for modeling and verifying schedulability of heterogeneous multi-core systems and the method we present uses timed-automata (TA) to model tasks and resources of heterogeneous systems considering their special features. Also this method allows us to establish complex dependences between tasks and use different scheduling policies. After that we choose CPU-GPU heterogeneous multi-core systems as an example and we model three TA networks according to three levels of this system, which are real-time tasks, resources and scheduling management modules. Finally, we use UPPAAL to verify if the model we established satisfies habitudes. According to our method, we present a link between model checking methods and schedulability analysis method for heterogeneous multicore real-time systems and we can automatically and accurately verify the schedulability of selected systems.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 13

A simplified BDD analysis method (SimBDD) is proposed to solve the problem of temporal and spatial explosion and low computational efficiency of getting the fault tree's minimum cut sets in BDD method. On the basis of using SimBDD to get the minimum cutting sets, the failure paths of the fault tree are obtained by the hierarchical processing method. Tests are taken on large fault trees with this method. The results demonstrate that the method is efficient.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 11

Assurance case has been widely used to justify safety, security and other properties of systems. The extent we can trust the assurance case, i.e., confidence in assurance case, is still an open issue in the area. This paper proposes a subjective logic based approach to assess the confidence in assurance case. Compare to other quantitative tools such as Bayesian Belief Network or Dempster-Shafer theory, subjective logic can (1) handle uncertainty and vagueness that usually are contained in expert opinions, and (2) provide appropriate logic operators to deal with different argument types. In the paper, we firstly define the concepts of confidence, sufficiency and necessity to measure the affecting factors to the confidence. Then, we identify four basic argument types and define confidence propagation rules for them. The confidence in assurance case can be calculated iteratively from the bottom up. The application of the proposed approach is illustrated by an industry case study.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 27

The color and shape features provide more contributions for vehicle images in criminal investigation generally. A criminal investigation library that was used in studying criminal tools was set up, in which the number of the car images is more than one thousand. In this paper, the color of the car pictures was quantized by the non-uniform quantization, the triangular and the trapezoidal fuzzy membership degree functions respectively at the first step, and then the Euclidean distance and the weighted distance similarity measures methods are used at the second step, the last step is that the car images were retrieved by the above six algorithms in the criminal investigation library. The results show that the weighted distance similarity measure algorithm based trapezoidal membership degree is better than others; meanwhile the precision and recall are significantly higher than other else. The triangular fuzzy quantization algorithm is not very different from the non-uniform quantization algorithm, and even the precision and recall ratio is smaller than the latter. Under the same quantization condition, the weighted distance algorithm is slightly better than the Euclidean distance, but the difference is not great.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 14

Single reliability growth model usually only captures partial knowledge of a failure process. A combinational model tries to capture more knowledge by integrating two or more reliability growth models. Unlike the existing linear combinational models that simply adds up the weighted results by G-O, M-O and L-V model, this paper proposes the combinational model from G-O and S-Shaped model, at the level of failure distribution to reduce fitting errors and to maintain the mathematical properties of non-homogenous Poisson process. To evaluate the effectiveness of the proposed model, we use the failure data sets (21 projects) available in public sources. Ten out of the twenty-one projects, which pass the distribution test and have feasible solutions in parameter estimation, are selected to conduct experiments. We use mean squared error (MSE) to evaluate the historical predictive validity. The results show that our model is consistently stable and has lower MSE. It reduces 51.3% MSE of G-O, 67.2% MSE of S-Shaped, and over 56% MSE of the three linear combinational models in average. The proposed model tends to have a larger estimation of the expected number of failures, which can overcome the under estimation by G-O and S-Shaped model in some degree.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 28

Fault-tolerant embedded systems can provide the correct service with the active faults. It is important to verify the ability of fault tolerance in system design phase especially for safety-critical systems. Besides, the verification of models can reduce the cost of the system development. The Architecture Analysis and Design Language (AADL) and its Error Model Annex (EMV2) provide the ability to model a fault tolerance system. The error event, error propagation and error state machine provided by the AADL Error Model Annex can model an embedded system. However, there is a problem that whether the model satisfies the requirement of fault tolerance or not. We design a component-based algorithm to verify the ability of fault tolerance. The error and warning messages will be produced by our algorithm. Finally, a plugin based on the Osate2 tool and a case study are given.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 27

In this paper, a novel accelerating Ant Colony Optimization (ACO) algorithm based on High-Level Synthesis (HLS) on FPGA (Field Programmable Gate Array) is proposed. The proposed algorithm (HACO-F) is implemented by C/C++ programming language and calculated by floating-point. For the sake of accelerating, the algorithm mainly employs the data optimization strategy to redefine the variables precision in HACO-F to reduce resource utilization and energy consumption. Then, we explore a loop optimization strategy including pipeline and unroll to parallelize loops in HACO-F to decrease computation time. The experimental results show that the HACO-F algorithm can achieve more than 6 times accelerating performance than that of the AS (Ant System) at the same search ability. The resource utilization in HACO-F is 1% FF, 4% LUT, and 9% BRAM decrease. The total on-chip energy consumption of HACO-F is reduced by 23.9%.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 16

Reliability-based design optimization (RBDO) plays an essential role in structure and system design. However, its application in practical engineering is hindered by the huge computational cost in performance function evaluation. In this paper, a moving optimal radial sampling (MORS) method is proposed for RBDO problems to substantially improve the computational efficiency of Monte Carlo simulation (MCS). In MORS, the failure probability and its gradient are calculated using radius based importance sampling (RBIS) method. The initial radius is selected according to the target reliability, which can also be used to check the feasibility of probabilistic constraints afterwards. The arc search scheme in enhanced performance measure approach (PMA+) and linear interpolation scheme are used to calculate the optimal radius of RBIS. After the failure probability and its gradient are calculated, the optimal design is obtained using sequential approximation programming (SAP). The computational capability of the proposed MORS method is demonstrated using a honeycomb crashworthiness design application, a nonlinear mathematical problem and a speed reducer design. The comparison results show that the proposed MORS-SAP method is very efficient and accurate.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 32

In view of the lack of subjectivity and accuracy in the traditional micro-blog opinion leader recognition method to measure the important factors of users, a new micro-blog opinion leader recognition method is proposed. This paper used the linked data to describe the micro-blog data, used the association rule mining algorithm to quantitatively determine the important factors that affected the users’ ranking, and constructed the opinion leader recognition model according to the index scoring method. Experiments show that our method using linked data identifies the opinion leaders the same as the standard leaders, is more accurate and has better feasibility than that of traditional data.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 12

In order to formalize the correctness of real-time systems, strong timed bisimulation in TCCS has been proposed to characterize the relation between implementation and specification. The usual action and time delay must be the same in strong timed bisimulation. However, in some real situations, many real-timed systems can not satisfy the exact match. In this paper, in order to characterize the approximate usual action and time delay, the strong timed bisimulation in TCCS is generalized to numerical version. Firstly, the definition of global timed bisimulation index of a binary relation is established to describe the relation between implementation and specification. Secondly, in order to quantify the approximate degree between implementation and specification, the global timed lambda-bisimulation is defined. Finally, the congruence of the global timed lambda-bisimulation is proven to guarantee the modular development and hierarchic design methods which are used in the real software development.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 24

In recent days, rail transit accidents happen from time to time, but the causes are difficult to be found. According to the stochastic and real-time characteristics of equipment faults, three layer models based on stochastic hybrid automata (SHA) are proposed for interlocking systems. The three layer models consist of a system model, a monitoring model and a fault prediction model. The accidents caused by the equipment faults are predicted by simulating these models together on UPPAAL-SMC platform. The main contributions of this paper include: (1) extracting model patterns for interlocking systems (2) presenting a pattern-based system model generation process and an automatic generation method of monitoring model based on time constraints and (3) defining the accidents prediction model of collision accidents to predict the accidents and monitoring accident causes through model simulation.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 19

In order to realize the scientific management of the light environment and achieve the purpose of saving energy and increasing efficiency for the hotel, this paper designs and implements an intelligent lighting control system for the room, manual control and automatic control of dual management for the room lighting [6]. The system uses the multi-layer distributed structure, combining organically with the embedded technology, Internet of things technology, sensor technology, CAN bus technology and so on, and the embedded ARM microprocessor is the core of the system, combining with the intelligent gateway module, I/O weak current digital input and output module, dimming module, illuminance sensor module and power module, and communicating with each other by the CAN bus, to realize the intelligent management of the room lighting. The Lighting Control System for Smart Hotel Rooms is facing up to the high star hotel, to improve the management level and service quality of the hotel, and save energy, scientific and intelligent management, and enhance the competitiveness for the hotel, it will be an important part of the Smart Hotel System which has a certain application value [14]. As a new and most promising light source, all kinds of lighting have attracted more and more attention for their energy saving, environmental protection and other advantages.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 20

Model based software testing is one of the most popular software quality assurance techniques adopted by industrial practitioners. The SysML Activity Diagram (AD) can describe dynamic behaviors of a software system under testing in an intuitive way. That is, the AD is a promising foundation for generating test case to test the target software system. Unfortunately, there are few effective AD based testing approaches for industrial practitioners due to the lack of automated generation technique and powerful tool support, especially for whose shape is out of structure. To tackle this problem, we propose an automated generation approach with a supporting tool. For a specific AD, we first transform it into an intermediate representation form — Intermediate Black box Model (IBM). Then the IBM is used to generate test cases automatically. The approach presented in this paper can make up the deficiency of automated test case generation with the unstructured SysML AD.

(This paper was presented at the Third International Symposium on System and Software Reliability.)
Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 21

"Traffic measurement is the basis of analysis and prediction of network traffic. Its accuracy directly affects the reliability of upper applications. This paper proposes a rate-adaptive traffic measurement method called AdaRate. It adjusts the polling period according to fluctuations of flow rates which is calculated with a sliding window queue. We noticed that some fluctuations of flow rates may be dropped in AdaRate while the polling period is too long. Then we propose RAdaRate. It adjusts the polling period according to AdaRate or sets the polling period to minimum randomly. We compared the performance of AdaRate and RAdaRate with SWT in CeMon and Payless in Mininet when monitoring those VBR flows with frequent and large fluctuation. The results show that the accuracy of RAdaRate is the best while the overhead of AdaRate is the least. The overhead of AdaRate is 68.75% less than periodic polling.

(This paper was presented at the Third International Symposium on System and Software Reliability.) "
Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017
References: 16

The processing of streaming data implies new requirements concerning restrictive processing time, limited amount of memory and one scan of incoming instances. One of the biggest challenges facing data stream learning is to deal with concept drift, i.e., the underlying distribution of the data may be evolving over time. Most of the approaches in the literature are under the hypothesis that the distribution of classes is balance. Unfortunately, the class imbalance issue is common in the real-world. And the imbalance issue further increases the difficulty of solving the concept drift problem. Motivated by this challenge, a novel ensemble classification for mining imbalanced streaming data is proposed to overcome both issues simultaneously. The algorithm utilizes the under-sampling and over-sampling techniques to balance the positive and negative instances. Moreover, dynamic weighting strategy was adopted to deal with concept drift. The experimental results on synthetic and real datasets demonstrate that our proposed method performs better than competitive algorithms, especially in situations where there exist concept drift and class imbalance.

Submitted on July 25, 2017; Revised on August 30, 2017; Accepted on September 15, 2017(This paper was presented at the Third International Symposium on System and Software Reliability.
References: 26

Underwater Wireless Sensor Networks (UWSNs) are increasingly applied to explore underwater environments and resources. Some characteristics of mobile UWSNs, such as energy restriction, low transmission rate, node mobility and multiplicity, affect networks' performance significantly. In this paper, a Mobile-MultiSink (MMS) routing protocol is proposed for UWSNs with multiple mobile sink nodes. In the proposed MMS routing protocol, sensor nodes report events to multiple sink nodes using dynamic routing. There are two phases in this protocol, the layering phase and the construction phase. In the layering phase, each sink node establishes a layered structure, which needs to be refreshed periodically. Taking into consideration that the packet delivery rate (PDR) will reduce because of the mobility of nodes, we stipulate that every sink node should refresh its layered structure every time the PDR comes below a critical value. In the communication phase, the intermediate relay nodes are selected to forward data to the target sink node. This phase contains the selection of target sink node and the selection of relay nodes. We evaluate the performance of MMS using NS-3. Our findings demonstrate that the proposed protocol has better packet delivery rate and lower energy consumption in UWSNs with multiple mobile sink nodes.

Submitted on April 29, 2017; Revised on July 12, 2017; Accepted on August 23, 2017
References: 19

Vehicle routing problem is one of hot issues in the past decade; the paper proposes an important variant of vehicle routing problem, which we name time probabilistic vehicle routing problem. We find the time spent on the road from point A to point B always changes because of dynamic road condition, dynamic weather et al. However, the probability is certain that the vehicle can travel from point A to Point B within certain time intervals. Ant colony algorithm is an efficient distributed algorithm, and has been widely used in various fields. The ant colony algorithm is introduced to obtain the optimal solution by observing the characteristics of the problem. The Ants also search for the next transfer node according to probability in ant colony algorithm. Ant colony algorithm can solve this kind of problem effectively.

Submitted on April 29, 2017; Revised on August 12, 2017; Accepted on September 23, 2017
References: 13

In consideration that Uyghur symbols are poorly recognized and recognition algorithms are impertinent for connecting characters, five kinds of features are extracted from Uyghur symbols after feature analysis and pre-processing, including 8-directional features, fuzzy features and primitive features. According to characteristics of Uyghur, features of Uyghur symbols such as aspect ratio and dynamic speed are extracted. Features are selected by LDA transformation, and the methods for judging relationships among Uyghur primitives are determined. Minimum distance classifier and MQDF classifier are suggested to be used. By recognizing primitives, effectiveness of algorithms is experimentally verified with online Uyghur recognition algorithms of decision fusion strategies by classifiers.

Submitted on May 29, 2017; Revised on July 12, 2017; Accepted on September 17, 2017
References: 12