Search

At present, when the device management application programs the devices (such as mobile terminals, Internet of things terminals and devices, etc.), buffer overflow will inevitably occur due to the defects of filter input condition setting, variable type conversion error, logical judgment error, pointer reference error and so on. For this kind of software and its running environment, it is difficult to reduce the false positive rate and false negative rate with traditional static detection method for buffer overflow vulnerability, while the coverage rate of dynamic detection method is still insufficient and it is difficult to achieve full automation. In view of this, this paper proposes an automatic dynamic detection method based on boundary testing, which has complete test data set and full coverage of defects. With this method, the input test points of the software system under test are automatically traversed, and each input test point is analyzed automatically to generate complete test data; driven by the above complete test data, the software under test runs automatically, in which the embedded dynamic detection code automatically judges the conditions of overflow occurrence, and returns the overflow information including the location of the error code before the overflow really occurs. Because the overflow can be located accurately without real overflow occurrence, this method can ensure the normal detection of the next input test point, thus ensuring the continuity of the whole automatic detection process and the full coverage of buffer overflow detection. The test results show that all the indexes meet the requirements of the method and design.

Because of the continuous promotion of cloud computing applications, the demand for data processing in cloud computing is increasing. Users have higher requirements for the service quality of cloud computing, the high efficiency of cloud computing task scheduling algorithm plays a key role in the cloud computing. How to scheduling the computing resources efficiently, all tasks can be completed in the least time and cost is an important issue in cloud computing research. In this paper, a method of initial optimization on the crossover mutation probability of adaptive genetic algorithm (AGA) using binary coded chromosomes is proposed. Through experiments, the improved adaptive genetic algorithm is compared with the adaptive genetic algorithm (AGA) and the standard genetic algorithm (SGA). The experimental results show that the improved algorithm is an effective cloud computing task scheduling algorithm.

Cloud computing resource scheduling is a complex NP problem and difficult to solve. In order to shorten task completion time in cloud resource scheduling, an improved LDW-PSO (linearly decreasing weight-particle swarm optimization) algorithm is proposed. Firstly, for the fact that PSO algorithm is easy to fall into local convergence, based on the linearly decreasing weight strategy, the constant disturbance is added to increase the inertia weight, for the purpose of getting rid of local search and begining global search. Secondly, in order to avoid the situation that particles highly gather around the optimal particles, resulting in being similar and damaging the diversity of particle swarm, thus, by changing inertia weight mixed with random individuals Adaptively in a certain probability, it could better maintain the diversity of population. Finally, through different simulation tests on Matlab2010a platform, proving that the improved LDW-PSO algorithm can get a more accurate solution and optimize completion time in cloud computing resource scheduling.

MapReduce is an abstract model of processing big data with the parallel computing framework. However it is heavy that the model for the data intensive computing, DIC, depends on the data transmission between the cluster's nodes which causes the overload on the master node handling the messages used to communicate with the nodes in the cluster. Therefore, this paper proposes an improved model of MapReduce based on message middleware, to improve the performance of DIC. This paper describes the principles and mechanism of the improved model in detail.

The Component Library(CL)is the key to realize software reuse, the CL available on the network is decentralized, independent, not easy to access at present and existing component libraries face infrastructure supply and maintenance challenges. The authors use cloud computing model, which contribute to resolve the issues of visiting different component libraries and the scalability of component library resources. In this paper, we discuss the plan that placed the CL in the cloud infrastructure and put forward a hierarchical organizational model of component resources based on cloud storage to achieve component integration services. We also explore the component library retrieval framework in the cloud and verify the cloud model applied to component library retrieval based on the Hadoop platform can improve the system performance and achieve the scalability of the computing power.

The architectural description and design decide the quality and life of the software. Cross-cutting concerns as aspects crosscut multiple architectural components. Due to these cross-cutting concerns the metrics for cohesion and coupling are affected, the scenario result in lower values for modifiability and reuse quality attributes. Accordingly, the concept of aspect orientation is required to be addressed in architecture design level. But now a lot of methods about architectural description fail to accommodate the description for cross-cutting concerns. In this paper, on the based of standard for ldquoRecommended practice for architectural description of software-intensive systemrdquo in IEEE 1471-2000, we propose the Use-Case Based Aspect-Oriented Conceptual Model for Architecture Description to explicitly address aspects in architecture design. It is more favorable to maintain, reuse and evolve software system. And we applied this model into Web Service, shaped Aspect-Oriented Extension Architecture Description Model for Web Services to show the advantages of this approach.

Introduce the pigs' payoffs in society, analyse the negative infection of pigs' payoffs on the peer-to-peer network, base on usability, reliability, security and extensible speciality. And give a solution for the question base on the incentive mechanism. Finally, future works about the mechanisms discouraging pigs' payoffs in P2P networks are outlined.

The maintenance of software complexity and management of its evolution is two most important points to be solved by software engineering. The separation of concerns is a kind of solution. Thus aspect oriented software development (AOSD) was aroused and shed new software evolution problems. For example separating or integrating crosscutting concerns. In this paper we focus on design a model weaving framework to support aspect-oriented dynamic software evolution in architecture level. Hence we propose a MDA based model weaving framework (MMWF), to formalize the underlying mechanisms of aspect dynamic weaving in architecture level. That separates concerns and improves the abstract level by modeling the crosscutting concerns to aspect model and dynamic integrate them into base component. And it is well regulated and executable as well as achieve the software easy evolution and maintenance.

With the rapid development of the current information networks, there are various databases, operating systems and network platforms coexisting in the networks. Therefore, it is important to find a remote interactive way to manipulate the various databases on a uniform operating platform. We demonstrated the feasibility of such a way in the three-tier client/server architecture (BWD) based on the theory of telecommunications management networks (TMN), and the simulation of which in an experimental environment has been realized.