Search

Redundant code not only causes noise in code debugging which confuses developers, but also correlates with the presence of traditional severe software errors. RC-Finder, a redundancy detection system for large-scale code is proposed to detect six kinds of redundancy. This paper analyzes each kind of redundant code, and provides the detailed algorithm respectively. The experiments on large scale open source software systems show that RC-Finder can find redundant code efficiently. With RC-Finder, it is very convenient for developers to detect and correct these kinds of defects, and thereby to further guarantee the software quality.

Detecting bad smells in program design and implementation is a challenging task. Manual detection is proved to be time-consuming and inaccurate under complex situation. Weight based distance metrics and relevant conceptions are introduced in this paper, and the automatic approach for bad smells detection is proposed based on Jaccard distance. The conception of distance between entities and classes is defined and relevant computing formulas are applied in detecting. New weight based distance metrics theory is proposed to detect feature envy bad smell. This improved approach can express more detailed design quality and invoking relationship than the original distance metrics theory. With these improvements the automation of bad smells detection can be achieved with high accuracy. And then the approach is applied to detect bad smells in JFreeChart open source code. The experimental results show that the weight based distance metrics theory can detect the bad smell more accurately with low time complexity.

With the increasing of software requirements, software is becoming larger and larger. It becomes an important problem to maintain the software. Thus understanding the program exactly, rapidly and comprehensively plays an important role. This paper has presented the related research of program comprehension, summarized six typical program comprehension strategies and compared these strategies.

In order to extract the peaks of PHD, a novel method STPHD has been proposed recently. This method can provide more accurate target state estimates than the general clustering algorithm such as k-means clustering. This paper presents a version of STPHD for multi-sensor scene and makes two contributions. First, we generalize the STPHD algorithm to a multi-sensor scenario with an existing framework of fusion. The framework includes an association step and a fusion step. This generation can get better performance in accuracy. But the association step is time-consuming. The second contribution is a novel model for computing the cost of two sets of particles with sub-weights in the association step. The numerical simulation results show that the proposed method can significantly reduce the time cost with a very slight loss in accuracy compared with the previous methods.

Software errors can potentially lead to disastrous consequences. Unfortunately, debugging software errors can be difficult and time-consuming. A comprehension oriented software fault location approach (COFL) is proposed in this paper to provide automated assistance in bug location. It not only locates program predicates predicting bugs, but also provides high efficiency demand-driven data flow and control flow analysis to help developers understand the causes and contexts of bugs.

According to the background of network processor XDNP, this article describes a design of Fast Bus Interface(FBI) verification platform based on VMM architecture. All of the components and codes in the verification platform are achieved by SystemVerilog and assertions are also used in the process of verification. Through resonable usage of SVA it bacomes much easier to check out errors occuring in the executive process of FBI quickly and exactly. And then, the valuable functional verification results are obtained.

With the reduction of interconnect dimensions, time-dependent dielectric breakdown (TDDB) has become more important. An improved TDDB lifetime model is proposed according to a physics mechanism based on the model of Wen Wu, et al.. The TDDB lifetime temperature dependence obtained by the original model is contrary to the experimental data. The improved model presented here introduces a correctional factor and takes into account the effects of electric fields, temperature, and interconnect line spacing. The predictions of the new model regarding TDDB lifetime vs. temperature agree well with the previously published experimental data. The improved model has important implications for interconnect technology and design.

AutoLEP, an automated learning and examination environment for programming is introduced. It differs from the other existing systems mainly in that it adopts a novel marking mechanism. It combines static analysis with dynamic testing to analyze students’ programs. So it can evaluate not only the testing results but also the constructs of the programs. It can give reasonable marks to programs with syntactical or logical errors. AutoLEP has been used in the C programming course since 2004 at Harbin Institute of Technology. The advantages of using AutoLEP are that the students are assessed on what they have to do in practice, rather than on their theoretical knowledge, and that the marking load on the examiner is drastically reduced.

Code variations are widely believed to impede program analysis. This paper introduces a program normalization approach to remove code variations. Semantic-preserving transformations are performed on the system dependence graphs of programs. As a result, various syntactically different but semantically equivalent constructs are transformed to the same system dependence graph representation, so that code variations are removed. This approach establishes a good framework for testing the semantic equivalence of source codes and it can facilitate program analysis.