Nowadays rail transit noise is one of the major environmental problems in the urban areas. The primary impact of transportation noise is related to annoyance instead of other effects such as hearing damage. Experiments studies have found that rail transit noise is mainly focused on the mid-frequency and low-frequency (50Hz-2 kHz), SPL of high-frequency noise decreases with the growth of frequency. As public awareness and action is growing, numerous measures have been taken to reduce rail transit noise in the development of new urban railway lines. Generally, noise barriers are more effective at high frequencies than at low frequencies, which generates conflicts since urban rail transit noise mainly occurs at low frequency and mid-low frequency. The simplest way to control low-frequency noise is to increase the height of noise barriers, which may lead to aesthetic and security issues. Rail transportation authorities of any countries in the world usually prevent from erecting very high barriers. According to the previous relevant studies, there are some other methods to improve noise reduction of a barrier by structural change. The slope of the top part of the noise barrier has been shown to provide improved performance, with the greatest attenuation of low-frequency sound for a slope of 120º. Besides, using perforated sheets within the diffuser can also shift the low-frequency of the barrier to lower frequencies as well. Combination of installation of absorbing material and structural change can also improve the performance of barrier at low frequencies by installing the material hard surface with an air gap. Overall, structural change is very crucial to low-frequency noise reduction and the principle of these methods referred above need to be analyzed in order to improve noise reduction of the barrier at low frequency. In order to validate whether it is effective to low-frequency noise reduction by the structural change, scale models of each method should be designed and tested. At the same time, it spends too much manpower, material and financial resources on laboratory tests of these scale models. Therefore numerical method should be utilized to simulate the sound field around the noise barrier. Among these numerical methods, boundary element method (BEM) and finite element-infinite element method (FEM-IEM) are widely applied in sound radiation problems. BEM has important advantages over the methods based on a geometrical theory of diffraction approach. Arbitrary shapes and surface acoustic properties can accurately be represented by BEM with high accuracy. Infinite element method is the extension of the finite element method. It can be applied to solve the inhomogeneous acoustic wave equation and it is more applicable to large computational structural acoustic problems. With the analysis and the comparison of several typical structural changes means by BEM or FEM-IEM, some helpful results would be provided for this research to design the optimum structure of noise barriers in the rail transit area. In conclusion, the numerical model of the straight barrier is firstly established with the validation of scale model experiments or field tests. Field tests of straight barrier or other kinds of noise barriers installed on the operating lines should be carried out in China, aiming to figure out the ILs (Insertion Loss) of barriers at sensitive positions. Results of these field tests will be analyzed and discussed not only for numerical model validation but also for evaluating the actual acoustic performance of noise barriers. Then, the solutions of low-frequency noise reduction are applied to evaluate the actual effects of noise reduction by numerical simulation. Finally, the optimum solution of the noise barrier in rail transit field is designed out with the combination of these effective technologies. A field test of the optimum solution should be carried out.