Research On Wireless Channel Parameter Extraction Under High-speed Railway Scenario

With the rapid development of high-speed railway in our country, the role it played in people’s life is becoming more and more important. The high-speed railway wireless channel environment learning and the high-speed railway channel modeling are two very urgent problems for us to solve. Before determining the channel model, it is needed to determine the parameters of the transmission channel, which contains amplitude, propagation delay, Azimuth of Departure (AoD), Azimuth of Arrival (AoA) and Doppler frequency. This procedure needs an advanced high resolution channel parameter extraction method.In this article, we first investigate three fundamental methods in categories. Worth mentioning within the first category is the Multiple Signal Classification (MUSIC), which makes full use of the orthogonal characteristics between the signal subspace and the noise subspace and it can be used to extract the incidence direction of the incoherent signal. Estimating Signal Parameters via Rotational Invariance Techniques (ESPRIT) is the typical method of the second category. In this method, the receiving array is divided into two sub-arrays. With the rotation invariant characteristic of the signal subspace, ESPRIT can be used to extract the horizontal azimuth of coherent signals. However, the common weakness of the two algorithms is that they are designed for extracting the azimuth parameter. At last, Expectation Maximization (EM) and Space-Alternating Generalized Expectation Maximization (SAGE) are the representative methods of the third category. Based on the Maximum Likelihood (ML) algorithm, these two methods jointly extracts the multi-dimensional parameters through maximizing the expectation. Moreover, this article gives the simulation results and analysis of those three kinds of methods.After above works, this paper explores the channel parameter extraction algorithm which is based on WINNER II D2a channel model under the high-speed railway scenario. Under the propagation scenario D2a, the channel can be divided into8clusters, and each cluster is made up of20reflection signals whose locations are relatively concentrated. That means the angle values of20rays are very close. In this condition, the SAGE algorithm can’t extract the channel parameter with high resolution. To solve this problem, in this article, we have creatively proposed an improved SIC-SAGE algorithm, which can successfully solve the problem of low resolution of basic SAGE algorithm. In this new proposed algorithm, we first use the ESPRIT algorithm to extract the angle estimation value, and then we introduce the Serial Interference Cancellation to acquire the initial value of SAGE algorithm and give the deduction of the formula. According to our simulation, we can get the conclusion that improved SIC-SAGE algorithm can raise the convergence rate and extraction accuracy significantly. At last, we analyze the parameters which impact the Doppler frequency under high-speed railway scenario. The conclusion is that when the signal-noise-rate (SNR) is high and the number of antenna is large, the improved SIC-SAGE algorithm we provide can get a low Root Mean Square Error Estimation (RMSEE).