Study And Implementation On Several Critical Problems About Wireless Signal Monitoring

With the advances of modern communication technology, the technical requirements for radio management departments rise at the same time. Given this situation, the Radio Management of China is attempting to propose constructive and practical solutions based on the existing radio frequency signals monitoring methods and algorithms. So that they could handle the problems occurred. At present time, there are two technical problems for the signals monitoring of radio frequency from the open space. One is the accurate, automatic and efficient signal modulation classification and the other one is passive, portable and exact locating of the signal sources. Radio management is relative with the national military and civilian security. These two technical issues are of great significant, which could be classified into signal Automatic Modulation Classification (AMC) and passive locating of signals.For the signals modulation classification issue, there are two major methods including the Maximum Likelihood method and Characteristic based method. The research targets are often limited to several space-time mixed modulation signals which could be classified well. However, the existing algorithms are short of effective classification performance for signals with multi-modulation modes. Moreover, there is no good method to balance the confliction between algorithm performance and computational amount. The passive location methods are classified into three types. According to location parameters, they include time-based, direction-based and receptive-intensity-based methods. There are positive and passive locating methods if classified based on whether to emit detection waves. These methods neglect the study on space signal source non-collaborative passive locating and they have higher requirements on locating equipment and less portability.This thesis propose improved solutions to the above two issues based on in-depth study of existing methods and algorithms. Our methods have more feasibility and are easier to be implemented for the practical applications.Firstly, High Order Cumulants (HOC) and Artificial Neural Network (ANN) are used to classify 23 modulation modes. Our methods extract HOC as signal characteristics combined with other time-frequency features and cite Radial Base Function (RBF) ANN as a classifier. In addition, a tradeoff ratio is proposed to measure algorithm classification performance and computational amount. Real signals are classified after down-sampling. The computational amount is greatly reduced on the premise of ensuring classification accuracy performance.Secondly, we apply time-difference-of-arrival (TDOA)-based estimates to locate the signal sources passively. This method use complex-correlation statistics to improve the system precision. We are attempting to develop practical system on the DSP platform. The equipment requirements are much decreased with increased portability on the premise of ensuring locating performance.These two algorithms both implemented in the real system. The automatic modulation classification system can classify simulated signals of 23 modulation modes with classify rate of 97%. For real signals with 10 modulation modes, the classify rate is above 90%. The locating precision of passive locating system can reach to 30m for the signals whose bandwidth is larger than 100KHz in real environments. The feasibility of the two proposed algorithms has been testified through simulation and real tests.