LTE signal classification and center frequency detection without Priori information

Creating a cognitive Long Term Evolution (LTE) network on an arbitrary frequency requires an LTE specific classifier. This paper explains a novel LTE signal classification method that requires no a priori information about the signal parameters including its frequency. The classification method consists of multiple steps. The first step is using a correlation-based discriminator which exploits the characteristics of the LTE cyclic prefix, cell-specific reference symbol sequence and physical broadcast channel. A set of candidate LTE signal frequencies and confidence factors are determined at the end of this step. The candidate frequency set is narrowed down in the second step using the confidence factors. The third step is to estimate the center frequency of the LTE signal. This step relies on known signal characteristics including the synchronization sequences. The time-domain LTE signal is represented in frequency and time grids and two-dimensional cross-correlation is performed. The algorithms were simulated and then implemented on a real-time platform. The results show that an LTE signal with signal-to-noise ratio (SNR) of less than 0 dB can be classified within 0.23 sec.